Silica-supported Ti chloride tetrahydrofuranates, precursors of Ziegler–Natta catalysts

DaltonTransactions

PAPER

Cite this: Dalton Trans., 2013, 42, 12706

Received 5th March 2013,Accepted 8th May 2013

DOI: 10.1039/c3dt50603g

www.rsc.org/dalton

Silica-supported Ti chloride tetrahydrofuranates,precursors of Ziegler–Natta catalysts†

Kalaivani Seenivasan,a Erik Gallo,a,b Andrea Piovano,c Jenny G. Vitillo,a

Anna Sommazzi,d Silvia Bordiga,a Carlo Lamberti,a Pieter Glatzelb andElena Groppo*a

The structural and electronic properties of silica-supported titanium chloride tetrahydrofuranates

samples, obtained by impregnating a polymer-grade dehydroxylated silica with TiCl4(thf )2 and TiCl3(thf )3complexes, precursors of Ziegler–Natta catalysts, are investigated by means of FT-IR, XAS, XES and diffuse

reflectance UV-Vis spectroscopy, coupled with DFT calculations. The properties of the two silica-supported

samples are very similar, irrespective of the starting precursor. In both cases, most of the chlorine ligands

originally surrounding the Ti sites are substituted by oxygen ligands upon grafting on silica. As a conse-

quence, the electronic properties of silica-supported Ti sites are largely different from those of the corres-

ponding precursors, and in both cases most of the grafted Ti sites have a formal oxidation state of +4.

The whole set of experimental data provide evidence that mono-nuclear Ti species are mainly present at

the silica surface.

1. Introduction

Reactions of silica surface with organometallic species are offundamental importance in many areas of chemistry, and inparticular in heterogeneous catalysis.1–6 Amorphous silica iswidely used as a support for many heterogeneous catalysts,because of its high surface area, thermal and mechanicalstability. In most of the cases, the active phase is formed uponreaction of the well defined organometallic precursors withsurface silanol groups, whose concentration and type can betuned by changing the temperature of the pre-treatments.7

When the grafting procedure is performed on a highly de-hydroxylated silica, the grafted metal species can assume asingle-site character.2–4,8 Several heterogeneous catalysts forolefin polymerization are supported on amorphous silica; thesilica types involved are porous, with high specific surface

areas and potentially reactive surface hydroxyl groups almostentirely upon the pore walls.9 The first developed polyolefincatalyst based on silica was the Phillips catalyst for ethylenepolymerization, where the active sites are diluted chromiumcentres.10–12 In such a case it was demonstrated that silicadoes not play the role of an inert support only, but directlyinfluences the properties of the grafted chromiumspecies in terms of accessibility, coordination ability andflexibility.10–14 Amorphous silica was successively employed asa support also for Ti-based Ziegler–Natta catalysts for ethylenepolymerization.9,15,16 The simple combination of titaniumtetrachloride on silica yielded low-reactivity catalysts; however,the combination of a magnesium compound with a poroussilica material, followed by reaction with titanium tetra-chloride, resulted in a catalyst showing an enhanced reactivityand the excellent handling and polymer particle controlcharacteristic of Phillips’ chromium catalysts.9,15,16 In contrastto chromium-based catalysts, the role of silica surface chem-istry in Ziegler–Natta catalysts is often overlooked.

In the frame of a wider work devoted to the physical–chemi-cal characterization of silica-supported Ziegler–Natta catalystsbased on tetrahydrofuranates of TiCl4 and MgCl2,

17 we per-formed a detailed spectroscopic investigation on the reactivityof titanium chloride tetrahydrofuranates (TiCl4(thf)2 andTiCl3(thf)3, thf = tetrahydrofurane) towards a polymer-gradesilica and on the structure of the resulting Ti-grafted sites.Reactivity of surface hydroxyl groups of silica towards titaniumchloride and other metal chlorides was studied in the past byseveral research groups, mainly by FT-IR spectroscopy and

†Electronic supplementary information (ESI) available: Additional experimentaldetails, XRPD patterns, FT-IR data of interaction with H2O and CO asprobe molecules, details on EXAFS and vtc-XES dta analysis. See DOI:10.1039/c3dt50603g

aDepartment of Chemistry, NIS Centre of Excellence and INSTM University of Torino,

via Quarello 15, I-10135 Torino, Italy. E-mail: [email protected];

Fax: (+)39 011 6707855; Tel: (+) 39 011 6708373bEuropean Synchrotron Radiation Facility, 6 Rue Jules Horowitz, 38043 Grenoble,

FrancecInstitut Laue Langevin, 6 Rue Jules Horowitz BP 156, F-38042 Grenoble Cedex 9,

FrancedCentro Ricerche per le Energie non Convenzionali, Istituto ENI Donegani, Via

Fauser, 4 - 28100 Novara, Italy

12706 | Dalton Trans., 2013, 42, 12706–12713 This journal is © The Royal Society of Chemistry 2013

Publ

ishe

d on

09

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TÀ

DE

GL

I ST

UD

I D

I T

OR

INO

on

20/0

8/20

13 0

9:55

:59.

View Article OnlineView Journal | View Issue

chemical analysis, and was used as a means of determiningthe surface structure and the number of isolated andH-bonded hydroxyl groups of silica.18–23 In particular, IRstudies demonstrated the occurrence of a dissociative chemi-sorption of TiCl4 on dehydrated silica, although it was notshown conclusively whether the isolated or the H-bondedhydroxyl groups are the most reactive toward TiCl4.

