Thermosensitive Drug Delivery System SBA-15-PEI for Controlled Release of Nonsteroidal Anti-Inflammatory Drug Diclofenac Sodium Salt: A Comparative Study

Mesoporous SBA-15 silica material was prepared by the sol–gel method and functionalized with thermosensitive polyethylenimine polymers with different molecular weight (g·mol−1): 800 (SBA-15(C)-800), 1300 (SBA-15(C)-1300) and 2000 (SBA-15(C)-2000). The nonsteroidal anti-inflammatory drug (NSAID) diclofenac sodium was selected as a model drug and encapsulated into the pores of prepared supports. Materials were characterized by the combination of infrared spectroscopy (IR), atomic force microscopy (AFM), transmission electron microscopy (TEM), photon cross-correlation spectroscopy (PCCS), nitrogen adsorption/desorption analysis, thermogravimetry (TG), differential scanning calorimetry (DSC) and small-angle X-ray diffraction (SA-XRD) experiments. The drug release from prepared matrixes was realized in two model media differing in pH, namely small intestine environment/simulated body fluid (pH = 7.4) and simulated gastric fluid (pH = 2), and at different temperatures, namely normal body temperature (T = 37 °C) and inflammatory temperature (T = 42 °C). The process of drug loading into the pores of prepared materials from the diclofenac sodium salt solutions with different concentrations and subsequent quantitative determination of released drugs was analyzed by UV-VIS spectroscopy. Analysis of prepared SBA-15 materials modified with polyethylenimines in solution showed a high ability to store large amounts of the drug, up to 230 wt.%. Experimental results showed their high drug release into the solution at pH = 7.4 for both temperatures, which is related to the high solubility of diclofenac sodium in a slightly alkaline environment. At pH = 2, a difference in drug release rate was observed between both temperatures. Indeed, at a higher temperature, the release rates and the amount of released drug were 2–3 times higher than those observed at a lower temperature. Different kinetic models were used to fit the obtained drug release data to determine the drug release rate and its release mechanism. Moreover, the drug release properties of prepared compounds were compared to a commercially available medicament under the same experimental conditions.


Introduction
Oncological and/or infectious diseases often cause inflammation in the human body [1]. Inflammation can have a local character or spread throughout the whole human body, accompanied by symptoms such as pain, fever and other complications. Anti-inflammatory support performed in solution revealed high DIC storage capacity up to 230 wt.%. The in vitro release properties were assessed in two model media differing in pH, namely small intestine environment/simulated body fluid (pH = 7.4) and simulated gastric juice (pH = 2), and at different temperatures, namely T = 37 • C simulating normal human body temperature and T = 42 • C representing inflammatory temperature. These two temperatures were chosen as external stimuli for the controlled release of diclofenac sodium. Indeed, at a lower temperature (T = 37 • C), the PEI molecules form hydrogen bonds with each other and the entry into the pores of the material should be blocked (closed conformation). At a higher temperature (T = 42 • C), the hydrogen bonding system should be temporarily disrupted, the pores of the carriers should be opened, accelerating the release of the drug (open conformation). In addition, the data obtained from drug release studies were fitted using different kinetic models to calculate the release rate and release mechanism of diclofenac sodium from porous matrixes.

Preparation of SBA-15
First, 13.1 g of surfactant P-123 was dissolved in a mixture of 382 mL of hydrochloric acid (2M solution) and 98 mL of deionized water. The reaction mixture was stirred for 48 h at 35 • C and 580 rpm. After the dissolution of Pluronic, 28 mL of TEOS was added dropwise 1 mL/min into the mixture under stirring. The whole reaction was stirred at 580 rpm and temperature 35 • C for 24 h. The reaction mixture was then placed in an oven which was preheated at 80°C, and the material was aged for 24 h. After the aging process, powder material was filtered off and the solid residue was washed several times with deionized water and dried in the stream of air (yield 17.3 g). The as-synthesized sample is denoted SBA-15(AS) in the following.

Extraction/Calcination Process
After the material drying procedure, surfactant molecules located in pores were removed by the combination of extraction and calcination processes. The first step was an extraction with dry toluene to purify the silica from contamination of impurities that occurred during the synthesis. The second step was sample extraction with HCl (2M) to remove the surfactant from the pores. The third step of the extraction was performed with tetrahydrofuran to remove the remaining surfactant molecules and other impurities. After extraction processes, the sample was dried in an oven at 70 • C and further calcined. The calcination process was performed in an oven with the following temperature program: First, the sample was slowly heated, with a heating rate of 0.5 • C·min −1 , to 150 • C for 2 h. Slow heating was chosen to remove residual solvents from the mesopores of the material. The next step was thermolysis of the residual surfactant by heating to 600 • C with a slow heating rate of 1.25 • C·min −1 . After heating the material at 600 • C for 8 h, the material was slowly cooled down to ambient temperature. The weight of the calcined material was 3.585 g, and the sample after the calcination process was denoted as SBA-15(C).

