Local Structural Changes in Aluminum Isopropoxide Fluoride

Mar 26, 2009 - Different solid aluminum isopropoxide fluorides AlFx(OiPr)3−x with varying Al/F molar ratios (from 4:1 to 1:3) were synthesized follo...
0 downloads 0 Views 2MB Size
6426

J. Phys. Chem. C 2009, 113, 6426–6438

Local Structural Changes in Aluminum Isopropoxide Fluoride Xerogels and Solids as a Consequence of the Progressive Fluorination Degree R. Ko¨nig, G. Scholz, and E. Kemnitz* Institut fu¨r Chemie, Humboldt-UniVersita¨t zu Berlin, Brook Taylor-Str. 2, D-12489 Berlin, Germany ReceiVed: NoVember 20, 2008; ReVised Manuscript ReceiVed: January 15, 2009

Different solid aluminum isopropoxide fluorides AlFx(OiPr)3-x with varying Al/F molar ratios (from 4:1 to 1:3) were synthesized following the fluorolytic sol-gel synthesis route. The subsequent characterization by multinuclear (27A1, 19F, 1H, and 13C) and multidimensional solid-state MAS NMR techniques enables the characterization of local structural features in these solids. The mechanism of the fluorolytic sol-gel process deduced earlier was strongly supported by following changes of the local structures in these compounds in dependence on their composition. For the first time, the existence of four-fold fluorine-coordinated Al species (AlF4) was proven as intermediates in the case of low fluorine supply. Additionally, a plausible pathway for the formation of monofluorinated Al3(OiPr)8F · D (DMSO, pyridine) is given, which was isolated as single crystals recently. The appropriate spectra point out that similar species exist also in powders of pyridine- and DMSO-free reaction mixtures. 1. Introduction High-surface aluminum fluoride (HS-AlF3) is one of the most prominent representatives of highly disordered metal fluorides.1,2 These metal fluorides were obtained by a fluorolytic sol-gel synthesis route utilizing nonaqueous or aqueous HF and metal alkoxides. The synthesis of HS-AlF3 involves two steps: The sol-gel reaction gives, after vacuum-drying, a xerogel characterized as an aluminum alkoxide fluoride, AlFx(OR)3-x · ROH. To obtain HS-AlF3, a second reaction step is necessary, which involves a “post-fluorination” using gaseous CHClF2 or HF.3 Usually, the as-prepared metal fluorides exhibit extraordinary properties: high surface areas and small particle sizes (nanoscale). For HS-AlF3, they are accompanied by a comparatively high Lewis acidity. Additionally, the fluorolytic sol-gel synthesis route offers several possibilities to tune the surface and bulk properties of these materials and, with them, the amount, strength, and kind (Lewis/Brønsted) of acid/basic centers.4-6 For such reasons, these materials are potential candidates for several catalytic applications. However, the AlFx(OR)3-x · ROH xerogel and HS-AlF3 are X-ray-amorphous. Hence, structural information and the detection of changes of local structures are not deducible on the basis of this standard technique. On the other hand, the knowledge of the distinct local structures and following their changes in the course of the fluorination processes are key factors for understanding the origin of their high Lewis acidity. Generally, liquid and solid-state NMR techniques are approved methods to identify and understand local structures on a molecular level. Consequently, our first studies were focused on the sol-gel process and multinuclear liquid-state NMR experiments. On that basis, a plausible reaction pathway for the sol-gel fluorination was given. Starting with aluminum isopropoxide, which initially exists in solution mainly in its tetrameric and partly in its trimeric form (Scheme 1, 1 and 2), the aluminum alkoxide fluoride gel is formed by OR versus F exchange (fluorolysis), followed by the formation of * Corresponding author. Fax: (+49) [email protected].

30-2093-7277.

E-mail:

oligomeric/polymeric units built up by corner-shared Al(OR,F)6 octahedra (see Scheme 1).7 In the following work, local structures in the solid aluminum isopropoxide fluoride (the intermediate of HS-AlF3) and, for comparison, local structures in the wet jelly-like aluminum isopropoxide fluoride gel were identified. The main characteristic structural features of the AlFx(OR)3-x · ROH xerogel are preformed in the corresponding wet gel.8,9 On the other hand, Ru¨diger et al. isolated a DMSO-stabilized, partly fluorinated crystalline aluminum isopropoxide fluoride that is possibly a relative of the trimeric aluminum isopropoxide form (see Scheme 1, 3, D ) DMSO).10 Recently, we found that 3 is also formed in the presence of other donor molecules like pyridine (D ) pyridine) at low F content.11 As a consequence of the last findings, the characterization of solid aluminum isopropoxide fluorides at early fluorination states (e.g., Al/F ratios of 4:1, 2:1) without donor molecules should allow detailed insights into changes of local structures formed in the sol-gel process, starting from solid aluminum isopropoxide and ending with the AlFx(OR)3-x · ROH xerogel. Additionally, the recently developed “chemical shift scales” simplify the attribution and assignment of octahedral AlFx(OiPr)6-x species present in these solids. This holds as well for the analysis of the 19F as for the 27A1 MAS NMR spectra.12,13 Therefore, the intentions of this paper are (i) to address local structures in solids with different molar ratios (Al/F) starting from solid Al(OiPr)3, (ii) to follow their structural changes with higher degree of fluorination in the solid aluminum alkoxide fluorides, (iii) to compare the obtained results with the derived reaction pathway for the fluorolytic sol-gel process for the liquid/gel-like sols and gels,7 and (iv) to discover the origin and nature of local structures in the precursor of HS-AlF3. For these purposes, a multinuclear solid-state MAS NMR study was performed using 1H, 13C, 27A1, and 19F as sensitive spin probes, which also includes different solid-state NMR techniques, such as 27A1 triple-quantum MAS (3QMAS), 19 Ff27Al cross-polarization (CP) MAS, 19F-19F spin-exchange MAS(2DEXSY),and19F-27AlHETCORMASNMRexperiments.

10.1021/jp810190k CCC: $40.75  2009 American Chemical Society Published on Web 03/26/2009

Aluminum Isopropoxide Fluoride Xerogels and Solids

J. Phys. Chem. C, Vol. 113, No. 16, 2009 6427

SCHEME 1: Derived Pathway for the Sol-Gel Fluorination and Possible Structural Units Of Al(OiPr)3 in Solution with and without F

TABLE 1: Molar Ratios of the Initial Reaction Mixtures, Appearances, and Elemental Analysis of the Products sample a b c d e f g

i

Al(O Pr)3

AlFx(OiPr)3-x · iPrOH

nAl:nF ratio

appearance of the reaction mixture

appearance of the product

C/%

H/%

F/%

4:1 2:1 1:1 2:3 1:2 1:3

clear sol with white solid clear sol clear sol clear, slightly viscous gel opaque gel opaque, viscous gel

transparent solid transparent solid transparent solid transparent solid fine powdered xerogel fine powdered xerogel

53 49 46 42 39 32 25

9.8 8.7 8.4 8.1 7.2 6.4 5.9

3 7 12 18 22 34

2. Experimental Section i

Sample Preparation. Commercially available Al(O Pr)3 (Aldrich) was used as delivered, isopropanol was dried with sodium, and nonaqueous isopropanolic HF was synthesized through introduction of gaseous HF in dried iPrOH, as described in ref 3. Furthermore, all preparations were carried out under Schlenk conditions. Solid aluminum isopropoxide was dissolved in iPrOH (c(Al) ) 0.3 mol/L) by slightly heating to 80 °C. After cooling the solution in an ice bath, isopropanolic HF was added. Depending on the molar ratio of the educts, the reaction mixture was either a clear sol with unreacted precipitate, a clear sol, or a highly viscous opaque gel. Drying under vacuum at 70 °C (water bath) eventually leads to solid aluminum isopropoxide fluorides. Since the X-ray-amorphous products are sensitive to moisture and the evaporation of solvent and hydrolysis of remaining alkoxide groups would result in the formation of aluminum hydroxy fluorides, the samples were transferred and stored in a glovebox.