18–23 For-mation of both mono-functional uSiO-TiCl3 and bi-functional(uSiO)2-TiCl2 species was postulated, whose proportiondepends mainly on the silica dehydration temperature and onthe reaction temperature. Hydrochloric acid was found as a by-product in all the cases, and in some cases chlorination ofsilica was also observed. A similar reactivity was reported forseveral organometallic complexes of the general formulaXxMLn (where M = transition metal, X = halogen and L =ligand) that were found to react with surface silanols uSiOHof highly dehydroxylated silica to yield uSiOMXx−1Ln speciesalong with HX.1,3,4,24–27 For tetrahedral XxMLn complexes, thegrafting process usually occurs without major changes interms of structure and geometry of the grafted fragment; thatis, tetrahedral d0 ML4 complexes remain mostly tetrahedralupon grafting for a large range of metals.

The case of Ti chloride tetrahydrofuranate complexes dis-cussed in the following is less straightforward for at least tworeasons. First, the starting complexes are six-folded co-ordinated and are less reactive with respect to the pure metalchlorides due to the presence of the thf ligands. Secondly, theemployed silica is not highly dehydroxylated and, beingporous, contains a large amount of internal hydroxyl groupsinteracting with each other. Therefore, the number of possiblestructures resulting from the grafting of the Ti complexes istheoretically much larger than those hypothesized for TiCl4alone on highly dehydroxylated silica and also the two extremepossibilities of no-grafting or formation of multi-nuclear Tispecies should be taken into account. While the occurrence ofTi grafting through surface uSiOH groups can be easilydemonstrated by FT-IR spectroscopy (by looking at the con-sumption of the IR absorption bands due to surface uSiOHgroups), insight into the geometric and electronic structure ofthe grafted Ti sites can be obtained only by coupling manycharacterization techniques, such as X-ray absorption spectro-scopy (XAS),28 X-ray emission spectroscopy (XES)29 and diffusereflectance UV-Vis spectroscopy.30

2. Experimental2.1 Materials

Davison sylopol silica 955 grade (surface area = 276 m2 g−1,pore volume = 1.76 ml g−1, average pore diameter = 266 Å,average particle size = 31 μm) was used as support, after a pre-treatment in air at 550 °C for 8 hours, followed by a coolingstep carried out in a nitrogen atmosphere. TiCl4(thf)2 andTiCl3(thf)3 precursors were synthesized by following the recipereported elsewhere.31,32 The TiIV and TiIII chloride tetrahydro-furanates were dissolved in dry tetrahydrofurane (thf) and

impregnated on dehydroxylated SiO2 in a controlled atmos-phere, using a Schlenk technique. In both cases, Ti loadingwas 2 wt%. The excess of the solvent was further removed bygently heating the sample up to ca. 60 °C.

2.2 Techniques

X-ray powder diffraction patterns were collected with aPW3050/60 X′Pert PRO MPD diffractometer from PANalyticalworking in Debye–Scherrer geometry, using as X-ray source aCu anode. The samples were measured as powders inside a0.8 mm boron-silicate capillary sealed in an inert atmosphere.

FT-IR spectra were acquired in transmission mode on aBruker Vertex70 spectrophotometer, at a resolution of 2 cm−1.The samples were measured in the form of self-supportingpellets inside a quartz cell in a controlled atmosphere. UV-Vis-NIR spectra were collected in diffuse reflectance mode on aCary5000 Varian spectrophotometer. All the samples weremeasured in powdered form inside a home-made cell havingan optical window (suprasil quartz) and allowing one toperform measurements in controlled atmosphere. Silica-sup-ported samples were measured without dilution, whereas non-supported samples were diluted in Teflon.

X-ray absorption (XAS) experiments at the Ti K-edge wereperformed at the BM23 beamline of the ESRF facility (Greno-ble, F). The EXAFS spectra of the two reference samples werecollected in transmission mode, whereas those of the silica-supported samples were collected in fluorescence. The EXAFSpart of the spectra was collected up to 12 Å−1 with a variablesampling step in energy, resulting in Δk = 0.03 Å−1. For eachsample, three equivalent EXAFS spectra were acquired andaveraged before the data analysis. EXAFS data analysis was per-formed using the Athena and Arthemis software.33 Phase andamplitudes were calculated by FEFF6.0 code.34

XES experiments were performed at beamline ID26 of theEuropean Synchrotron Radiation Facility (ESRF, France). Theincident energy (Ω) was selected by means of a pair of cryo-genically cooled Si(311) single crystals (higher harmonics weresuppressed by three Si mirrors operating in total reflection).The fluorescence photon energy (ω) was selected using anemission spectrometer working in vertical Rowland geometryemploying five Ge(331) spherically bent analyzer crystals ofradius 1000 mm covering 70–110 degrees in the horizontalscattering plane. The emitted photons were detected using anavalanche photo-diode. The total energy bandwidth was 0.9 eV(as determined from the full width at half maximum of theelastically scattered peak). The background of the vtc-XESspectra, due to the Kβ1,3 peak tail, was subtracted according tothe procedure discussed in ref. 35 and 36. DFT calculationswere performed within the one electron approximation usingthe ORCA 2008 code.37 Details are given in ref. 35 and ESI.†The beam size on the sample was approximately 0.8 mm hori-zontally and 0.2 mm vertically. For both XAS and valence-to-core XES (vtc-XES) experiments, the samples were measured inthe form of self-supporting pellets prepared inside a glove-boxand placed inside a home-made cell with kapton windows;before measurements, the cell was outgassed in order to remove

Dalton Transactions Paper

This journal is © The Royal Society of Chemistry 2013 Dalton Trans., 2013, 42, 12706–12713 | 12707

Publ

ishe

d on

09

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TÀ

DE

GL

I ST

UD

I D

I T

OR

INO

on

20/0

8/20

13 0

9:55

:59.

