Sol-gel synthesis employing fluoride ion as the catalyst affords the advantage of increasing the reaction speed compared to other routes [18]. The reactions follow the SN2 type mechanism, where the F − ion acts as a nucleophile, attacking the silicon alkoxide, with subsequent release of water or alcohol, when in an acidic medium. The generated species becomes quite susceptible to nucleophilic attack by a water molecule due to the strongly withdrawing inductive effect of fluoride, leading to a five-coordinated transition state. The hydrolysis reaction is completed after catalyst regeneration. The condensation reaction goes through a hexacoordinated transition state.
3.1 Structural characterization of hybrid silica-based materials
The produced silicas were first evaluated by infrared spectroscopy, as shown in Fig. 1, and the assignments of the main bands identified in the spectra are summarized in Table 2.
Table 2
Band assignments of the FTIR spectra in Fig. 1 [20–24].
Wavenumber (cm− 1) | Assignment | Wavenumber (cm− 1) | Assignment |
3400 | ν O-H (H2O, SiOH) | 1270 | δ H-C-H (Si-CH3) |
2960 | νas C-H (CH3) | 1300 − 1000 | νas Si-O-(Si) |
2918 | νas C-H (CH2) | 950 | ν Si-OH e ν Si-O− |
2850 | νs C-H (CH2) | 837 | ρ CH3, ν Si-C |
1640 | δ H-O-H | 800 | νs Si-O-Si |
1550–1640 | δ H-N-H | 560 | ν Si-O, δ O-Si-O |
1465 | δ H-C-H | | |
The spectral region between 1300 cm− 1 and 850 cm− 1 represents the bands of greatest interest for structural evaluation of silica-based materials. In the case of asymmetric stretching of siloxane groups, it is possible to decompose this band into longitudinal optical (LO4 and LO6) and transverse optical (TO4 and TO6) components, which are divided into doublets due to the presence of four- and six-member rings in the silica network, respectively.
To quantify the previous analyses, it was necessary to perform a deconvolution of the bands following the Gaussian adjustment model after normalizing the spectra. The profiles found are presented in Fig. 2, where the stretching (Si-O−) and Si-OH were integrated as a single band, which was identified as ν Si-Od, as its area can indicate the degree of hydrophilicity of the sample according to Eq. 2 [20].
The values of the integrals of the longitudinal and transverse vibrational modes of the bands related to the asymmetric stretching of siloxane make it possible to determine the percentages of six- and four-member cyclic arrangements according to Eq. 3. The results are listed in Table 3.
Table 3 Deconvolution of the 850–1300 cm-1 spectral region of the FTIR spectra of all the samples: wavenumber (cm-1), integrated area of components (A), and variance (r²).
| C0 | | C1 | | C8 | | C18 | | APTES |
| Control | Sensor | | Control | Sensor | | Control | Sensor | | Control | Sensor | | Control | Sensor |
LO6 | 1224 | 1222 | | 1216 | 1213 | | 1221 | 1220 | | 1217 | 1218 | | 1216 | 1220 |
A | 7.4 | 3.8 | | 5.8 | 9.4 | | 5.4 | 3.8 | | 8.0 | 5.4 | | 6.3 | 4.0 |
LO4 | 1162 | 1149 | | 1142 | 1139 | | 1161 | 1153 | | 1147 | 1148 | | 1155 | 1161 |
A | 30.5 | 48.2 | | 47.1 | 45.2 | | 34.2 | 42.6 | | 39.3 | 42.5 | | 28.4 | 28.0 |
TO4 | 1078 | 1073 | | 1069 | 1071 | | 1077 | 1075 | | 1074 | 1074 | | 1073 | 1073 |
A | 69.5 | 52.4 | | 50.7 | 49.5 | | 69.7 | 59.5 | | 55.1 | 52.4 | | 57.7 | 62.1 |
TO6 | 1042 | 1040 | | 1036 | 1038 | | 1039 | 1038 | | 1046 | 1042 | | 1038 | 1039 |
A | 11.9 | 14.2 | | 17.3 | 16.9 | | 15.2 | 16.7 | | 12.2 | 16.6 | | 25.3 | 26.0 |
ν Si-Od | 957 | 956 | | 951 | 952 | | 951 | 952 | | 952 | 954 | | 954 | 952 |
A | 11.5 | 13.4 | | 12.8 | 13.1 | | 12.2 | 12.7 | | 7.9 | 10.3 | | 19.2 | 18.6 |
r² | 0.9997 | 0.9992 | | 0.9999 | 0.9998 | | 0.9997 | 0.9998 | | 0.9998 | 0.9992 | | 0.9996 | 0.9996 |
%(SiO)6 | 16.2 | 15.2 | | 19.1 | 21.7 | | 16.6 | 16.7 | | 17.6 | 18.9 | | 26.8 | 25.0 |
%Si-Od | 8.8 | 10.2 | | 9.6 | 9.7 | | 8.9 | 9.4 | | 6.5 | 8.1 | | 14.0 | 13.4 |
Four-membered arrangements are found in greater proportions in all the samples, and their formation is reported to be thermodynamically favorable [21]. With the addition of organoalkoxysilanes to the reaction, the percentage of six-membered siloxane units is expected to increase as a consequence of better accommodation of longer chains within the network [25]. For the evaluated samples, the proportion of these arrangements was significantly affected only by the APTES hybrid silica. The same was observed for the groups responsible for the increase in hydrophilicity in the silica network (%Si-Od) due to the interaction of silanol with the amino groups of the organosilane during synthesis.