All samples discussed here are listed in Table 1. Generally, the Al/F molar ratios used for descriptive purposes in the text always correspond to the initial Al(OiPr)3 to HF molar ratio of the appropriate sols/gels, as given in Table 1. Elemental analysis was performed with a LECO CHNS-932 combustion equipment for C, H, and a fluoride-sensitive electrode (F); the latter afforded pulping of the solids into a soluble form with Na2CO3/K2CO3. NMR Methods. 19F, 27A1, 1H, and 13C spectra were recorded on a Bruker Avance 400 spectrometer using a Bruker 2.5 mm probe (Bruker Biospin) and standard zirconia rotors (Larmor frequencies of the respective nuclei at B ) 9.4 T, as well as the references, are given in Table 2). Existent background signals for 1H and 19F were suppressed either by application of a phasecycled depth pulse sequence introduced by Cory14 or by application of a rotor-synchronized echo pulse sequence. The π/2 pulse durations for 1H and 19F were determined as 2.6 (1H) and 2.2 µs (19F). 27A1 spectra were recorded with an excitation

6428 J. Phys. Chem. C, Vol. 113, No. 16, 2009 TABLE 2: General NMR Parameters 1

H C 19 F 27 A1 13

I

νL (B0 ) 9.4 T)

given chemical shifts with respect to

1/2 1/2 1/2 5/2

400.1 MHz 100.6 MHz 376.4 MHz 104.3 MHz

TMS TMS CFCl3 Al(H2O)63+

pulse duration of 1 µs (eπ/6) to ensure, preferably, quantitative excitation of all species. Typical spectral widths were 400 kHz (1H,19F) up to 1 MHz 27 ( A1). Chosen recycle delays were 0.5 to 1 s for Al and up to 10 s for 1H or 19F. In any case, spectral changes for longer delays were checked. 27 A1{19F} decoupling experiments were performed at a spinning speed of 8 kHz with a tppm decoupling sequence and a rf field of about 90 kHz. For 19Ff27Al CP MAS NMR experiments, a contact time of 300 µs was used (νrot ) 10 kHz). The CP match conditions were optimized using an amorphous aluminum fluoride sample synthesized in analogy to the way reported in ref 3. In preparation for the 19F-27Al HETCOR experiments (based on WISE procedure), the CP conditions at νrot ) 25 kHz were optimized on each sample for the rf field of the 27A1 nucleus (other conditions: contact times 160 and 300 µs, rf field 19F ) 67 kHz). 1 Hf13C CP MAS NMR experiments were performed with a 4 mm probe and contact times of 1.15 ms for the amorphous samples and 5 ms for Al(OiPr)3. 27 Al triple-quantum MAS NMR spectra were achieved using a three-pulse z-filter sequence. For the creation of the 3Q coherence, a pulse with a duration of 3 µs and, for the ((3Qf0Q) conversion, a pulse with a duration of 1.3 µs, both at a rf field strength of 125 kHz, were applied. The selective z-filter pulse lasted for 8 µs (rf field ) 15 kHz). After a twodimensional Fourier transformation, the spectra were sheared, resulting in pure absorption mode 2D contour plots. For the two-dimensional 19F-19F spin-exchange MAS NMR experiments, different contact times ranging from 1 µs to 10 ms were applied. The applied MAS frequencies, accumulation numbers (na), and, when necessary, further experimental details (TD1, time domain size for F1, and number of slices) are given in the particular spectra. If denoted, spectra were fitted with the DMFIT program in its actual version.15 3. Results Figure 1 shows the 1H MAS NMR spectra and the 1Hf13C CP MAS NMR spectra obtained for the different aluminum isopropoxide fluorides, starting with the spectra of the aluminum isopropoxide precursor (a) and ending with the spectra for the amorphous aluminum isopropoxide fluoride, xerogel (g), which is the precursor of HS-AlF3. The single samples, along with the chemical composition, as determined by elemental analysis, are given in Table 1. From the crystal structure of aluminum isopropoxide, one can deduce two different kinds of isopropoxide groups: terminal groups coordinating to one aluminum center and bridging groups linking the central Al with the outer tetrahedrally coordinated Al atoms in a ratio of 1:1 for the tetramer. Nevertheless, the 1H solid-state NMR spectrum exhibits only two main contributions at δiso ) 1.2 and 4.4 ppm, easily assignable to CH3 and CHO groups of both bridging and terminal isopropoxide groups (Figure 1, 1H, a). With rising F content, the signals become slightly narrower for intermediate F content (c and d) and, again,

Ko¨nig et al. broader for the xerogels (f and g). Additionally, a slight shift of the signals to lower field is observable; it seems more of an optical illusion when comparing a and g, but it is a bit more obvious when comparing the spectra of c, d, and e. Moreover, as earlier reported,8 along with the two signals for the CH3 and CHO groups of the isopropoxide, two further signals at δiso ) 7.8 and 10.2 ppm become visible for the final xerogel, with the Al/F molar ratio of 1:3. The intensity ratio of the signals at the 1 H chemical shifts of 1.2, 4.4, and 7.8 ppm is nearly equal to 6:1:116 as it holds for isopropanol. In comparison to the 1H MAS NMR spectra, the appropriate 1 Hf13C CP MAS spectra are much more differentiated, as also shown in Figure 1. Generally, the CH3 groups are found in the region from 20 to 30 ppm, whereas the CHO groups can be found in the region between 60 and 70 ppm. In contrast to the proton spectra, the discrimination between contributions from terminal and bridging species is, in part, possible. Starting with the 1Hf13C CP MAS spectrum of isopropoxide (a), several signals can be distinguished. Following the assignment of Abraham,17 the signals at δ13C ) 66.1, 28.2, 27.6, 27.3, 25.8, and 25.5 ppm can be attributed to CHO and CH3 groups of the bridging isopropoxide groups and the signals at δ13C ) 63.4, 29.9, 29.6, 29.1, 29.0, 28.6, and 27.6 ppm to the terminal isopropoxide groups. For low F content (b and c), the essential features of the initial isopropoxide spectrum are still present. However, broadening and additional contributions in the lowfield area of the CHO region and the high-field area of the CH3 region are already visible (see 1Hf13C CP MAS NMR spectrum of sample c). Reaching an Al to F ratio of 1 (sample d), no further signals attributable to the pure aluminum isopropoxide are detectable, and the resulting 1Hf13C CP MAS spectrum looks quite different, indicating a different “chemical situation”. The spectrum now consists of mainly two pairs of broader signals belonging together: a smaller broad signal at about δiso ) 68 ppm paired with the signal at δiso ) 23 ppm in the methyl region and, in comparison to the latter, “narrower” signals at δiso ) 63 and 27 ppm. With rising F content (samples e and f), the latter are disappearing, and for the final xerogel with the Al/F ratio equal to 1:3, only the signals in the outer region remain. To probe the local aluminum and fluorine environment and to follow the changes during fluorination, 27A1 and 19F MAS NMR experiments were performed. Figure 2 shows the obtained spectra of the aluminum isopropoxide fluorides from a to g with rising F content, allowing a comparison of the aluminum spectra with the 19F MAS NMR spectra of all samples, with a focus on the central region. Again, for both nuclei, the largest spectral changes are visible going from an Al/F ratio of 2:1 to 1:1 (see Figure 2, samples c and d). The 27A1 MAS NMR spectra of samples b and c clearly still show the characteristics of the initial aluminum NMR spectrum of the tetrameric Al(OiPr)3 (Figure 2, a). Two signals occur, one narrow and one broad, with both exhibiting a line shape split by second-order quadrupolar interactions with δiso ) 61.5 ppm and νQ ) 1860 kHz (ηQ ) 0.14) for the three tetrahedrally coordinated Al species and δiso ) 2.5 ppm and νQ ) 290 kHz (ηQ ) 0) for the octahedrally coordinated Al species.16,17 Introducing small amounts of fluorine (samples b and c), in addition to the signals of the aluminum species of Al(OiPr)3, a peak at δ27A1 ) 38 ppm with a shoulder at δ27A1 ≈ 29 ppm rises (from b to c). For the sample with the initial Al/F molar ratio of 2:1, a further very small contribution with its maximum at about -7 ppm becomes visible.