View Article Online

the Ar, which absorbs most of the beam at this low energy. Thetwo precursors were diluted in a paraffin.

3. Results and discussion

The two SiO2-supported samples investigated in this work wereprepared by impregnating a dehydroxylated silica with a tetra-hydrofurane (thf) solution of either TiCl4(thf)2 or TiCl3(thf)3precursors (hereafter labelled TiIV and TiIII), resulting in bothcases in a Ti loading of 2 wt% (see the Experimental section).The employed silica is a polymer-grade silica dehydroxylated at550 °C, therefore containing a non-negligible amount of inter-acting uSiOH groups. The XRPD patterns of both TiCl4(thf)2/SiO2 and TiCl3(thf)3/SiO2 (hereafter TiIV/SiO2 and TiIII/SiO2)show only a broad peak centred around 2θ = 21°, due to theamorphous silica support (Fig. S1†). The absence of otherdiffraction peaks indicates that no crystalline domains arepresent (within the sensitivity of the technique).

FT-IR spectroscopy was used to evaluate the occurrence ofTi grafting, as already proposed in the literature.18–23 TheFT-IR spectra of TiIV/SiO2 and TiIII/SiO2 are compared to thatof the bare SiO2 support pre-activated at 550 °C in Fig. 1. TheFT-IR spectrum of SiO2 (light grey) shows several IR absorptionbands in the ν(OH) region: the sharp band at 3746 cm−1 is dueto isolated silanol groups, whereas the broad absorption bandhaving two maxima around 3670 and 3550 cm−1 indicates thepresence of nests of hydroxyl groups weakly interacting witheach other,38,39 as commonly found in porous silicas.9 Atlower wavenumber values, the spectrum is dominated by theintense Si–O vibrational modes of the framework (in the

1400–950 cm−1 and 850–770 cm−1 ranges) and by their over-tone modes (in the 2100–1550 cm−1 range).40

The FT-IR spectra of the TiIV/SiO2 and TiIII/SiO2 samples(dark grey and black in Fig. 1) differ from that of the bare SiO2

in both ν(OH) and framework vibrational mode regions. In par-ticular, the IR absorption bands of both isolated silanols andinternal hydroxyl groups have a lower intensity than in thepure support (more evident after subtracting the spectrum ofthe bare silica, Fig. 1b), proving that a fraction of the surfacehydroxyl groups react with the Ti chloride tetrahydrofuranateprecursors during the synthesis step. Simultaneously, a new IRabsorption band appears around 925 cm−1, well evident in anarrow frequency region of transparency (subtracted spectraare shown in Fig. 1c). Similar IR absorption bands were pre-viously observed upon gas-phase reaction of TiCl4 with de-hydrated silica and assigned to Si–O–Ti vibrations of bi-functional grafted (uSiO)2-TiCl2 species.19,20,22 In analogywith the chemistry of TiCl4, it can be hypothesized that thetwo Ti precursors react with the surface hydroxyl groups at thesilica surface during the synthesis in solution; likely, the HClreleased as a by-product remains in solution. Additional evi-dence that most of the chlorine ligands are lost as a conse-quence of Ti grafting is obtained by a FT-IR study of H2Oadsorption and reaction (Section S3 and Fig. S2†). H2O ismainly physisorbed on both TiIV/SiO2 and TiIII/SiO2 samplesand almost no hydrolysis reaction occurs, as it would beexpected in the presence of exposed chlorine ligands.

Finally, the IR spectra of both TiIV/SiO2 and TiIII/SiO2 showIR absorption bands in the 3000–2800 cm−1 region (ν(CH2)modes) and in the 1500–1350 cm−1 range (δ(CH2) modes), tes-tifying that thf molecules, or some species deriving from thfrearrangement,41 are still present in the sample. However, it isdifficult to verify whether thf is still attached to the Ti sites oris adsorbed on the silica support. In this regard, it should benoticed that independent FT-IR experiments of CO adsorbedat 100 K provide an evidence that the majority of grafted Tisites do not have coordination vacancies available for CO inser-tion (Section S4 and Fig. S3†), suggesting that they retain thesix-fold coordination geometry characteristic of the startingcomplexes. As a consequence, at least a fraction of the thfshould be coordinated to the Ti sites.

In the absence of any long-range order (see XRPD patternsin Fig. S1†), the local structure around the Ti atoms should beinvestigated by element selective spectroscopic techniques,such as Ti K-edge EXAFS. The potentiality of EXAFS in unravel-ling the local structure of the transition metal sites graftedon amorphous silica was largely demonstrated in thepast.1,2,13,28,42,43 Fig. 2 shows the phase-uncorrected Fouriertransforms (FT) of the k3χ(k) EXAFS functions for both TiIV/SiO2 and TiIII/SiO2 (black spectra) in both modulus and ima-ginary parts (parts a and b, respectively). Also the EXAFSspectra of the corresponding TiCl4(thf)2 and TiCl3(thf)3 precur-sors are shown for comparison (gray spectra). Both Ti chloridetetrahydrofuranate complexes are molecular crystals; in eachmolecular unit, Ti atoms are six-fold coordinated to the sametype of ligands (i.e. oxygen of the thf ring at around 2.0 Å and

Fig. 1 Part (a): FT-IR spectra (collected in an inert atmosphere) of TiIV/SiO2

(grey) and TiIII/SiO2 (black), compared to that of the SiO2 support pre-activatedat 550 °C (light grey) (TiIV: TiCl4(thf )2; Ti

III: TiCl3(thf )3). The spectra have beennormalized to the intensity of the absorption bands around 1900 cm−1 (firstovertones of the silica framework modes) in order to account for the thickness ofthe pellets, and a straight line, representing the scattering contribution, was sub-tracted to better compare the spectra with each other. Insets (b) and (c) show amagnification of the spectra of TiIV/SiO2 and TiIII/SiO2 in the 4000–2750 cm−1

and 950–850 cm−1 regions, after subtraction of the spectrum of bare SiO2.