Surface Grafting and Modification
The prepared porous material SBA-15(C) was subjected to surface grafting (see Figure 2) with 3-(chloropropyl)-trimethoxysilane (CPTES). Three grams of SBA-15(C) was first dispersed in 150 mL of dry toluene, and then 9 mL of 3-(chloropropyl)-trimethoxysilane was added dropwise. After the addition of the silane derivative, the reaction mixture was heated for 24 h under reflux at 130 • C with a stirring speed of 500 rpm. Subsequently, the reaction mixture was cooled down to ambient temperature, filtered off, washed with dry toluene and dried in an oven at 50 • C overnight. To eliminate the unreacted species of 3-(chloropropyl)-trimethoxysilane from grafted material, 8 h extraction with dry toluene was performed. The mesoporous material after the grafting process with chloropropyl groups was designated as SBA-15(C)-Cl (yield 3.5 g). Chloropropyl-grafted material (SBA-15(C)-Cl) was prepared as a precursor for postsynthetic modification of the surface with polyethylenimines (PEIs) with different molecular weights: 800, 1300 and 2000 g·mol −1 . For PEI binding, the surface must contain free chloropropyl function groups that can react with primary or secondary amine groups of polyethylenimines by a condensation reaction to form a covalent C-N bond and release a HCl molecule (see Figure 2). In this way, 0.8 g of SBA-15(C)-Cl was dispersed in 25 mL of dry toluene, and then 3.3 g of corresponding PEI (800, 1300 and 2000 g mol −1 ) was added and synthesis was performed under reflux at 130 • C for 24 h. Unreacted PEI molecules were removed from the pores of the prepared materials by extraction with 2M hydrochloric acid solution and then dried in an oven at 70 • C overnight. Materials after surface modification with polyethylenimines were designated as SBA-15(C)-800 (yield 3.30 g), SBA-15(C)-1300 (yield 3.47 g) and SBA-15(C)-2000 (yield 3.64 g).

Drug Encapsulation
Modified samples SBA-15(C)-800, SBA-15(C)-1300 and SBA-15(C)-2000 and calcined material SBA-15(C) were used as carriers for diclofenac sodium salt. The affinity of the drug to the surface of carriers was assessed using methanolic solutions of diclofenac sodium with different concentrations: 0.1, 0.05, 0.025, 0.0125 and 0.00625 mol·dm −3 . Methanol was used as a solvent due to the high solubility of diclofenac sodium salt (35 mg·cm −3 ). In general, 10 mg of samples SBA-15(C)-PEI-X (X = 800, 1300 and 2000)/SBA-15(C) were dispersed in 1 mL of drug solution in a plastic vial. Suspensions were stirred at 40 rpm (rotator) for 24 h at 42 • C (open conformation) while the evaporation of methanol was prevented. The suspensions were then centrifugated at 6000 rpm for 10 min; after centrifugation, the supernatants were separated, and the amount of adsorbed drugs was determined by UV-VIS spectroscopy, based on the calibration curve of DIC in methanol (see Figure S1 in Supplementary Materials).

Drug Release
Drug-loaded samples (~3 mg) were packed in a semipermeable membrane VISKING and inserted into 50 mL polypropylene vials covered with caps. The vials were filled with 50 mL of saline solutions with pH = 2 and pH = 7.4 and transferred into a preheated oven at 37 • C (closed conformation) or 42 • C (open conformation), and mixtures were gently stirred on a magnetic stirrer. In general, the release amount of the drug was analyzed at selected time intervals: 0.5, 1.5, 3.5, 5.5, 7.5, 9.5 and 24 h.
The amount of the drug released was determined using UV-VIS spectroscopy. Before determining the concentration of DIC, calibration curves were constructed based on solutions (saline at pH = 7.4/2/5.5) with various concentrations of diclofenac sodium (see Figure S2 in Supplementary Materials). The correlation coefficient r 2 = 0.9999 for saline solution at pH = 7.4, r 2 = 0.9966 for saline solution at pH = 2 and r 2 = 0.9999 for saline solution at pH = 5.5 confirmed the linearity of calibration curves. pH = 5.5 was used in drug release experiments performed under dynamic conditions.

Drug Solubility
The diclofenac sodium solubility was studied at 37 and 42 • C and pH 2 and 7.4. One hundred milligrams of drug was dispersed in 10 mL of saline solution with different pH values in glass vials. Vials were sealed and sonicated for 3 min. After sonification, vials were put in the preheated ovens at 37 and 42 • C. The prepared suspensions were stirred under selected conditions (1. 37 • C, pH = 2; 2. 37 • C, pH = 7.4; 3. 42 • C, pH = 2; 4. 42 • C, pH = 7.4) for 24 h. After that, the remaining undissolved drug was filtered with polytetrafluoroethylene (PTFE) micro filter with 0.2 µm porosity and 20 mm diameter. Subsequently, 100 µL of supernatant from the individual solutions was diluted in water in a 1:1000 ratio (drug solution/saline solution) and analyzed by UV-VIS spectroscopy (Analytik Jena GmbH, Jena, Germany). The amount of drug in the samples was determined based on absorbance using calibration curves.

