Nucleophilic substitution of chloro- by azido groups on the silica surface
Nucleophilic substitution of chloroalkyl-modified silica monoliths to azide-containing monoliths (SiO2–(CH2)1,3–Cl → SiO2–(CH2)1,3–N3) was conducted in a saturated solution of NaN3 in DMF on monolithic silica gels that had been treated with trimethylchlorosilane (). During the course of this reaction, the macroscopic morphology of the monoliths was retained, and no significant influence on the macroporous network was observed. The gels had been treated with trimethylchlorosilane to remove reactive silanol groups and facilitate drying of the monoliths.
Schematic description of the nucleophilic substitution reaction for chloromethyl-modified silica pore surfaces.
Gels modified with chloromethyl groups (CMTMS) or chloropropyl groups (CPES) were subjected to the azide solutions. The number of azide groups per nm2
was evaluated according to a previously published method and was found to be in the range of 0.7 nm−2
(3.0 mmol CMTMS), 1.2 nm−2
(4.5 mmol CMTMS), 1.3 nm−2
(6.0 mmol CMTMS) and 0.7 nm−2
(3.0 mmol CPES) at a reaction temperature of 60 °C [29
]. The presence of the newly inserted azide functionalities was also confirmed by IR–ATR spectroscopy. This type of reaction has previously been reviewed for a variety of different silica surfaces, however, the influence of the reaction on the structural properties of the material has been mostly neglected [30
This influence of the nucleophilic substitution on the porous structure of the meso/macroporous monoliths, with special emphasis on the long range hexagonal ordering of the mesopores, was evaluated by nitrogen sorption and small angle X-ray scattering (SAXS) analyses. shows the nitrogen adsorption/desorption isotherms at 77 K for modified silica gels before and after nucleophilic substitution (SiO2
–Cl → SiO2
). The isotherms for the chloroalkyl-containing precursor materials are of type IV with H2 hysteresis loops according to the classification of Sing et al. [31
], whereas the same samples after conversion of the chlorides into azides display hysteresis loops of H1 type indicating a narrow distribution of pores. In addition, the isotherms for SiO2
exhibit stretching along the volume axis, adsorption and desorption isotherms display a sharper capillary condensation step and the relative pressure of the pore filling is shifted to larger values compared to the corresponding SiO2
–Cl. These variations in the hysteresis loops indicate an increase in the pore diameter and still a narrow pore size distribution for SiO2
Figure 2 Nitrogen sorption isotherms (taken at 77 K, left) and SAXS patterns (right) of SiO2–(CH2)n–Cl and SiO2–(CH2)n–N3 gels after nucleophilic substitution in a saturated DMF/NaN3 solution at 60 °C. n = 1 (top): Prepared (more ...)
The H2 type hysteresis loops obtained for SiO2
–Cl suggest rather complex pore structures with interconnected pores of different size and shape, e.g., spherical mesopores interconnected by smaller windows or large pore channels with undulating walls are possible. The pore sizes calculated from the desorption branch of the isotherm, applying the Barrett–Joyner–Halenda (BJH) model, are in the range of 3.5–4.7 nm for all samples. However, for pore diameters smaller than 5 nm (in our case presumably given by the small interconnecting windows; see , D
) the relative pressure at which desorption occurs is strongly influenced by fluid cavitations and instability of the meniscus [32
]. In addition, the BJH model is based on the Kelvin equation, which describes the relationship between the relative vapour pressure in equilibrium and the radius of curvature of the meniscus [33
]. Since a stable fluid meniscus with a given radius of curvature cannot be guaranteed for the desorption process in all systems, and the risk of obtaining physically meaningless results exists, the adsorption branch was also used to calculate the pore size distribution. This is not the case for the azido-functionalized samples (SiO2
) with pore sizes larger than 5.4 nm for all samples. Here, the calculation using the desorption isotherm is favoured, since desorption processes are thermodynamically more stable compared to the corresponding adsorption processes.