Paper Dalton Transactions

12708 | Dalton Trans., 2013, 42, 12706–12713 This journal is © The Royal Society of Chemistry 2013

Publ

ishe

d on

09

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TÀ

DE

GL

I ST

UD

I D

I T

OR

INO

on

20/0

8/20

13 0

9:55

:59.

View Article Online

chlorine at around 2.3 Å), although in a different relativeamount.17 The similarity in the local structure of the two pre-cursors explains why the corresponding EXAFS spectra are verysimilar to each other. They are dominated by a first shellsignal centered around 1.8 Å (not corrected in phase), which isthe result of the sum of the six single-scattering contributionsderiving from the first shell ligands. Although the molecularstructure of the two precursors is well known from XRPD, aquantitative analysis of the EXAFS signals was not possiblewithin the available data quality because of the strong corre-lations among the fitted EXAFS variables.

The EXAFS spectra of the TiIV/SiO2 and TiIII/SiO2 samplesare different from those of the precursors, providing strikingevidence that the two complexes are not simply physisorbedon the silica surface. In both cases, the |FT| is characterized bytwo main contributions, around 1.4 and 1.9 Å, respectively,whereas no signals are observed at longer distances. Thespectra of TiIV/SiO2 and TiIII/SiO2 are very similar to eachother, although the relative intensity of the two contributionsis slightly different in the two cases. As discussed for the refer-ence compounds, a quantitative analysis of the EXAFS data ofTiIV/SiO2 and TiIII/SiO2 samples was affected by the strong cor-relation among the fitted variables and thus is not reliable.Nevertheless, a series of fits was performed by changing in asystematic way the relative number of oxygen and chlorineligands. In such a way we introduced arbitrary constraints thatsignificantly reduced the correlation among the remainingparameters. The results are discussed in Section S5 and sum-marized in Tables S1 and S2.† Contrary to many successfulexamples in the literature where EXAFS is used to determinethe local structure around the absorbing metal species, in thepresent case we can use the EXAFS data only to conclude thatthe ligand sphere around the Ti sites contains more oxygen

(contributing to the first shell peak) and less chlorine ligands(responsible for the contribution around 1.9 Å, not correctedin phase) than in the corresponding TiIV and TiIII precursors.This finding is in agreement with the hypothesis that the Tichloride tetrahydrofuranate complexes do graft to the silicasurface via elimination of HCl. Finally, the absence of anysignal at longer distances suggests that mono-nuclear Tispecies are mainly present at the silica surface.

Since EXAFS spectroscopy was not conclusive to investigatethe local structure around the Ti sites, we turned to valence-to-core X-ray emission spectroscopy (vtc-XES), which has beenproved to be effective in the identification of the metal–ligandenvironment also when, contrary to XAS, the atomic numberof the ligands is very close (e.g. XES can distinguish betweenO, N and C ligands).44–46 vtc-XES is a second order opticalprocess. For a 3d-transition metal, it can be induced by theabsorption of an X-ray photon with energy higher than the 1sphoto-excitation threshold; the so formed core hole is thenfilled by a valence electron lying below the Fermi level.35,36,47

Note that the vtc-XES features reflect the valence band densityof occupied electronic states projected onto Ti p orbitalangular momentum; thus, molecular orbitals with no metal pcontribution cannot be detected.35,36

Fig. 2c shows the vtc-XES of both TiIV/SiO2 and TiIII/SiO2,compared to those of the corresponding TiIV and TiIII precur-sors. Two main regions can be, in general, identified in a vtc-XES spectrum, which are called Kβ″ and Kβ2,5 (white and greyboxes in Fig. 2c, respectively). The Kβ″ fluorescence lines aremainly due to transitions involving molecular orbitals (MOs)with ligands s-atomic character. They can be used for ligandidentification.35,36,44–46,48–50 Two Kβ″ lines are present in thevtc-XES spectrum of TiIV precursor (A and B in Fig. 2c). Accord-ing to previous works on Ti compounds,35,44 A and B are

Fig. 2 Part (a): Phase-uncorrected |FT| of the k3χ(k) EXAFS function for TiIV/SiO2 and TiIII/SiO2 (black). The EXAFS spectra of the two Ti chloride tetrahydrofuranatesprecursors are shown for comparison (grey). The spectra were FT-transformed in the Δk = 2.0–10.5 Å−1 range. The spectra have been vertically translated for clarity.Part (b): the same as part (a) for the Imm(FT). Part (c): vtc-XES spectra of TiIV/SiO2 and TiIII/SiO2 (black) and of the corresponding TiIV and TiIII precursors (grey). Thespectra have been vertically translated for clarity. The Kβ’’ and Kβ2,5 regions are indicated with white and grey boxes, respectively. A and B features identify oxygenand chlorine ligands, respectively. TiIV: TiCl4(thf )2; Ti

III: TiCl3(thf )3.

Dalton Transactions Paper

This journal is © The Royal Society of Chemistry 2013 Dalton Trans., 2013, 42, 12706–12713 | 12709

Publ

ishe

d on

09

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TÀ

DE

GL

I ST

UD

I D

I T

OR

INO

on

20/0

8/20

13 0

9:55

:59.