When curcumin was incorporated into the matrices, a small disturbance in the formed silica network was observed, demonstrating that the condensation reactions were not significantly affected by the presence of these compounds at the studied concentrations.
The structural characteristics obtained from the FTIR spectra were correlated with the NMR 29Si spectra to better evaluate the role of the sol-gel route. The spectra corresponding to the silica matrices, as well as the sensor encapsulated in nonhybrid silica, are shown in Fig. 3, from which it was possible to identify the trifunctional (Ti) and tetrafunctional (Qi) silicon species indicated in Table 4, which generate signals with different chemical shifts in NMR spectra according to the degree of condensation in the molecules [26].
The presence of trifunctional species in the hybrid silicas confirmed the maintenance of the Si-C covalent bond after the hydrolysis of alkoxysilanes.
Table 4
NMR 29Si peak localization of silica-based matrices and sensors produced from curcumin entrapped in bare (nonmodified) silica.
Sample | Chemical shift (ppm) |
T1 | T2 | T3 | Q2 | Q3 | Q4 |
C0 | - | - | - | - | -102.6 | -112.9 |
C1 | - | -54.5 | -63.3 | - | -102.2 | -112.0 |
C8 | - | -56.9 | -64.9 | -92.1 | -101.5 | -110.9 |
C18 | -56.0 | -65.8 | -70.3 | -94.0 | -103.4 | -113.8 |
APTES | - | -68.2 | -75.2 | - | -109.5 | -118.5 |
C0_C | - | - | - | -92.6 | -101.4 | -111.3 |
3.2 Textural analysis
Another important technique used in the characterization of hybrid silicas was nitrogen porosimetry. The N2 adsorption-desorption isotherms presented in Fig. 4 (a) reveal similarities between the samples. According to the IUPAC [27] classification, C0, C1, C8 and APTES silicas correspond to type IV(a), which defines them as mesoporous materials, while C18 is presented as a combination of type II and IV(a) isotherms; therefore, these silicas are essentially nonporous or macroporous solids containing mesopores on their surface. The hysteresis presented by the materials appears as a result of capillary condensation, and their shapes provide information about the pore structure. In the case of the sample produced only from TEOS, type H1 hysteresis indicates the existence of uniform mesopores, with a narrow range of size distributions. Hybrid silicas, on the other hand, present nonparallel lines for the adsorption and desorption phenomena as a result of the heterogeneity in the pores. Solids containing methyl, octyl and 3-aminopropylsilane groups exhibit H2(a) hysteresis due to pore blockage, while silica containing octadecylsilane exhibits H3 hysteresis due to macropores that are not completely filled during condensation.
When analyzing the isotherms of the encapsulated materials illustrated in Fig. 4 (b), it is observed that they follow the same IUPAC classification as the corresponding silica, therefore presenting the same nature and characteristics as the pore network of their matrices. This shows that the reactions were governed mainly by the alkoxysilanes present, corroborating the conclusions drawn from the analysis of the FTIR spectra, from which little influence was found from the encapsulated compounds on the network structure of the obtained silica.
The shapes of the sensors' hysteresis are also similar to those of their respective matrices, with the exception of samples encapsulated in C0 and C1 silicas, where the hysteresis becomes a combination of H2(a) and H5, the latter being associated with a structure containing both open or partially blocked pores [27], which is reported less frequently in the literature but can be explained by the presence of hybrid components within the network.
Based on the isotherms, it is also possible to observe the differences in the N2 adsorption capacities of these materials. The surface areas, calculated by the BET method, and the average pore width, determined by the BJH method, are presented in Table 5.
The introduction of organic chains to the silica network generally results in an increase in surface area, which can be related to the increase in the percentage of cyclic arrangements of six siloxane units, which results in a more porous structure [28]. However, the surface area of C18 silica is an exception due to the obstruction of its pores by the longest chain in the series.
Although curcumin did not significantly affect the structure of the siloxane network, its incorporation led to changes in surface area or in the porosity of the obtained solids, thus demonstrating the influence of the compounds on the adsorption capacity of the resulting solids.