Aluminum Isopropoxide Fluoride Xerogels and Solids

J. Phys. Chem. C, Vol. 113, No. 16, 2009 6429

Figure 1. Development of the 1H MAS NMR (νrot ) 25 kHz, na ) 16-32) and 1Hf13C CP MAS NMR spectra (νrot ) 10 kHz, na ) 800 and 8 for a (na is the number of accumulations)) of the different aluminum isopropoxide fluorides, from pure Al(OiPr)3 (a) to the aluminum isopropoxide fluoride (g) with rising F content. The initial Al/F molar ratio is given within the figure.

The latter clearly becomes more and more dominant with higher F content, and for the Al to F molar ratio of 1:3 (Figure 2, g), it is nearly the only obvious contribution in the 27A1 MAS NMR spectrum recorded at B0 ) 9.4 T. The signal shape with a (slightly high-field-shifted) maximum at about δ27A1 ≈ -16 ppm is typical for aluminum in a disordered environment (asymmetric upfield decay), and the position clearly indicates, in general, six-fold-coordinated Al atoms predominantly surrounded by fluorine. However, solid-state NMR investigations concerning a more specific view on local structures in the xerogel were recently presented and have proven the existence of different defined AlFx(OiPr)6-x species.8,9 Following the changes from d to f (Al/F ratio of 1:1 to 1:2), the corresponding spectra consist mainly of a broad envelope, which, for all three compounds, exhibits similar features covering a region from 50 to -75 ppm. Apparently, this results from a superimposition of several four-fold-, six-fold-, and possibly five-fold-coordinated Al species in a disordered environment, including distributions of chemical shifts and quadrupolar coupling constants for each. With higher F content, the sharp peak at δ27A1 ) 1 ppm decreases, indicating residual Al(OiPr)6 species. Along with this, the amounts of Al species in the amorphous compounds with coordination numbers of 4 or 5 decrease also with rising F content. Interestingly, the 27A1 MAS NMR spectrum of the aluminum isopropoxide fluoride with the Al/F ratio of 1 has nearly no spinning side bands (not shown). Looking at the 19F MAS NMR spectra of the aluminum isopropoxide fluorides (Figure 2, 19F), the spectra exhibit several sharp signals for low F content (samples b and c): one signal group at δiso ) -156 ppm with downfield shoulders at about δiso ) -153 and -149 ppm; a further signal at δiso ) -188 ppm; small contributions at δiso ) -162.3, -164.4, and -171.0 ppm; and a very sharp signal at δiso ) -123.4 ppm. However,

the line widths of the main signals are a little bit smaller for the compound with the higher F content (sample c). Additionally, a broad signal “in the underground”, which is not due to background signal, ranging from about -130 to -190 ppm is already visible. The spectrum for sample d (with an initial aluminum to fluorine ratio equal to 1) has changed dramatically. The spectrum is dominated by a shouldered envelope with its maximum at δiso ) -156 ppm and several shoulders at δiso ) -138, -148, and -182 ppm. Two more “sharper” signals are visible at δiso ) -162 and -171 ppm. For comparative purposes, an 19F MAS NMR spectrum of long-aged aluminum isopropoxide fluoride with an Al/F ratio of 1 is shown (Figure 2, 19F, sample d, dotted line) exhibiting, in addition, the sharp peak at -123 ppm. With rising F content (from d to g), the envelope of all signals slightly changes; the amplitudes of the sharper signals at -162 and -171 ppm decrease, and the same is observable for the shoulders at -138 and -148 ppm. On the other hand, the relative intensities of contributions at δiso ) -161 and -172 ppm increase with rising F content. Eventually, as earlier reported, the signal of the final precursor with the F to Al ratio of 3 has three mainly visible contributions at δiso ) -161, -154, and in the area of about -172 ppm.8 For a clearer understanding of the spectral features of the intermediates, samples c, d, and f (Al/F ) 2:1, 1:1, and 1:2, respectively) were chosen for advanced MAS NMR experiments, including F-decoupled, CP MAS, 3QMAS, 19F-19F spinexchange, and 19F-27Al HETCOR experiments. For comparison purposes, these experiments were also performed with the xerogelsthe precursor of HS-AlF3, sample g. To probe the influence and proximity of fluorine to aluminum nuclei, F-decoupled aluminum MAS NMR experiments and 19 Ff27Al CP MAS NMR experiments were carried out. Together with the appropriate single-pulse spectra, these are

6430 J. Phys. Chem. C, Vol. 113, No. 16, 2009

Ko¨nig et al.

Figure 2. Development of the 27A1 MAS NMR and 19F MAS NMR spectra of the different aluminum isopropoxide fluorides, from pure Al(OiPr)3 (a) to the aluminum isopropoxide fluoride (g) with rising F content, each with a focus on the central region. The initial Al/F molar ratio is given within the figure (27A1: νrot ) 25 kHz (a, 20 kHz), na ) 15 000-12 0000. 19F: νrot ) 25 kHz, na ) 192).

shown in Figure 3. For the fluorine-decoupled 27A1 MAS NMR spectra, only minor effects are observable (see Figure 3B, spectra for d and f), meaning the more accentuated downfield shoulders at about δ27A1 ) 38 and 14 ppm. On the other hand, an enhanced resolution is observable for the CP MAS NMR experiments, and aluminum species in the proximity of fluorine can easily be identified. Contrary to the single-pulse experiment, the 19 Ff27Al CP MAS NMR spectrum of sample c shows only two contributions, one symmetric peak at δ27A1 ) 38 ppm and one less intense peak at δ ≈ -6 ppm with an asymmetric upfield decay; the signals of aluminum isopropoxide species are no longer visible. Furthermore, for both AlFx(OiPr)3-x · iPrOH samples with Al/F ratios of 1:1 and 1:2 but not for the final xerogel, the signal for the species at δ27A1 ) 38 ppm and a further one in the region of about 14 ppm are detectable. Also, for sample d, the sharp signal at about 1 ppm is no longer visible. Comparing the 19Ff27Al CP MAS spectra of the latter three samples in the octahedral region, at first view, for sample g, the maximum of the single signal can be found at δ27A1 ) -12 ppm. Looking at the spectra of the compounds with lower F content, a shoulder at δ27A1 ) -2 ppm is obvious, indicating at least a further octahedral AlFx(OiPr)6-x species. The intensity of the latter is decreasing with rising F content. In contrast to the single-pulse spectra, for all samples, the most upfield part is considerably less pronounced in the CP MAS NMR spectra. A further possibility for resolution enhancement of the 27A1 MAS NMR spectra is the application of MQMAS techniques, separating single species. Theoretically, utilizing the sheared two-dimensional spectra, the isotropic chemical shifts of the single species can be taken from the projection of the species ridge along the F1 dimension, whereas the projection along the

F2 dimension displays the “normal” 1D spectrum, including the second-order quadrupolar line shapes of all species. The extraction of the quadrupolar parameters (δiso, νQ) can be performed in a graphical way, as presented by Amoureux,18 or by using the particular (F1;F2) coordinates of the barycenter of the observed peak and calculating the parameters [δiso ) (17δF1 + 10δF2)/27 (1) and PQ ) (85)/900 · νL · (δF1 - δF2)1/2 (2) with νQη ) PQ · 3/[(2I - 1) · 2I] (3)]. The obtained z-filtered 3QMAS spectra for samples d and f are shown in Figures 4 (sample d) and 5 (sample f). The advantage of an easier extraction of a single species at moderate magnetic fields (B0 ) 9.4 T) is, in part, visible. A summary of the identified species in the two different isopropoxide fluorides is given in Table 3. For samples d and f, a variety of species can be identified, including mainly species with chemical shifts typical for fivefold- and six-fold-coordinated Al species. In sample d (Al/F ratio of 1), a residual amount of the symmetric AlO6 unit in Al(OiPr)3 is also detected (Table 3, d, species 4), characterized by comparable quadrupolar parameters (δiso ) 2.3 ppm, νQ ) 239 kHz), as published by Abraham.17 However, the detected intensities of some of the assigned species are comparably low. Only a few can be clearly and unambiguously assigned in the appropriate 3QMAS spectra. The latter are given in Table 3 as bold text (species 1-4). Nevertheless, some indications exist (see the appropriate 27A1 MAS NMR single-pulse spectra, which are displayed in Figure 2) that the peaks that are assigned with a higher degree of uncertainty to predominantly four-fold-coordinated Al species (and which are given as italic entries in parentheses, species

Aluminum Isopropoxide Fluoride Xerogels and Solids

J. Phys. Chem. C, Vol. 113, No. 16, 2009 6431

Figure 3. (A) 19Ff27Al CP MAS NMR spectra (solid line) in comparison to the 27A1 single-pulse spectra (dotted line) for samples c, d, f, and g (na for the CP spectra ) 16 000-60 000). For additional orientation, guiding lines at δ27A1 ) 38, 14, and -12 ppm are indicated. (B) 27A1{19F} MAS NMR spectra of samples d and f (na ) 50 000). Asterisks denote spinning side bands.