View Article Online

related to transitions involving MOs having primarily O(2s)and Cl(2s) character, respectively; hence, they identify thepresence of oxygen and chlorine ligands in the first coordi-nation sphere of the Ti sites. Feature A is not observed in thevtc-XES spectrum of TiIII precursor. In this case, in fact, theMOs have mainly O(2s) character and no/little metal contri-bution and thus cannot be observed.44 The Kβ″ regions of thevtc-XES spectra for TiIV/SiO2 and TiIII/SiO2 are very differentfrom those of the corresponding TiIV and TiIII precursors, butsimilar to each other. In particular, in both cases, feature A(identifying oxygen ligands) is very intense and broad, whereasfeature B (identifying chlorine ligands) is much less visiblethan in the spectra of the two precursors. Therefore, the Kβ″lines of the vtc-XES spectra provide striking evidence that aligand exchange takes place when TiIV and TiIII precursorsreact with SiO2: most of the chlorine ligands are substituted byoxygen ligands.

Additional insights can be gained by observing the featuresthat compose the Kβ2,5 region of the vtc-XES spectra (grey boxin Fig. 2c). In general, the Kβ2,5 lines are primarily due to tran-sitions involving molecular orbitals having a p-type ligandatomic character and therefore they are particularly sensitive tochanges in the valence orbitals of the material under investi-gation.46,48 In the present case, quantum mechanics calcu-lations (see below and ESI†) suggest that the Kβ2,5 lines aredue to transitions involving MOs having either an O(2p)–C-(2s2p) character or a Cl(3p) atomic character. Also in thisregion the spectra of the two precursors are remarkablydifferent from those of the corresponding silica-supported

samples, suggesting that the electronic properties and theligand environment of the Ti sites are changed.

Since the electronic properties of materials are usually thedomain of UV-Vis and XANES spectroscopy, we carried out adetailed investigation by means of both techniques. TheUV-Vis spectra of the two TiIV and TiIII chloride tetrahydrofura-nates precursors (grey spectra in Fig. 3a and b) were alreadydiscussed in our previous work.17,43 Both precursors display awell defined colour: TiIV is bright yellow as a consequence ofthe intense Cl(2p) → Ti(3d) charge-transfer transition havingthe edge (arbitrarily defined as the maximum of the firstderivative) at around 22 600 cm−1 (2.80 eV). In contrast, thepale blue colour of TiIII has its origin from the double and welldefined d–d absorption band centred around 14 000 cm−1

(1.74 eV), which is typical of TiIII species (d1 transition metal)in a six-fold coordination;51–53 the intense Cl(2p) → Ti(3d)charge transfer falls entirely in the UV region (edge at26 000 cm−1, 3.22 eV). XANES spectroscopy gives complemen-tary information on transitions involving un-occupied states(Fig. 3c and d). The XANES spectra of TiIV and TiIII precursorsdiffer in both the pre-edge and the edge regions. In the pre-edge region, two weak peaks are observed in both cases, whichare assigned to 1s → 3pd transitions. These peaks have a lowerintensity for the more symmetric TiIII precursor (C3 symmetry)than for the TiIV one (C1 symmetry). The edge position shifts tolower energies by about 2 eV when going from TiIII to TiIV pre-cursor, as expected on the basis of the formal oxidation state.

Both UV-Vis and XANES spectra of TiIV/SiO2 and TiIII/SiO2

are remarkably different from those of the corresponding

Fig. 3 UV-Vis diffuse reflectance spectra (parts a and b), XANES spectra (parts c and d), and Kβ1,3 XES spectra (parts e and f ) for TiIV/SiO2 and TiIII/SiO2 (black spectra)compared to those of the corresponding TiIV and TiIII precursors (grey); TiIV: TiCl4(thf )2; Ti

III: TiCl3(thf )3. Inset in part (b) shows an enlargement of the absorption bandsin the d–d region. Insets in parts (c) and (d) display an enlargement of the pre-edge peaks in the XANES spectra. Vertical dotted lines in (e) and (f ) highlight the positionof the maxima.

Paper Dalton Transactions

12710 | Dalton Trans., 2013, 42, 12706–12713 This journal is © The Royal Society of Chemistry 2013

Publ

ishe

d on

09

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TÀ

DE

GL

I ST

UD

I D

I T

OR

INO

on

20/0

8/20

13 0

9:55

:59.

View Article Online

precursors. Starting from the UV-Vis spectra (Fig. 3a and b), inboth cases the intense edge in the charge-transfer region shiftsat higher wavenumbers (i.e. the energy necessary for transfer-ring one electron from the ligands to Ti increases); for TiIII/SiO2 also the absorption bands due to d–d transitions upwardshift and broaden. These drastic changes can be qualitativelyexplained by using the semi-empirical Jorgensen’s rules,54 i.e.the energy of a charge-transfer transition is a function of theoptical electronegativity values of the metal, χopt(M) (which inturn is a function of the metal coordination number), and ofthe ligand, χopt(X), following the equation: ν(cm−1) =30 000 cm−1 [χopt(X) − χopt(M)]. The shift of the charge-transferband towards higher wavenumber when the Ti precursors aregrafted on SiO2 can be explained by two hypotheses only:54–56

(i) the charge-transfer band is still due to a Cl(2p) → Ti(3d)transition (χopt(X) constant), but the Ti sites have now a lowercoordination number (χopt(M) decreases); or (ii) the Ti sitesstill have a six-fold coordination (χopt(M) constant), but aligand exchange occurs (from chlorine to oxygen, χopt(X)increases). Only the second hypothesis is compatible with theexperimental results discussed above. Therefore, UV-Visspectroscopy provides additional evidence that both TiIV andTiIII precursors graft to the silica surface by exchanging chlor-ine ligands with oxygen belonging to the silica surface.