Structural characteristics of SiO2–(CH2)1,3–Cl compared to corresponding SiO2–(CH2)1,3–N3, obtained from nitrogen sorption analysis at 77 K.
gives all pore sizes as calculated from the adsorption and desorption isotherms. As expected, the calculation from the adsorption isotherm led to larger pore diameters for all samples with differences in the desorption pore size in the range of 1.7 to 2.4 nm for the methyl-spacer samples (n = 1), and in the range of 1.0 to 1.3 nm for the propyl-spacer samples (n = 3). Regardless of which sorption branch was applied for the calculation, a significant enlargement in the mesopore diameter after nucleophilic substitution in the range of 2.7 to 3.7 nm for methyl-spacer samples and in the range of 1.5 to 1.9 nm for propyl-spacer samples was observed. For instance, the chloromethyl-modified sample (3.0 mmol CMTMS) showed a pore diameter D
BJH,Ads of 5.54 nm prior to nucleophilic substitution and after conversion into the azides an increase to D
BJH,Ads = 9.17 nm was detected. The larger amount of nitrogen adsorbed at relative pressures above p/p
0 = 0.3 indicates a dramatic increase of the specific pore volumes (V
max and V
meso) after nucleophilic substitution, but relatively constant specific surface areas were observed from the pressure range p/p
0 = 0.05–0.30). V
max and V
meso followed the same trend and showed only slight deviations in their values, thus only V
max is discussed in the course of this work. The difference in V
max between samples before and after nucleophilic substitution was in the range of 300 to 440 cm3 g−1 for methyl-spacer samples and in the range of 270 cm3 g−1 for propyl-spacer samples ().
The decreasing C
-value, indicative of the adsorbent–adsorbate interactions, for gels prepared from a silica-precursor solution containing 3.0–6.0 mmol CMTMS follows the trend expected for gels with increasing coverage of the silica surface with organic groups. For nitrogen sorption on non-modified silica materials, the C
values are typically in the range 80–150 [34
also shows the SAXS patterns for the modified silica gels before and after nucleophilic substitution of the chlorides into azides (SiO2
–Cl → SiO2
). For all samples, higher order reflections were found, indicating long range ordering of the pore system. SiO2
exhibited the characteristic Bragg reflection sequence for a 2-D hexagonal ordering of 1 : 31/2
: 2 : 71/2
… and the reflections were indexed to the (10)-, (11)- and (20)-crystallographic planes [35
]. A comparison with SAXS patterns of the corresponding SiO2
–Cl precursor material clearly indicates mesostructural changes during the process of nucleophilic substitution. Both, SiO2
–Cl as well as SiO2
showed typical diffraction patterns for a 2-D hexagonal ordering of the pores. However, the relative intensities of the reflections were different for the chloroalkyl-modified silica gels compared to the corresponding azido-modified gels. The intensity of the (11)-reflection was reduced (almost to zero) compared to the corresponding azidoalkyl-modified silica gels. Furthermore, an additional higher order reflection was found for the chloroalkyl-modified precursor material that can be indexed to the (21)-crystallographic plane.
The differences in the reflection intensities before and after nucleophilic substitution () are attributed to the different form factors arising from differences in the respective pore wall thicknesses and pore diameters. One approach to describe these intensities is a two-phase model (pore and silica), where the form factor can be analytically solved (for more details see Supporting Information File 1
]). This model has been previously used to determine the pore diameter and pore-wall thickness of surface functionalized silica gel monoliths [37
Another approach is based on the reconstruction of the electron densities from a Fourier series and the appropriate choice of the phases [38
]. This has been experimentally and theoretically used to model the electron density across the pore for modified and unmodified MCM-41 and SBA-15 materials [11
] (for detailed information on SAXS data evaluation, see Supporting Information File 1
The best solution for the three observed reflections in our case was −+−, which differs to the phase shift from −−++ for the first four coefficients for the SBA 15 material observed by Flodström et al. or −++− for the MCM 41 material [11
]. This could be due to the variations in the synthesis conditions of the different materials.