Figure 4. Sheared 27A1 3QMAS NMR spectrum of sample d (Al/F ) 1:1), along with the positive projections along F2 and F1. The region of the less intense peaks is zoomed-out. For orientation, some guiding lines and the species corresponding to Table 3 are given. Bold: species unambiguously identified. Italic: possible position of a species (νrot ) 25 kHz, TD1 ) 256, na ) 6816 (TD1 ) time domain size in F1, number of slices)).

(5), (6), and (7), in Table 3) are not only noise and are assignable to further species. That means species with isotropic chemical shifts in the typical region of tetrahedrally coordinated Al species could not unambiguously be detected by 3QMAS. However, especially the latter can be identified by the application of MAS NMR methods at higher magnetic fields, due to the reduction of

quadrupolar line broadening and a general enhanced resolution.19 The detected low intensities in the 3QMAS NMR spectra in this case are due to the high quadrupolar frequencies of the corresponding species, so additionally, higher rf field strengths in the MQMAS experiment would be necessary to excite these species in a quantitative manner.20 This fact is accompanied by the general low occurrence of each single species in the isopropoxide fluorides and the before mentioned distribution effects (chemical shifts and quadrupolar frequencies). Interestingly, with higher F content, Al sites in ordered surroundings with nearly crystalline behavior can be found (see Figure 5, inset and Table 3, sample f, species 2) exhibiting a second-order quadrupolar line shape. A simulation of the sum of the surrounding of its ridges (approximately (5 ridges) gives a species with an isotropic chemical shift of about δiso ≈ 0 ppm and a quadrupolar frequency of about 1000 kHz (η ≈ 0) with some degree of uncertainty (for details, see Figure 5). To identify the neighborhood of several fluorine species in the amorphous alkoxide fluorides d and f, 19F-19F spinexchange MAS NMR experiments were performed. The appropriate spectra are given in Figures 6 (sample d) and 7 (sample f). A magnetization transfer from one species to another at short contact times causes cross-peaks in the two-dimensional spectra, which is only possible if the considered fluorine nuclei are in the sterical vicinity. For the aluminum isopropoxide fluoride d (Al/F ratio of 1), even at longer contact times, nearly no magnetization transfer between the different F-sites is observable. The two-dimensional 19F-19F spin-exchange MAS NMR spectrum (Figure 6) is only characterized by the diagonal ridge, meaning that several fluorine species are predominantly not in proximity to the others and are more or less isolated. For the AlFx(OiPr)3-x compound with the Al/F ratio equal to 1:2 (f), the corresponding 19F-19F EXSY MAS NMR spectrum (Figure

6432 J. Phys. Chem. C, Vol. 113, No. 16, 2009

Ko¨nig et al.

Figure 5. Sheared 27A1 3QMAS NMR spectrum of sample f (Al/F ) 1:2), along with the positive projections along F2 and F1. For orientation, some guiding lines and the species corresponding to Table 3 are given. Bold: species unambiguously identified. Italic: possible position of a species. The insets show the sum of the approximately (5 slices around species 4, along with a possible decomposition and estimated values for the quadrupolar parameters (νrot ) 25 kHz, TD1 ) 256, na ) 5040).

TABLE 3: 27Al MAS NMR Parameters of Samples d and f As Obtained by 3QMAS sample

species

d

1

δ (F1)/ppm δ (F2)/ppm PQ/MHz νQη/kHz δ (iso)/ppm

36.9 14.3 5.1 762 28.5 Al: CN ) 5

f

1

δ (F1)/ppm δ (F2)/ppm PQ/MHz νQη/kHz δ (iso)/ppm

43.6 14.3 5.8 867 32.7 Al: CN ) 5

2

3

4.0 -2.3 -6.7 -11.0 3.5 3.2 524 473 0.0 -5.5 Al: CN )

2

3

16.6 1.3 -35.0 -11.0 7.7 3.7 1151 562 -2.5 -3.3 Al: CN )

4 3.4 0.5 1.8 273 2.3 6

(5)a

(6)a

(7)a

44.4 43.4 30.1 36.5 10.5 7.7 3.0 6.1 5.1 450 919 758 41.5 31.2 21.8 Al: CN ) 4 and/or 5

4

(5)a

-3.4 -12.2 3.2 475 -6.7 6

46.5 34.6 3.7 553 42.1 Al: CN

(6)a 29.2 14.7 4.1 610 23.8 ) 4 and/or 5

a

Italic numbers in parentheses: species with low intensity; identification with higher uncertainty.

7) features a symmetric shape concerning the diagonal ridge: the F-species with the signal at δiso ) -154 ppm “sees” the F-site with the signal at -161 ppm, which is in sterical vicinity to the species with the resonance at about -172 ppm. Similar results were found for sample g (not shown here).21 Furthermore, delayed rotor-synchronized echo experiments were performed utilizing different spin-spin relaxation times (T2) of the particular species. Selected spectra are displayed as insets in Figures 6 and 7. For sample d, a slight resolution enhancement was achieved; the maximum of the primary signal at δiso ) -156 ppm is less pronounced, but all F-sites are also present at long delays, having similar T2 relaxation times, except the very sharp signals at -162 and -171 ppm. For sample f, the fluorine species at δiso ≈ -172 ppm is also present at longer delays (see Figure 7, inset). A similar behavior was recently observed for terminal F-sites present in highly disordered highsurface AlF3 samples, resonating in the high-field area from

-190 to -210 ppm (not shown here; the latter nicely agree with theoretically predicted chemical shifts of terminal F-sites in such compounds22,23). Additionally, a residual amount of the species with the narrow signal at the chemical shift at δiso ) -162 ppm is still present. Expecting a correlation of fluorine with aluminum species, some 19F-27Al HETCOR MAS NMR experiments were performed with samples c and f. Figures 8 and 9 show the obtained spectra. For sample c (Al/F ratio ) 2:1), a clear correlation of the peak for the aluminum species at δ27A1 ) 38 ppm and the fluorine peaks in the region of δiso ≈ -155 ppm are visible. However, the dashed lines mark “interesting” areas in this spectrum (Figure 8): First, no correlation peak is observable for the fluorine species at -190 ppm, and second, the six-foldcoordinated aluminum species at δ27A1 ) -6 ppm has a very low intensity; no direct correlation peak is identifiable, too. The 19 F-27Al HETCOR spectrum obtained for sample f (Figure 9) is rather featureless, and the several F-sites with their maxima at δiso) -154, -161, and -172 ppm overlap and correspond with Al species in the “six-fold area”. However, as the signals are not resolved very well, discrimination is hardly possible. No contributions/correlations to four-fold or five-fold Al species are visible. However, the optimal 19Ff27Al CP-matching conditions for each species contained (which are the prerequisite for the WISE-based HETCOR program) are also dependent on the quadrupolar frequency. This behavior is nicely demonstrated in Figure 10. One optimum for the species at δ27A1 ≈ 38 ppm could be found for high applied rf fields (low amounts for the attenuation given in dB), where the octahedral species is not visible yet. For the experiment (the spectrum is given as Figure 8), those CP conditions were chosen, which should allow the detection of both Al species. However, due to the fact that the two Al-F species strongly differ in their characteristics (quadrupolar parameters, coordination number, and chemical environment), their relaxation behavior is quite different. The species resonating at -6 ppm has a fast relaxation behavior and is only visible in the first few slices of the HETCOR spectrum obtained

Aluminum Isopropoxide Fluoride Xerogels and Solids

J. Phys. Chem. C, Vol. 113, No. 16, 2009 6433

Figure 6. 19F-19F 2D spin-exchange MAS NMR spectrum of sample d (Al/ F ) 1:1) at νrot ) 25 kHz and a contact time of 10 ms (TD1 ) 128, na ) 240). Inset: 19F rotor-synchronized delayed echo experiments are shown for the same sample (νrot ) 25 kHz). A: Additional rotor periods before echo detection, 20. B: 40 (na ) 256).