The evolution of the UV-Vis spectra upon grafting of TiIV

and TiIII precursors on silica is mirrored by the evolution ofthe XANES spectra, as shown in Fig. 3c and d. In both cases:(i) the pre-edge peak is enhanced in intensity, suggesting thatthe supported Ti sites have a six-folded coordination more dis-torted than that of the Ti precursors; (ii) the threshold-edgesignificantly shifts to higher energy. It is usually difficult todisentangle the electronic and geometric information con-tained in a XANES spectrum. Another method that can beused to characterize the formal oxidation state of 3d-tran-sition-metals is by the analysis of the shape and energy posi-tion of X-ray emission lines such as Kα1, Kα2 and Kβ1,3, whichare less sensitive to the symmetry of the system under investi-gation compared to XANES.46,57 It has been shown that Kα1,Kα2 and Kβ1,3 provide similar information, but the Kβ1,3 is theline more sensitive to the ligand environment of the metalion.58,59 The Kβ1,3 spectra of TiIV/SiO2 and TiIII/SiO2 are shownin Fig. 3e and f and compared to those of the two precursors.The maxima of the Kβ1,3 lines for the two precursors areshifted by about 0.4 eV. No changes are observed in the Kβ1,3spectrum of TiIV/SiO2 when compared to that of the corres-ponding TiIV precursor. In contrast, the spectrum of TiIII/SiO2

is shifted towards low energy with respect to that of TiIII. Theenergy position of its maximum is close to that of TiIV/SiO2,providing evidence that the formal oxidation state of thegrafted Ti sites is higher than +3. We propose that the majorpart of the Ti sites in TiIII/SiO2 and TiIV/SiO2 may have thesame formal oxidation number, i.e. +4.

On the basis of the information obtained so far, we carriedout a systematic series of theoretical calculations based onground state density functional theory (Section S6), which hasbeen shown to be effective at reproducing the vtc-XES spectral

features.50 We computed all the reasonable monomericmodels for the Ti local environment in TiIV/SiO2, where Ti isfour-, five- and six-folded coordinated, while varying thenumber of the three possible ligands (i.e. oxygen of the silicasurface, oxygen of thf and chlorine). It is recalled that the vtc-XES spectrum of TiIV/SiO2 is similar to that of TiIII/SiO2, seeFig. 2c. The list of adopted models is summarized in Table S3†as a function of the Ti coordination number. Also a tentativedimeric model was computed in order to evaluate the distinc-tive feature for bridging chlorine ligands. Examples of com-puted vtc-XES for TiIV/SiO2 are shown in Fig. 4 in comparisonto the experimental spectrum. The relative agreement betweenthe experimental spectrum and the computed ones was evalu-ated by using a first moment analysis of the experimental vs.theoretical features.29,60 The quality of the calculation isexpressed in terms of the quality factor Θ, as discussed inSection S1; the best model corresponds to Θ = 1.0, while theworst has Θ = 4.0. In general, a small Θ value (<1.5) is obtainedfor models characterized by a small number of chlorineligands (maximum 2); the corresponding simulated spectra are

Fig. 4 Experimental XES spectrum of TiIV/SiO2 (exp) and selection of spectrasimulated by means of the minimal (monomeric) clusters p4, p5, s3, s6, and s7and of a tentative dimeric model. The labels refer to the names of the clustersreported in Table S3.† The spectra have been vertically translated for clarity. Inorder to evaluate the agreement between the simulated and the experimentalvtc-XES spectra the zero of the energy was shifted to the maximum of the Kβ’’characteristic of oxygen ligands.

Dalton Transactions Paper

This journal is © The Royal Society of Chemistry 2013 Dalton Trans., 2013, 42, 12706–12713 | 12711

Publ

ishe

d on

09

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TÀ

DE

GL

I ST

UD

I D

I T

OR

INO

on

20/0

8/20

13 0

9:55

:59.

View Article Online

shown in Fig. 4. In the best model (s6) the Ti site is grafted tothe silica surface through three oxygen ligands; the six-foldcoordination sphere is completed by two thf and one chlorineligands. The tentative dimeric model is not able to reproducethe main features of the experimental spectrum, in both Kβ″and Kβ2,5 regions. Although the level of calculation could befurther improved, and the small models shown in Fig. 4 mustbe taken only as a schematic representation of the morecomplex situation found in amorphous silica, it is evidentthat vtc-XES spectroscopy coupled with DFT calculation ispotentially able to provide structural details of the grafted Tispecies.36

4. Conclusions

The work reports on a systematic investigation of the structuraland electronic properties of silica-supported titanium chloridetetrahydrofuranates samples, obtained by impregnating apolymer-grade dehydroxylated silica with TiCl4(thf)2 andTiCl3(thf)3 complexes, precursors of Ziegler–Natta catalysts.While the occurrence of titanium grafting through surfaceuSiOH groups was easily demonstrated by FT-IR spectroscopy,the determination of the local structure of the grafted Ti sitesand of their electronic properties required the synergic appli-cation of many complementary techniques, coupled with DFTcalculations. All the experimental data do suggest that bothstructural and electronic properties of silica-supportedsamples are very similar, irrespective of the starting precursor,i.e. TiCl4(thf)2 or TiCl3(thf)3. In both cases, most of the chlor-ine ligands originally surrounding the Ti sites are substitutedby oxygen ligands upon grafting on silica, as happens for themore reactive and geometrically different TiCl4. The electronicproperties of silica-supported Ti sites are largely different fromthose of the corresponding precursors, and in both cases mostof the grafted Ti sites have a formal oxidation state of +4.Besides the interest in Ziegler–Natta catalysis, such an investi-gation could be of high relevance in the field of organometallicchemistry and reactivity of metal-oxide surfaces towardsorganometallic compounds. The results discussed hereindemonstrate that when dealing with the reactivity of molecularcomplexes towards high surface area amorphous supports,care must be taken in analyzing experimental results. In manycases, the observed spectra contain both electronic and geo-metric information on the supported metal sites, and it isoften difficult to disentangle the two. The synergic use ofcomplementary experimental techniques is a valuable instru-ment to get insights into the properties of the grafted sites.43