Table 5
Surface area (SBET) and porosity of the materials obtained from nitrogen physisorption isotherms.
Sample | SBET [m² g− 1] | | Average pore width (Å) |
Control | Sensor | | Control | Sensor |
C0 | 298.7 | 350.6 | | 74.1 | 61.0 |
C1 | 530.8 | 551.2 | | 45.3 | 49.7 |
C8 | 452.6 | 472.2 | | 44.1 | 44.7 |
C18 | 187.8 | 211.1 | | 126.3 | 118.7 |
APTES | 376.9 | 298.9 | | 53.3 | 55.1 |
3.3 Particle size evaluation
The dynamic light scattering (DLS) technique was employed to evaluate the size and distribution of the silica particles. The results are presented in Table 6, where it is observed that only the organoalkoxysilane with polar groups affected the average particle size, indicating a disadvantage in nucleation caused by the presence of APTES groups. High polydispersity indices denote a broadening in the particle size distribution, ranging from a moderate (0.1 ≤ PDI ≤ 0.4) to a broad (PDI > 0.4) polydispersity distribution [29].
The same evaluation procedure was used for the sensors containing curcumin, although a direct relationship could not be established with the values obtained from the matrices due to the light absorption and fluorescence presented by the encapsulated compound, which led to possible deviations in the measurements. However, similar to silicas, Table 6 shows that the largest particles among all encapsulates were obtained for sensors containing organic polar groups.
The polydispersity indices were also quite high, indicating heterogeneity in the particle size distribution, with sedimentation of particles with larger diameters occurring in some cases.
Table 6
Sample | Average particle size (nm) | | Polydispersity index, PDI |
Control | Sensor | | Control | Sensor |
C0 | 705 | 823 | | 0.493 | 0.303 |
C1 | 756 | 716 | | 0.465 | 0.347 |
C8 | 747 | 799 | | 0.327 | 0.568 |
C18 | 720 | 742 | | 0.241 | 0.328 |
APTES | 2265 | 1550 | | 0.546 | 0.367 |
3.4 Curcumin distribution characterization
Confocal laser scanning microscopy was used to evaluate the distribution of curcumin in the silica matrices. This technique is based on the fluorescence emitted by molecules when excited using a laser of suitable wavelength and provides a z-axis scan. Figure 5 shows the resulting projections of several focal planes for all the sensors, through which a relatively homogeneous distribution of curcumin can be verified across the entire length of the grains. However, the fluorescence intensities could not be compared to verify the contents of this compound in each matrix since the laser power varied between the samples. Different intensities can be observed only within the grains of the same sample, indicating that some agglomerations of curcumin occurred mainly in the sample encapsulated in hybrid silica containing APTES groups.
3.5 Optical properties of the sensors
The immobilization of compounds in a solid matrix can cause a shift in the maximum absorption bands in the ultraviolet‒visible region as a result of interactions between the organic molecules and the solid support, such as the formation of hydrogen bonds, steric effects, surface acidity, and polarity of the medium [30]. Therefore, diffuse reflectance UV‒Vis spectroscopy was used to evaluate the behavior of curcumin encapsulated in silica and hybrid silicas. Figure 6 shows the results obtained after applying the Kubelka–Munk function (K/S) shown in Eq. 4, where the reflectance values (Rλ) obtained from the measurements were converted into values approximately proportional to the absorbance of the materials [31].
K and S are parameters related to absorption and light scattering, respectively.
The curcumin spectrum has a maximum absorption band in the visible region centered at 476 nm, attributed to the π-π* electronic transition [28]. After encapsulation in the silica matrix, hypsochromic shifts are observed, leading to an increase in the band gap, with the most pronounced effects being observed for matrices containing C18 and APTES groups. The results demonstrated the interaction of the compounds with the silica matrices in which they were immobilized.
3.6 Performance evaluation of the curcumin sensors
The performance of the encapsulated materials was evaluated in comparison to that of free curcumin. The color changes are shown in Fig. 7, as well as the reversibility of the sensors. Compared with those of free curcumin, the color variations in encapsulated curcumin samples are much clearer, although all the results demonstrate ΔE*ab values greater than 2, indicating that an inexperienced observer is able to perceive the change in color [19]. This difference is probably due to the greater diffusion of ammonia and water vapors through the pores of the samples, the hydrophilicity of the silica network and the high surface area allowing greater interaction with the indicator, although the sample encapsulated in a matrix bearing an octadecylsilane group exhibited the greatest color variation. However, this conclusion is supported by the fact that curcumin agglomerates inside the sensor grains containing 3-aminopropylsilane groups, as demonstrated by confocal laser scanning microscopy, resulting in a reduction in ΔE*ab compared to that of the other silica matrices.
Regarding the reversibility of the sensors, a large variation is noted in the first 24 hours of measurements, with most samples returning close to their original color, which in turn allows the sensors to be reused.