Figure 7. 19F-19F 2D spin-exchange MAS NMR spectrum of sample f (Al/F ) 1:2) at νrot ) 25 kHz and contact time of 10 ms (TD1 ) 128, na ) 64). Inset: 19F rotor-synchronized delayed echo experiments are shown for the same sample (νrot ) 25 kHz). A: Additional rotor periods before echo detection, 20. B: 40 (na ) 256 and 1024).

by a Fourier transformation only in the F2 dimension (but not in F1; spectrum is not shown here). After the Fourier transformation in F1, the information of the species is no longer visible in the final spectrum. Consequently, the observed 19F-27Al HETCOR spectrum of sample c is dominated by the species resonating at 38 ppm, whereas the cross-peak for the species at -6 ppm is only hardly visible. Additionally, the detection is limited, due to the low amount of the species. These general differences between the two species can also be deduced comparing the observed line shapes in the single-pulse spectra (Figures 2 and 3, c). The signal of the species at 38 ppm is narrower and much more symmetric (fwhm ≈ 900 kHz). Instead, the second line at δ27A1 ) -6 ppm exhibits a Czjzektype line shape and finally is much more broad (fwhm > 2.5 kHz); this hints at the different chemical nature of both Al species. That means a complete HETCOR spectrum for a sample containing several AlFx(OiPr)CN-x species (CN ) 4, 5, 6), each with its own set of quadrupolar parameters, would require more than one experiment. This fact is accompanied by the necessary huge number of accumulations (amorphous samples), which is multiplied for the two-dimensional spectra by the number of experiments (F1 dimension).

4. Discussion Taking all results into account, we can identify three stages of the sol-gel fluorination process, which are discussed in the following paragraphs. The first stage is primarily represented by samples b and c, with Al to F molar ratios higher than 1 (meaning low F content). Samples d and e, at the second stage, are discussed separately as they mark for the solids the changeover to the third stage, the aluminum isopropoxide fluoride xerogels, samples f and g. Taking only the development of the 1H MAS and the 1Hf13C CP MAS NMR spectra into account, this classification can, in part, be deduced, and the 1H MAS NMR spectra have to be discussed together with the 1Hf13C CP MAS spectra. The spectral changes (peak maxima, line forms) deducible only from the 1H MAS NMR spectra are only minor, and the superimposition of the several signals makes it difficult to discuss them separately (see Figure 1, 1H). However, when we observed the 1 Hf13C CP MAS NMR spectra of samples b and c, they still clearly show the peak pattern obtained for the pure solid aluminum isopropoxide a, indicating that not all the Al(OiPr)3 was consumed, and aluminum/fluorine ratios of 4:1 and 2:1, respectively, mean that every tetrameric aluminum isopropoxide

6434 J. Phys. Chem. C, Vol. 113, No. 16, 2009

Figure 8. 19F-27Al HETCOR MAS NMR spectrum of sample c, along with the positive projections for each dimension (Al/F ) 2:1, TD1 ) 128, na ) 1536, νrot ) 10 kHz, contact time ) 300 µs, rf27A1 ) 15 kHz (power level pl1 ≈ 20 dB). As additional markers, guiding lines at δiso ) -190 (19F) and δ27A1 ) -6 ppm (27A1) are given.

Figure 9. 19F-27Al HETCOR MAS NMR spectrum of sample f, along with the positive projections for each dimension (Al/F ) 1:2, TD1 ) 128, na ) 3528, νrot ) 25 kHz, contact time ) 160 µs, rf27A1 ≈ 10 kHz. The lines indicate possible cross-peaks of the superimposed species.

molecule could react with one (or two) HF molecule. Additionally, small contributions in the upfield area of the methyl region (≈23 ppm) and the downfield area of the CHO region (≈67 ppm) are visible in the spectrum of sample c. Going to a higher degree of fluorination (samples d-g), a transformation of the general spectral pattern is observable. The sharp, divided peaks of the initial aluminum isopropoxide are no longer present. Instead, two peak pairs, as mentioned in the Results, dominate the spectra. The assignment of the latter is possible comparing the observed chemical shifts with chemical shifts measured for some reference points in this system containing iPrX (O, N) species. These samples are listed in Table 4. As a general trend, the observed chemical shifts of the CH3 groups move to lower chemical shifts, in contrast to a shift of δiso to higher values for the CHX species. This trend is enforced by the higher the formal coordination degree of X (formal X is coordinated by C, H, or Al in this system). Known relevant examples are, for instance,

Ko¨nig et al.

Figure 10. Two-dimensional illustration of an optimization of an 19 Ff27Al CP MAS NMR experiment. In this case, for sample c (Al/F ) 2:1) at a constant contact time and rf field of the F nucleus ) 67 kHz, varying the rf field of the Al nucleus (expressed as pl1). More than one maximum of the optimal CP conditions for the different species is shown. For the 19F-27Al HETCOR experiment (c, Figure 8), the CP conditions that should match both Al species were chosen.

the 13C shifts observed for the protonated and unprotonated i PrNH224 and the observed chemical shifts for the different iPrO species of the tetrameric Al(OiPr)3 (see Table 4 and ref 7). Interestingly enough, the same behavior can be observed for the iPrOH · HF system used for the synthesis, supporting a partial existence of an equilibrium with a protonated form, iPrOH2+25 as also described by Quarterman.26 Following that, we assign the inner peaks (Figure 1, samples d-f, δiso ) 27 and 63 ppm) to terminal iPrO groups predominantly coordinating to Al, whereas the outer ones (δiso ) 23 and 68 ppm) we assign to bridging isopropoxide groups. These bridging iPrO groups were not only described by groups linking two Al species (as in the tetrameric Al(OiPr)3), but even more by iPrOH species coordinating Al, which is deducible from the integration of the single peaks of the observed 1H MAS NMR spectrum, as displayed in Figure 1 and Table 4 (sample g, 1H).16 With the formation of the network in the xerogels, the amount of the terminal groups decreases, and for the final xerogel, the change to iPrOH molecules incorporated in the gel network is nearly complete. As described, iPrOH molecules interact via H bonds with the network (with other iPrOH molecules or Al-F-Al bridges).8,12 Possibly, the slight downfield shift of the signals in the 1H MAS NMR spectra supports the transition (Figure 1, 1H, from c to d) from terminal iPrO groups to coordinating solvent molecules, in general, and gives a hint that the fluorination starts at bridging isopropoxide molecules, as proven earlier.7 Eventually, the spectra support the finalization of the xerogel network with strong H-bonded solvent molecules. With respect to the 27Al and 19F MAS NMR spectra, the proposed model (three stages for the sol-gel fluorination process) is supported by the following findings. As mentioned above, the greatest change can be observed by introducing more and more fluorine and changing the Al to F molar ratio to 1 (see Figure 2). For the samples with less fluorine content (b and c), the spectral features of the initial Al(OiPr)3 are still present in the 27A1 MAS NMR spectra, clearly identifiable by the signals for their AlO6 and AlO4 species split by secondorder quadrupolar interactions (Figure 2, 27A1, a-c).17 However, additionally, a rising signal at 38 ppm is detected, which is