Acknowledgements

We would like to thank Prof. Adriano Zecchina for the everydaydiscussions and for his fruitful contribution to spectroscopyover the years. We are grateful to Olivier Mathon (BM23 atESRF), Mauro Rovezzi and Christophe Lapras (ID26 at ESRF)

for the technical help during the XAS/XES experiment, and toHarald Muller (Chemical Lab at ESRF) for providing us theindispensable (and always perfect) glove-box during the experi-ments at ESRF.

Notes and references

1 J. M. Basset, F. Lefebvre and C. Santini, Coord. Chem. Rev.,1998, 180, 1703.

2 C. Coperet, M. Chabanas, R. P. Saint-Arroman andJ. M. Basset, Angew. Chem., Int. Ed., 2003, 42, 156.

3 F. Rascon, R. Wischert and C. Coperet, Chem. Sci., 2011, 2,1449.

4 D. Gajan and C. Coperet, New J. Chem., 2011, 35, 2403.5 J. Amor Nait Ajjou and S. L. Scott, Organometallics, 1997,

16, 86.6 C. A. Demmelmaier, R. E. White, J. A. van Bokhoven and

S. L. Scottt, J. Phys. Chem. C, 2008, 112, 6439.7 L. T. Zhuravlev, Colloids Surf., A, 2000, 173, 1.8 J. M. Thomas, R. Raja and D. W. Lewis, Angew. Chem., Int.

Ed., 2005, 44, 6456.9 T. J. Pullukat and R. E. Hoff, Catal. Rev. Sci. Eng., 1999, 41,

389.10 M. P. McDaniel, Adv. Catal., 2010, 53, 123.11 E. Groppo, C. Lamberti, S. Bordiga, G. Spoto and

A. Zecchina, Chem. Rev., 2005, 105, 115.12 A. Zecchina and E. Groppo, Proc. R. Soc. London, Ser. A,

Math. Phys. Eng. Sci., 2012, 468, 2087.13 D. Gianolio, E. Groppo, J. G. Vitillo, A. Damin, S. Bordiga,

A. Zecchina and C. Lamberti, Chem. Commun., 2010, 46,976.

14 E. Groppo, C. Lamberti, G. Spoto, S. Bordiga, G. Magnaccaand A. Zecchina, J. Catal., 2005, 236, 233.

15 G. G. Hlatky, Chem. Rev., 2000, 100, 1347.16 M. Y. Lee and S. L. Scott, Chem.–Eur. J., 2011, 17, 4632.17 K. Seenivasan, A. Sommazzi, F. Bonino, S. Bordiga and

E. Groppo, Chem.–Eur. J., 2011, 17, 8648.18 B. A. Morrow and A. H. Hardin, J. Phys. Chem., 1979, 83,

3135.19 J. B. Kinney and R. H. Staley, J. Phys. Chem., 1983, 87, 3735.20 S. Haukka, E. L. Lakomaa, O. Jylha, J. Vilhunen and

S. Hornytzkyj, Langmuir, 1993, 9, 3497.21 S. Haukka, E. L. Lakomaa and A. Root, J. Phys. Chem., 1993,

97, 5085.22 A. Kytokivi and S. Haukka, J. Phys. Chem. B, 1997, 101,

10365.23 K. Schrijnemakers, P. Van Der Voort and E. F. Vansant,

Phys. Chem. Chem. Phys., 1999, 2569.24 F. Bini, C. Rosier, R. P. Saint-Arroman, E. Neumann,

C. Dablemont, A. de Mallmann, F. Lefebvre, G. P. Niccolai,J. M. Basset, M. Crocker and J. K. Buijink, Organometallics,2006, 25, 3743.

25 G. Tosin, C. C. Santini, M. Taoufik, A. De Mallmann andJ. M. Basset, Organometallics, 2006, 25, 3324.

Paper Dalton Transactions

12712 | Dalton Trans., 2013, 42, 12706–12713 This journal is © The Royal Society of Chemistry 2013

Publ

ishe

d on

09

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TÀ

DE

GL

I ST

UD

I D

I T

OR

INO

on

20/0

8/20

13 0

9:55

:59.

View Article Online

26 R. P. Saint-Arroman, J. M. Basset, F. Lefebvre andB. Didillon, Appl. Catal. A, 2005, 290, 181.

27 E. Le Roux, M. Chabanas, A. Baudouin, A. de Mallmann,C. Coperet, E. A. Quadrelli, J. Thivolle-Cazat, J. M. Basset,W. Lukens, A. Lesage, L. Emsley and G. J. Sunley, J. Am.Chem. Soc., 2004, 126, 13391.