As an example, in , the electron density reconstructions are shown for the SiO2–CH2–Cl and SiO2–CH2–N3 gels, with SiO2–CH2–Cl exhibiting a smaller pore with a steeper slope of the electron densities, whereas the corresponding substituted gel (SiO2–CH2–N3) has a broader distribution, which indicates a larger pore with a higher surface roughness.
Figure 3 Electron density reconstructions for modified silica gels (SiO2–CH2–Cl and SiO2–CH2–N3) that have been prepared from a silica precursor solution containing 3.0 (left), 4.5 (middle) and 6.0 (right) mmol CMTMS. The electron (more ...)
One would expect the electron density to converge to a constant value within the silica phase. Unfortunately, due to the limited number of peaks available for the reconstruction, the resolution was limited [38
]. A large constant region would require the sum of a large number of Fourier coefficients, i.e., a large number of diffraction peaks, which are not available for our type of materials. Thus this is an inevitable inherent weakness of the model.
The change in the ratio of the silica wall thickness to the pore diameter, during the nucleophilic substitution process, was also evidenced by nitrogen sorption analysis. An increase in pore diameter was observed (), while simultaneously a reduction of the pore wall thickness was detected for SiO2–(CH2)1,3–N3 compared to SiO2–(CH2)1,3–Cl ().
Comparison of the structural characteristics of SiO2−(CH2)1,3−Cl and the corresponding SiO2−(CH2)1,3−N3.
In the SAXS experiments, this led to the striking appearance of the (11)-reflection and the disappearance of the (21)-reflection. However, whereas sorption analysis indicated a strong decrease of the pore wall thickness, this effect was much less pronounced for the SAXS measurements. One possible explanation could be an additional surface roughness of the pores, which is also in coincidence with the electron density reconstruction (). The model description in SAXS (see Supporting Information File 1
) as a two-phase material, i.e., cylindrical pores of identical radius embedded in a silica matrix, leads to the measurement of a mean radius and averages out any differences in the radii or effects from surface inhomogeneities or roughness along the length or cross section of the pore. The radius obtained from SAXS could then lie intermediate between the radius obtained in the BJH analysis, from the adsorption branch and that from the desorption branch, as shown in . In the desorption branch, the BJH analysis is restricted, by the presence of small pores or surface roughness, to give the smallest pore size ().
SAXS averages the surface inhomogeneities to a mean radius in the two-phase model and leads therefore to a slightly larger diameter than the nitrogen sorption analysis (desorption branch).
Not only did the reflection intensity change during nucleophilic substitution, but also the relative position of the scattering vector q(hk) shifted to smaller values (), indicating an increase of the repeating unit distance. For instance, the chloromethyl-modified sample (4.5 mmol CMTMS) showed a shift of the scattering vector q
(10) from 0.65 to 0.61 nm−1 after conversion of the chlorides into the azides, corresponding to an increase of the d
(10)-spacing from 9.66 nm to 10.26 nm. The d
(10)-spacing was used to calculated the lattice constant, which was also found to increase during nucleophilic substitution (). For example, the chloromethyl-modified sample showed an increase in the lattice constant from 11.15 nm to 11.85 nm during nucleophilic substitution.
Structural properties of SiO2–(CH2)1,3–Cl compared to corresponding SiO2–(CH2)1,3–N3 obtained from SAXS analysis.
The structural parameters obtained from nitrogen sorption and SAXS analyses suggest a process of mesostructural changes during conversion of SiO2–(CH2)1,3–Cl into the corresponding SiO2–(CH2)1,3–N3. shows schematically a hexagonally organized porous material with cell parameters a, wall thickness t, repeating unit distance d
10 and pore diameter D as obtained from nitrogen sorption.
Schematic representation of a hexagonally organized pore system with the characteristic sizes. A similar arrangement is found for the samples by SAXS and transmission electron microscopy (TEM) analyses.