Aluminum Isopropoxide Fluoride Xerogels and Solids

J. Phys. Chem. C, Vol. 113, No. 16, 2009 6435

TABLE 4: NMR Parameters Obtained for Some Reference Points δ13C/ppm entry i

(CH3)2CXH-

δ1H/ppm CH3

e

OH

PrOHa PrOH · HFa i PrNH2b i PrNH3+b Al(OiPr)3

63.5 66.1 43.0 46.5

25.1 23.6 26.5 23.2

terminal OiPr bridging OiPr

63.4 66.1

in solid state 27.6-29.9 25.5-28.2 -

i

i

terminal O Pr bridging OiPr AlFx(OiPr)3-x · iPrOH (wet gel)a AlFx(OiPr)3-x · iPrOH (wet gel)d AlF2.3(OiPr)0.7 · z iPrOH (xerogel)d

63.2 66.1 63.5 63.7 64.4, 67.6, 68.8

27.7, 25.3, 25.1 25.2 23.0,

5.8, d, 1H 9.0, s, 2H -

in solutionc 27.9 26.4 5.8, br, 1H 5.0, br, 1H 24.6 10.3, 7.8, br, 1H

CH3e

CHORe 4.4, d, sep, 1H 4.5, sep, 1H -

1.6, d, 6H 1.7, d, 6H -

4.4, br

1.2, br

4.3 sep 4.5 sep 4.4, br, 1H 3.8, br, 1H 4.4, br, 1H

1.1, 1.3, 1.6, 1.0, 1.2,

d; 1.1, d d; 1.5, d br, 6H br, 1H br, 6H

formal coordination number of X 2 “3” 3 4

2 3 2 3 2 2 “3”

a

b

Secondary standard CDCl3 (in a lock-in tube or for HF-containing samples outside of a 4 mm PP tube, which contained the HF solution). See ref 24. c Solved in CDCl3. d See ref 8 (measured with solid-state spectrometer). e Peaks: s, singlet; d, duplet; sep, septet; br, broad.

provoked by a tetrahedrally coordinated aluminum site in the proximity to fluorine, as evidenced by an 19Ff27Al CP MAS NMR experiment (see Figure 3A, sample c). The signals for the AlO4 and AlO6 species are no longer detected, so we deduce that the Al species at 38 and -6 ppm are assignable to Al-F species with Al-F bonds. For the assignment of the species at 38 ppm, the assumption of a five-fold Al species would also be reasonable. For instance, the AlO5 species have chemical shifts in the region of about 38 ppm.17 On the other hand, signals of five-fold AlFx(OiPr)5-x species should be more and more highfield-shifted with the higher degree of Fx, similar to the known trends for octahedrally coordinated AlFxO6-x species.13,27 In addition, an AlO5 species resonating in the region of about 38 ppm should (i) not be detected in the 19F-27Al CP-based spectra and (ii) possibly exhibit another line shape (rather Czjzek-type than symmetric). The 19F MAS NMR spectra of b and c are dominated by very sharp signals (fwhm less than 1 kHz), which indicate ordered (crystal-like) local structures. The chemical shifts of the most intense signals are around δiso ) -155 and -190 ppm. An 19F-27Al HETCOR experiment (see Figure 8) unambiguously shows that the Al species resonating at δ27A1 ) 38 ppm correlates with the signal group at -155 ppm. The dotted lines in Figure 8 mark “interesting” areas of the HETCOR spectrum, and following them, no correlation peak was detectable for the F-species at -190 ppm. However, our own results,11 as well as the findings of Lacassagne for similar species occurring in NaF/ AlF3/Al2O3 melts, support an assignment of the peak at -190 ppm to terminal fluorine sites of different four-fold, five-fold, and six-fold Al species.28,29 Furthermore, different AlFxO4-x species could be identified, including F3Al-O-AlF3 (δ27A1 ) 50 ppm) and FAl-µO2-AlF2 (δ27A1 ) 59 ppm); for the AlF4 species, a chemical shift of 38 ppm is given.29 Therefore, we assign the signals at 38 ppm (Al) and -155 ppm to one species: AlF4 existent in these solids. The signal at δiso) -190 ppm is assumed to correspond to terminal F-sites, whereas the signal at -155 ppm possibly corresponds to bridging F-sites. This species is not an isolated ionic form, but it is involved in the structure. The typical band for the AlF4 anion at 785 cm-1, as described by Herron,30 was not detectable in the appropriate IR spectra. The latter just show the change of the IR spectra from that of pure Al(OiPr)3 to that of the xerogel, the aluminum isopropoxide fluoride (Al/F ) 1:3), which are shown in ref 3. However, Kao found, in the course of investigating dealumination processes of zeolites utilizing NH4F, species with

corresponding signals at δ27A1 ) 50 ppm and δ19F ) -173 and -182 ppm and assigned these to tetrahedrally coordinated Al-F species.31 Following theoretical investigations in the Al-F-O system made by Liu, F-sites of a four-fold-coordinated AlF species would have chemical shifts of δiso ) -197 ppm for bridging F-atoms and δiso ) -216 ppm for terminal species.22 Hence, we conclude that, for all F-sites of the AlF4 species existent at the early stages of the fluorination process, a strong involvement in H bonds, along with strong deshielding effects, affects the isotropic chemical shift. The signal at δiso ) -123 ppm (see Figure 2, 19F, b, c, and d (dotted)) is due to the formation of F- complexes with different counterions (H+, Na+) incorporated,31-33 resulting from the reaction of the sol with the glass wall. This process takes a longer period of time, as illustrated by 19 F spectra (see Figure 2, d) for an isopropoxide fluoride obtained from a few days aged sol (solid line) and a longer aged sol (dashed line). However, a coincidental formation of Na2SiF6 complexes (chemical shift from about -131 (aqueous solution)32 to -152 ppm (solid state)33) could not unambiguously be proven. Further small peaks at δiso ) -162 and -171 ppm become more intense and are best visible for sample d; the origin of these two will be discussed later. With a molar ratio of Al/F ) 1 in the aluminum isopropoxide fluoride, the beginning of an AlFx(OiPr)y network is formed. Nevertheless, in this solid, a huge variety of different AlFx(OiPr)CN-x species exist with CN ) 4, 5, and 6. They form, indeed, “bigger” units, but among themselves, they are more or less isolated. The broad enveloped signals in the appropriate 27A1 and 19F MAS NMR spectra (Figure 2, 27A1 and 19F, sample d) indicate this huge variety. The “27Al envelope” of sample d governs an area ranging from 60 to -25 ppm and is caused by the superimposition of several species, each accompanied by a superimposition of its quadrupolar parameters. This finding is at least supported by the 27A1 3QMAS experiment, which is shown in Figure 4. The formation of these bigger units starts in the very beginning. Both in the 19 F MAS NMR spectrum (Figure 2, 19F, sample c, broad underground), as well as in the appropriate 27A1 MAS NMR and 19Ff27Al CP MAS NMR spectra (Figure 2, 27A1 and Figure 3A, sample c), octahedral AlFx(OiPr)6-x species are evidenced. Following the 27A1 MAS NMR chemical shift correlation for octahedral AlFxO6-x species, these units formed consist of AlF3(OiPr)3 and AlF4(OiPr)2 species (sample c) and, later on, of AlF5(OiPr) species (sample d).13 This assignment is in

6436 J. Phys. Chem. C, Vol. 113, No. 16, 2009

Ko¨nig et al.