28 S. Bordiga, E. Groppo, G. Agostini, J. A. Van Bokhoven andC. Lamberti, Chem. Rev., 2013, 113, 1736.

29 P. Glatzel and U. Bergmann, Coord. Chem. Rev., 2005, 249,65.

30 R. A. Schoonheydt, Chem. Soc. Rev., 2010, 39, 5051.31 L. E. Manxzer, Inorg. Synth., 1982, 21, 135.32 L. E. Manxzer, Inorg. Synth., 1982, 21, 137.33 B. Ravel and M. Newville, J. Synchrotron Radiat., 2005, 12,

537.34 S. I. Zabinsky, J. J. Rehr, A. L. Ankudinov, R. C. Albers and

M. J. Eller, Phys. Rev. B: Condens. Matter, 1995, 52, 2995.35 E. Gallo, C. Lamberti and P. Glatzel, Phys. Chem. Chem.

Phys., 2011, 13, 19409.36 E. Gallo, F. Bonino, J. C. Swarbrick, T. Petrenko, A. Piovano,

S. Bordiga, D. Gianolio, E. Groppo, F. Neese, C. Lambertiand P. Glatzel, ChemPhysChem, 2013, 14, 79.

37 F. Neese, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2012, 2,73.

38 S. Bordiga, I. Roggero, P. Ugliengo, A. Zecchina, V. Bolis,G. Artioli, R. Buzzoni, G. Marra, F. Rivetti, G. Spanò andC. Lamberti, J. Chem. Soc., Dalton Trans., 2000, 3921.

39 S. Bordiga, P. Ugliengo, A. Damin, C. Lamberti, G. Spoto,A. Zecchina, G. Spanò, R. Buzzoni, L. Dalloro and F. Rivetti,Top. Catal., 2001, 15, 43.

40 G. Ricchiardi, A. Damin, S. Bordiga, C. Lamberti, G. Spanò,F. Rivetti and A. Zecchina, J. Am. Chem. Soc., 2001, 123,11409.

41 E. Grau, A. Lesage, S. Norsic, C. Copéret, V. Monteil andP. Sautet, ACS Catal., 2013, 3, 52.

42 E. Groppo, C. Prestipino, F. Cesano, F. Bonino, S. Bordiga,C. Lamberti, P. C. Thüne, J. W. Niemantsverdriet andA. Zecchina, J. Catal., 2005, 230, 98.

43 E. Groppo, K. Seenivasan and C. Barzan, Catal. Sci.Technol., 2013, 3, 858.

44 J. C. Swarbrick, Y. Kvashnin, K. Schulte, K. Seenivasan,C. Lamberti and P. Glatzel, Inorg. Chem., 2010, 49, 8323.

45 G. Smolentsev, A. V. Soldatov, J. Messinger, K. Merz,T. Weyhermuller, U. Bergmann, Y. Pushkar, J. Yano,V. K. Yachandra and P. Glatzel, J. Am. Chem. Soc., 2009,131, 13161.

46 U. Bergmann, C. R. Horne, T. J. Collins, J. M. Workmanand S. P. Cramer, Chem. Phys. Lett., 1999, 302, 119.

47 P. Glatzel, T.-C. Weng, K. Kvashnina, J. Swarbrick,M. Sikora, E. Gallo, N. Smolentsev and R. A. Mori,J. Electron Spectrosc. Relat. Phenom., 2012, DOI: 10.1016/j.elspec.2012.09.004.

48 S. A. Pizarro, P. Glatzel, H. Visser, J. H. Robblee,G. Christou, U. Bergmann and V. K. Yachandra, Phys.Chem. Chem. Phys., 2004, 6, 4864.

49 N. Lee, T. Petrenko, U. Bergmann, F. Neese and S. DeBeer,J. Am. Chem. Soc., 2010, 132, 9715.

50 K. M. Lancaster, M. Roemelt, P. Ettenhuber, Y. Hu,M. W. Ribbe, F. Neese, U. Bergmann and S. DeBeer,Science, 2012, 334, 974.

51 R. J. H. Clark, D. Lewis, D. J. Machin and R. S. Nyholm,J. Chem. Soc., 1963, 379.

52 R. J. H. Clark, J. Chem. Soc., 1964, 417.53 B. J. Brisdon, G. A. Ozin and R. A. Walton, J. Chem. Soc. A,

1969, 342.54 C. K. Jorgensen, Prog. Inorg. Chem., 1970, 12, 101.55 C. K. Jorgensen, Halogen Chemistry, Academic Press,

New York, 1967, vol. 1.56 P. C. Crouch, G. W. A. Fowles and R. A. Walton, J. Chem.

Soc. A, 1969, 972.57 G. Peng, F. M. F. Degroot, K. Hamalainen, J. A. Moore,

X. Wang, M. M. Grush, J. B. Hastings, D. P. Siddons,W. H. Armstrong, O. C. Mullins and S. P. Cramer, J. Am.Chem. Soc., 1994, 116, 2914.

58 P. Glatzel, G. Smolentsev and G. Bunker, J. Phys.: Conf. Ser.,2009, 190, 012046.

59 G. Vanko, T. Neisius, G. Molnar, F. Renz, S. Karpati, A. Shuklaand F. M. F. de Groot, J. Phys. Chem. B, 2006, 110, 11647.

60 P. Glatzel, M. Sikora, G. Smolentsev and M. Fernandez-Garcia, Catal. Today, 2009, 145, 294.

Dalton Transactions Paper

This journal is © The Royal Society of Chemistry 2013 Dalton Trans., 2013, 42, 12706–12713 | 12713

Publ

ishe

d on

09

May

201

3. D

ownl

oade

d by

UN

IVE

RSI

TÀ

DE

GL

I ST

UD

I D

I T

OR

INO

on

20/0

8/20

13 0

9:55

:59.

View Article Online