Based on the nitrogen sorption and SAXS analysis, the following trends were observed during nucleophilic substitution: The mesopore diameter and maximal pore volume drastically increased, while the lattice constant showed only a small enlargement with a simultaneous decrease in pore wall thickness. The ratio of pore wall thickness to pore diameter decreased to such an extent, that the new electron density (phase shift of Fourier coefficients) involved a significant change in the reflection intensity. We assume that the reduction in pore wall thickness with a simultaneous increase in pore diameter can not simply be explained by dissolution processes of silica, because the lattice constants also increased during the nucleophilic substitution. Simple ageing of unmodified mesoscopically organized silica gels in azide-containing media allows us to demonstrate that the observed effects are not due to the inserted azide-functionalities, which are covalently attached to the silica surface, but rather to an exposure of a mesostructured silica matrix to azide ions, as presented vide infra.
Ageing of unmodified silica gels in azide-containing media
From the results obtained above for the chloroalkyl-modified silica gels, the cause of the structural changes cannot be identified clearly and without ambiguity. Therefore, the reaction conditions were changed step by step to isolate the influence of temperature, solvent, anion-cation pair and solvent-azide compositions to identify the critical parameter. Pure (not organically modified) silica gels were kept for 3 d at 60 °C (identical conditions as for the nucleophilic substitution described above) in solutions of NaN3
in different solvents ranging from N
-dimethylformamide (DMF), 1,1,3,3-tetramethylurea (TMU), 1,3-dimethyl-2-imidazolidinone (DMI) to H2
O. Reference samples were kept for 3 d at 60 °C in the respective solvents without azide and pure silica gels were aged at that temperature. In addition, sodium azide was changed to tetramethylammoniumazide ((H3
, TMAA). All gels were aged, treated with trimethylchlorosilane and dried. The structural characteristics of untreated, reference and silica gels that were exposed to the different reaction conditions were again determined by nitrogen sorption and SAXS analyses. Note that the untreated silica, reference silica and azide-treated silica gels originated from the same monolithic silica piece, which was divided into three parts. shows the nitrogen adsorption/desorption isotherms taken at 77 K for gels treated in DMF and DMI; detailed information on gels in H2
O and TMU is given in Supporting Information File 1
Nitrogen isotherms and SAXS patterns of untreated silica gels, reference silica gels (solvent/60 °C) and azide-treated silica gels (solvent/NaN3/60 °C) in different solvents: DMF (top) and DMI (bottom).
All isotherms are of type IV with H1 hysteresis loops according to the classification of Sing et al. [33
]. The reference samples that were heat treated in the various solvents showed higher pore volumes compared to untreated silica gels, but this effect was clearly intensified by the addition of NaN3
, as demonstrated by the stretching of the hysteresis loops along the volume axis. Furthermore, the addition of NaN3
led to a shift of the relative pressure of the pore filling, by capillary condensation, to higher values. Pore diameters were significantly increased by the treatment with azide-containing solvents (). Differences in pore diameters for the various samples calculated from the adsorption isotherm in the BJH model were 4.05 nm (DMF/NaN3
); 4.21 nm (TMU/NaN3
); 2.20 nm (DMI/NaN3
) and 2.25 nm (H2
). The analogous calculation from the desorption branch led to smaller, but still significant, values for the increase in the pore diameter, i.e., 2.26 nm (DMF/NaN3
); 1.77 nm (TMU/NaN3
); 1.13 nm (DMI/NaN3
) and 1.84 nm (H2
Structural characteristics of untreated silica, reference silica (solvent/60 °C) and azide-treated silica gels (solvent/NaN3/60 °C) from nitrogen sorption analysis at 77 K, solvents: DMF and DMI; azide: NaN3 and TMAA.