SCHEME 2: Extension of the Pathway Presented in Scheme 1 Based on the Results Obtained Here Comparing Possible Species in Sol/Gel State and in the Corresponding Solids. Three Fluorination States are Deducible: State 1, Al/F > 1; State 2, Al/F ≈ 1; Schemes of the Third State (Al/F ) 1:2; 1:3) were Given Earlier8

agreement with the observed position of some signals in the 19 F MAS NMR spectra (region from -140 to -163 ppm), comparing them with the 19F MAS NMR chemical shift correlation reported earlier.12 Nevertheless, the amount of this species in sample c is very low, and the distorted surrounding in irregular units provokes a fast relaxation behavior. Therefore, in the 19F-27Al HETCOR spectrum (Figure 8), no cross-peaks for the species at ≈ -6 ppm (see dashed line, Figure 8, 27A1 dimension) can be observed, although this species is present in the first slices (not shown here). It is presumable that the “peak” at -6/-165 ppm (Al/F) might be more than spectral noise. Eventually, for the first stage (F/Al ratio less than 1), we conclude the formation of AlF4 species that are incorporated in

a network structure with the consequence that tetrameric Al(OiPr)3 is still present. The tetramere stands in solution in equilibria with the trimeric form (see Scheme 2, state 1, tetrameres 1 and 2). These trimeres can be partly fluorinated, and if donor molecules are present, they are stabilized and can be isolated. The 19F MAS NMR spectra of pure compound 3 (Scheme 1, D ) pyridine) exhibit one single sharp peak at δiso ) -160 ppm for the terminal F-site of the AlF1O4N octahedron, whereas the 19F MAS NMR spectrum of an amorphous compound with pyridine as solvent (Al/F ) 1) shows, among others, signals at -160 and -173.5 ppm.11 Looking at the 19F MAS NMR spectrum of the compound with the molar ratio (Al/F ) 1:1) of the system under investigation (without further

Aluminum Isopropoxide Fluoride Xerogels and Solids donor molecules) (see Figure 2, sample d), two sharp signals at δiso ) -162 and -171 ppm are obvious. Further similar relationships can be found for the appropriate 27A1 MAS NMR spectra (not shown here). As mentioned above, these sites are already visible as small peaks in the 19F spectra of samples b and c. Furthermore, they are also present in the spectra at long delays while performing rotor-synchronized echo-MAS NMR experiments, which are shown as the inset in Figure 6 (graph B). With the consequence that these species must have a longer relaxation behavior, an ordered “crystalline-like” local structure with involvement of species similar to 3 (Scheme 2, state 1) seems plausible; however, it has not been isolated yet. The species with a signal maximum at -156 ppm has a faster relaxation behavior; nevertheless, after a delay of an additional 20 rotor periods, a lot of F-sites are still present (graph A). Bearing this fact in mind, it is easy to understand that nearly no cross-peaks are observable in the 19F-19F spin-exchange MAS NMR spectrum carried out with sample d, even at long contact times (10 ms). That means that most of the species in this solid are more or less isolated; no proximity of the certain F-sites to each other can be stated. Otherwise, a significant magnetization transfer and cross-peaks would be observed (see Figure 6 and Scheme 2, state 2). Furthermore, nearly no spinning side bands are observable in the 27A1 MAS NMR single-pulse spectrum, which is another indicator for the highly disordered and distorted structure. Nonetheless, from the 27Al 3QMAS NMR spectrum (Figure 4) and the 19Ff27Al CP MAS NMR spectrum (Figure 3, d), a conception of the present species is deducible. The amount of pure AlF4 species is only minor regarding only the 19 Ff27Al CP spectrum, but from the 3QMAS NMR spectrum (see Figure 4 and Table 3), a set of AlFx(OiPr)4-x, AlFx(OiPr)5-x, and AlFx(OiPr)6-x species (for the latter x ) 3-5) is expected. The information gained from 19F decoupling techniques is limited as the signals are not narrowed effectively enough (see Figure 3B), except for a small accentuation of the sites also visible in the appropriate CP spectra. Finally, moving to the third stage, local structures and characteristics of the solids change once more; a more and more stabilized network is formed with Al/F ratios equal to 1:2 and 1:3. The local structures found are predominantly preformed in the wet gels and remain in the corresponding xerogels. The amount and spread of four-fold and five-fold species decreases, ending up with six-fold AlFx(OiPr)6-x species (x ) 4 and 5), as deduced from the chemical shift correlation graphs.12,13 Interestingly enough, further apparently ordered structures are also attributable to the typically X-ray-amorphous xerogels, as stated earlier,8 and are clearly evidenced in the 27A1 3QMAS NMR spectrum of sample f. Species 2 (Figure 5, inset and Table 3, sample f) clearly exhibits a pattern split by second-order quadrupolar interactions, with a quadrupolar frequency in a range typical for pure complex crystalline AlF6 units containing fluorides organized in chains or layers.34,35 Nonetheless, it has to be noticed that the quadrupolar frequencies of the species observed here are comparatively high. Accompanied by the low abundance of some species in these solids, the expected intensity of the peaks in the 27A1 3QMAS spectra may be very small as the excitation efficiency of an MQMAS experiment depends also on the quadrupolar parameters.20 Therefore, the data of some of the identified species are given in italics (see Table 3), meaning that the assignment of these species is not unambiguous. For a more definite identification, MAS NMR experiments at higher magnetic field strengths will be utilized because quadrupolar broadening will be reduced

J. Phys. Chem. C, Vol. 113, No. 16, 2009 6437 and the obtained spectral resolution is enhanced. As mentioned above, for the final gels (Al/F ) 1:3), we found a stabilized network structure consisting mainly of AlFx(OiPr)6-x octahedra, which is already preformed in the wet gel. Consequently, in contrast to the 19F-19F spin-exchange NMR experiment of sample d (Figure 6, Al/F ratio of 1), cross-peaks were found for the main contributions in the corresponding experiment with sample f (see Figure 7, cross-peaks for species at -154 and -161 ppm and for -161 and -172 ppm, Al/F ratio of 1:2). That means that these sites are at least in sterical proximity to each other and are probably involved in the same local structures. Finally, rotor-synchronized NMR echo experiments (shown as the inset in Figure 7) reveal that the site at -171 ppm has a considerable other spin-spin relaxation behavior and resists also at long delays. Considering all characteristics of this signal (chemical shift, fwhm in comparison to the others, and relaxation behavior is similar, as observed for terminal F-sites in the post-fluorinated samples), these point to an assignment to terminal fluorine sites of AlFx(OiPr)6-x octahedra (x ≈ 4). Nevertheless, a superimposition with F-sites from AlF6 octahedra can not be excluded. Both the 27A1 MAS NMR spectra and the 19F MAS NMR spectra show a diversity of different species existent in the xerogel. However, the 19F-27Al HETCOR experiment (see Figure 9) exhibits only a broad cross-peak; different sites are not resolved very well and interfere with each other. This may be due to the general difficulties connected with HETCOR spectra and this kind of solids. 5. Conclusions Comparing the results of this study with the results obtained for the aluminum alkoxide fluoride sols and gels presented earlier7,8 and given in Scheme 1, some differences are worked out, especially regarding the early fluorination states. The main findings are additionally summarized in Scheme 2. (i) As the results obtained for samples b and c show, the early formation of tetrahedrally coordinated AlF4 species can be demonstrated for these solids. These species do not seem to exist in the corresponding sols. Instead, it is plausible that these species are stabilized as solvated AlFx(HOiPr)6-x3-x (x ≈ 4) species in the sols. Drying results in a partial loss of the solvent molecules and in the formation of “incorporated” AlF4 species. Additionally, further linking processes of the AlF4(HOiPr)2species lead to the formation of bigger units (the beginning of a gel-like network), which, as a consequence, are predominantly built of AlFx((H)OiPr)6-x octahedra (x ) 3-5). (ii) At early fluorination states, nonconverted Al(OiPr)3 molecules exist in the sols and gels to some degree, either in the tetrameric or in the linear trimeric form. In the following, the latter may be partly fluorinated, and if stabilizing donors (D) are accessible, even crystals of Al3(OiPr)8F · D can be isolated (Scheme 2, state 1, 3). Nevertheless, for aluminum isopropoxide fluorides resulting from sols without donor molecules, the formation of ordered similar Al3(OiPr)8F · iPrOH species is presumable, as evidenced by the corresponding sharp signals centered at -162 and -171 ppm (19F). (iii) The rise of the F content (molar ratio of Al/F ≈ 1) leads to an irregular, strongly disordered solid. The corresponding sol consists of a loose network imaginable in solution, as presented earlier in ref 7. Vacuum drying leads to the cleavage of iPrOH molecules coordinated to Al species existing in these sols and gels, forming a variety of four-fold AlFx((H)OiPr)4-x, five-fold AlFx((H)OiPr)5-x, and six-fold AlFx((H)OiPr)6-x species, as identified by 3QMAS. The units are sterically separated and, as in the 19F-19F EXSY NMR experiments, nearly no spinexchange could be observed (see Scheme 2, state 2).

6438 J. Phys. Chem. C, Vol. 113, No. 16, 2009 (iv) Finally, this gel-like network is strengthened and stabilized by cross-linking. Vacuum drying leads to xerogels that consist predominantly of six-fold-coordinated AlFx((H)OiPr)6-x species (x ) 3-5).8,9 Coordinating solvent molecules are stabilized by the formation of H bridges. The local structures of the xerogels are preformed in the wet gel. Nevertheless, ordered local structures are also observable for the xerogels. Structural models for both the gel-like wet isopropoxide fluoride and the dry xerogel were presented earlier.7,8 As stated, the results obtained for the different solid aluminum isopropoxide fluorides point out that a variety of different AlFx(OiPr)CN-x (CN ) coordination numbers 4, 5, and 6) exist in these kinds of solids. This fact makes it difficult to obtain well-resolved spectra and to unambiguously assign the certain species. The study and analysis of these solids at higher magnetic fields should provide more detailed information about the species and will be part of future work. Probably, these solids offer the opportunity to experimentally deduce a complete chemical shift scale for four-fold-, five-fold-, and six-fold-coordinated Al species in a mixed fluorine/(alk)oxide environment, depending on the degree of fluorination. Acknowledgment. The authors thank the Deutsche Forschungsgemeinschaft (project Ke 489/29-1), and the EU (sixth Framework Programme, FUNFLUOS, Contract No. NMP3-CT2004-5005575) for financial support. A. Pawlik and Prof. Dr. C. Ja¨ger (Bundesanstalt fu¨r Materialpru¨fung und -forschung, Richard Willsta¨tter-Str. 11, D-12489 Berlin) are kindly acknowledged for providing measurement time, fruitful discussions, and support with continuative MAS NMR experiments. J. Koch and A. Dimitrov are noticed for continuative experiments concerning the presence of pyridine. Dr. U. Hartmann, Dr. A. Zehl, U. Ka¨tel, and S. Ba¨ssler are recognized for performing the elemental analysis. Dr. D. Heidemann is acknowledged for fruitful discussions. References and Notes (1) Kemnitz, E.; Gross, U.; Ru¨diger, S.; Shekar, C. S. Angew. Chem., Int. Ed. 2003, 42, 4251–4254. (2) Ru¨diger, S.; Kemnitz, E. Dalton Trans. 2008, 1117–1127. (3) Ru¨diger, S.; Eltanany, G.; Gross, U.; Kemnitz, E. J. Sol-Gel Sci. Technol. 2007, 41, 299–311. (4) Scheurell, K.; Scholz, G.; Kemnitz, E. J. Solid State Chem. 2007, 180, 749–758.

Ko¨nig et al. (5) Wuttke, S.; Scholz, G.; Ru¨diger, S.; Kemnitz, E. J. Mater. Chem. 2007, 17, 4980–4988. (6) Patil, P. T.; Dimitrov, A.; Kirmse, H.; Neumann, W.; Kemnitz, E. Appl. Catal., B 2008, 78, 80–91. (7) Ko¨nig, R.; Scholz, G.; Thong, N. H.; Kemnitz, E. Chem. Mater. 2007, 19, 2229–2237. (8) Ko¨nig, R.; Scholz, G.; Kemnitz, E. Solid State Nucl. Magn. Reson. 2007, 32, 78–88. (9) Pawlik, A.; Ko¨nig, R.; Scholz, G.; Kemnitz, E.; Ja¨ger, C. J. Magn. Reson. 2008, submitted. (10) Ru¨diger, S.; Gross, U.; Feist, M.; Prescott, H. A.; Shekar, C. S.; Troyanov, S. I.; Kemnitz, E. J. Mater. Chem. 2005, 15, 588–597. (11) Koch, J. Diploma Thesis, Humboldt-Universita¨t zu Berlin, 2008. (12) Ko¨nig, R.; Scholz, G.; Bertram, R.; Kemnitz, E. J. Fluorine Chem. 2008, 129, 598–606. (13) Ko¨nig, R.; Scholz, G.; Pawlik, A.; Ja¨ger, C.; van Rossum, B.; Oschkinat, H.; Kemnitz, E. J. Phys. Chem. C 2008, 112, 15708–15720. (14) Cory, D. G.; Ritchey, W. M. J. Magn. Reson. 1988, 80, 128–132. (15) Massiot, D.; Fayon, F.; Capron, M.; King, I.; Le Calve, S.; Alonso, B.; Durand, J. O.; Bujoli, B.; Gan, Z. H.; Hoatson, G. Magn. Reson. Chem. 2002, 40, 70–76. (16) Ko¨nig, R. Diploma Thesis, Humboldt-Universita¨t zu Berlin, 2006. (17) Abraham, A.; Prins, R.; van Bokhoven, J. A.; van Eck, E. R. H.; Kentgens, A. P. M. J. Phys. Chem. B 2006, 110, 6553–6560. (18) Amoureux, J. P.; Fernandez, C. Solid State Nucl. Magn. Reson. 1998, 10, 211–223. (19) Ko¨nig, R.; Scholz, G.; Pawlik, A.; Kemnitz, E.; Ja¨ger, C. Paper in preparation. (20) Medek, A.; Harwood, J. S.; Frydman, L. J. Am. Chem. Soc. 1995, 117, 12779–12787. (21) Pawlik, A.; Ja¨ger, C. Paper in preparation. (22) Liu, Y.; Tossell, J. J. Phys. Chem. B 2003, 107, 11280–11289. (23) Bureau, B.; Silly, G.; Buzare´, J. Y.; Emery, J. Chem. Phys. 1999, 249, 89–104. (24) Goepper, M.; Guth, J. L. Zeolites 1991, 11, 477–482. (25) Ru¨diger, S.; Kemnitz, E. Dalton Trans. 2008, 9, 1117–1127. (26) Quarterman, L.; Katz, J. J.; Hyman, H. H. J. Phys. Chem. 1961, 65, 90–93. (27) Chupas, P. J.; Corbin, D. R.; Rao, V. N. M.; Hanson, J. C.; Grey, C. P. J. Phys. Chem. B 2003, 107, 8327–8336. (28) Robert, E.; Lacassagne, V.; Bessada, C.; Massiot, D.; Gilbert, B.; Coutures, J. P. Inorg. Chem. 1999, 38, 214–217. (29) Lacassagne, V.; Bessada, C.; Florian, P.; Bouvet, S.; Ollivier, B.; Coutures, J. P.; Massiot, D. J. Phys. Chem. B 2002, 106, 1862–1868. (30) Herron, N.; Thorn, D. L.; Harlow, R. L.; Davidson, F. J. Am. Chem. Soc. 1993, 115, 3028–3029. (31) Kao, H. M.; Liao, Y. C. J. Phys. Chem. C 2007, 111, 4495–4498. (32) Delmotte, L.; Soulard, M.; Guth, F.; Seive, A.; Lopez, A.; Guth, J. L. Zeolites 1990, 10, 778–783. (33) Guth, J. L.; Delmotte, L.; Soulard, M.; Brunard, N.; Joly, J. F.; Espinat, D. Zeolites 1992, 12, 929–935. (34) Body, M.; Legein, C.; Buzare´, J.-Y.; Silly, G. Eur. J. Inorg. Chem. 2007, 14, 1980–1988. (35) Mu¨ller, D.; Bentrup, U. Z. Anorg. Allg. Chem. 1989, 575, 17–25.

JP810190K