Interestingly, the specific surface area S
dramatically increased from 477 m2
to 876 m2
by treatment of silica gels with pure DMF, whereas by treatment with DMF/NaN3
value only slightly increased from 477 m2
to 592 m2
. For the series with DMF this behaviour was reproduced for several samples (). The sample series with TMU and H2
O showed the same behaviour, whereas for the series with DMI the sample with additional NaN3
exhibited the highest surface area (, and Supporting Information File 1
Higher order reflections were found in the SAXS patterns for every sample, with the characteristic sequence for a 2-D hexagonal ordering of 1 : 31/2
: 2 : 71/2
]. As noted before for the series with SiO2
–Cl and SiO2
, a variation in ratio of the radius of the high electron density region (that is the silica wall) to the inner pore volume was indicated by changes in relative reflection intensities. However, for the unmodified silica gels, neither untreated silica, reference silica nor azide-treated silica displayed the (21)-reflection or disappearance of the (11)-reflection as was seen for SiO2
–Cl or SiO2
This is also reflected in the electron density reconstruction (, and Supporting Information File 1
). We assume that this is due to differences in the electron density and pore wall thicknesses for unmodified silica compared to silica modified with organic functionalities covalently attached to the silica walls.
After treatment with solvent/NaN3
at 60 °C the relative positions of the scattering vectors q
shifted to smaller values (, and Supporting Information File 1
) indicating an increase of the repeating unit distances. This was accompanied by an increase in the lattice constants, with a = 13.49 nm for samples that have been treated in DMF with the addition of NaN3
, and a = 13.44 nm for the respective TMU and a = 13.39 nm for DMI samples. With H2
a slightly smaller lattice constant of 13.22 nm was observed (Supporting Information File 1
Structural properties as obtained from SAXS analysis of untreated silica, reference silica (solvent/60 °C) and azide-treated silica gels (solvent/NaN3 or TMAA/60 °C), solvents: DMF and DMI.
With DMF, TMU and DMI we deliberately chose aprotic solvents that would not solvate the azide ions. This is important when azides are made to react by nucleophilic substitution, since assuming a bimolecular mechanism (SN2), the rate constant will be increased by a unsolvated, and therefore not stabilized, nucleophilic agent. This is in agreement with the fact that when methanol was used as the solvent for the nucleophilic substitution, the yield was much lower. However, the results from the series of gels treated in H2O clearly demonstrate that the effect on the mesostructure is due to the azide ions, independent of the coordination environment of the azide.
Substitution of NaN3 by (H3C)4NN3 (tetramethylammoniumazide, TMAA) led to similar effects on the mesostructure as mentioned above. An unmodified silica gel was kept for 3 d at 60 °C in a solution of TMAA in DMF. A reference sample was kept for 3 d at 60 °C in DMF. The structural characteristics of untreated silica, reference silica and silica gels that were exposed to DMF/TMAA (all originating from the same gel monolith) were again determined by nitrogen sorption and SAXS analyses.
shows the nitrogen sorption isotherms and SAXS patterns. As observed previously for NaN3, the addition of TMAA leads to a shift of the relative pressure of the capillary condensation step to larger values, indicating an increase in mesopore diameter. Calculation from the desorption isotherm in the BJH model indicated an increase in the pore diameter from 6.23 to 7.33 nm, and calculation from the adsorption isotherm indicated an increase from 9.21 nm to 11.63 nm (). The specific surface area S
BET showed the same behaviour as for the series with DMF/NaN3 (). By treatment with pure DMF, a dramatic increase from 627 m2 g−1 to 906 m2 g−1 was observed, whereas by addition of the azide the S
BET value remained almost constant.
Nitrogen isotherms at 77 K and SAXS patterns of untreated silica, reference silica (DMF/60 °C) and azide-treated silica gels (DMF/TMAA/60 °C).
Higher order reflections with the same characteristic sequence for a 2-D hexagonal ordering of 1 : 31/2 : 2 : 71/2… were found. As observed for the series before, exposure to the azide compound led to a shift of positions for the scattering vectors q
(hk) to smaller values, indicating an increase of the repeating unit distance. In addition to that, an increase in the lattice constant was detected. A value of 13.08 nm was found for the sample that was treated with DMF and addition of TMAA in comparison to 12.75 nm for the untreated sample.
Exposure of silica gels to TMAA led to the same mesostructural effects as observed for NaN3. Therefore, substitution of a relatively small counter ion (Na+) by a sterically demanding counter ion ((H3C)4N+) did not change the observed effects of azides on mesoscopically organized silica gels.