Characterization
Infrared spectra of prepared materials were measured by FT-IR spectroscopy on Nicolet 6700 from Thermo Scientific (Thermo Scientific, Waltham, MA, USA) using KBr technique in the wavelength range of 4000-400 cm −1 . Samples for IR analysis were prepared in the form of KBr pellets with KBr/sample mass ratio 100:1 or 100:5. For selected samples, a higher mass ratio was chosen for better visibility of characteristic absorption bands. Before IR measurements, potassium bromide was dried at 700 • C for 4 h in an oven to remove water and freely cooled in a desiccator. All IR spectra of prepared materials were recorded by collecting 64 scans with a resolution of 4 cm −1 for a single spectrum at ambient temperature.
Nitrogen adsorption/desorption isotherms were measured at −196 • C and experiments were carried out using Nova 1200e from Quantachrome Instruments (Quantachrome, Miami, FL, USA) and ASAP 2020 Micromeritics apparatus (Micromeritics, Norcross, GA, USA). Before nitrogen adsorption/desorption measurements, samples were outgassed at different temperatures (150 • C for SBA-15(C) and extracted samples, 60 • C for surfacemodified materials). The Brunauer-Emmett-Teller (BET) specific surface area (S BET ) of each sample was calculated using nitrogen adsorption data in a p/p 0 = 0.05-0.20. The textural properties such as pore size diameter (d) and pore volume (V p ) and were evaluated using the Barrett-Joyner-Halenda (BJH) model from the nitrogen desorption isotherm.
The shape and surface morphology of prepared materials were analyzed using atomic force microscopy (AFM) on a Solver PRO instrument from NT-MDT Spectrum Instruments (NT-MDT, Moscow, Russia). The measurements were performed in contact and semicontact mode on samples attached to a surface-treated mica plate with polylysine.
Transmission electron microscopy was carried out on a microscope JEOL 2000FX (JEOL, Pleasanton, FA, USA). Prepared materials were ground and dispersed in ethanol. The material's suspension was dripped onto a carbon grid and freely air-dried.
The particle size distribution in the nano-range was determined by a photon crosscorrelation spectroscopy, which was carried out on a Nanophox particle size analyzer (Sympatec, Clausthal-Zellerfeld, Germany). A small amount of powder product was dispersed in the deionized water and ultrasonicated for 15 min. After that, the obtained suspensions were left standing for 10 min, and subsequently, part of the top fraction containing a significant amount of fine particles was transferred into the measurement cuvette and diluted with distilled water to obtain a suitable concentration for the experiment. A dispersant refractive index of 1.33 was used for the analysis, and the measurements were repeated three times for each sample. The polydispersity index (PDI) was calculated by dividing the square of the standard deviation of the corresponding peak by the square of the value of its central position on the x-axis.
Small-angle X-ray diffraction (SA-XRD) experiments (Rigaku, Tokyo, Japan) were performed on a Rigaku Ultima IV multipurpose diffractometer in transmission geometry using CuKα radiation (λ = 1.54056 Å). The thickness of the sample during the measurement was 2 mm, which was achieved by placing the sample in a metal holder, which was sealed on both sides with Kapton tape. SA-XRD measurements were realized by 2θ continuous scan at 0.2 • min −1 in the 2θ range from 0.1 to 3 • , and scattered photons from the samples were detected every 0.02 • using a sodium iodide scintillation detector.
TGA Q500 apparatus from TA Instruments (TA Instruments, New Castle, DE, USA) was used for determining the thermal properties of prepared materials. The thermogravimetric analysis was performed in the temperature range of 30-800 • C with a heating rate of 10 • C min −1 in an air atmosphere with a flow rate of 60 cm 3 ·min −1 using platinum crucibles.
The volume phase transition behavior of PEI-modified samples was investigated by differential scanning calorimetry (DSC) on a DSC Q2000 apparatus from TA Instruments (TA Instruments, New Castle, DE, USA). The samples were heated with a heating rate of 1 • C min −1 in a temperature range of 10-60 • C and nitrogen atmosphere.
UV-VIS spectroscopy was performed on a Specord 250 UV-VIS spectrometer from Analytik Jena AG (Analytik Jena GmbH, Jena, Germany) in a wavelength range of 240-350 nm. UV-VIS spectroscopy was used to determine the amount of adsorbed/released drug.

Synthesis of SBA-15 and Surface Modification
The sol-gel process was used for the synthesis of SBA-15 by hydrolysis and polycondensation reactions of TEOS in water using Pluronic-123 as surfactant and HCl as a catalyst [41]. Sol is a low viscosity liquid that can be transformed into a solid form. Over time, the colloidal and condensed silica are combined into a 3D structure to form a mesh. The gel form and its physical properties depend mainly on the particle size. When the gel is forming, the viscosity increases the strength and integrity of the silica material. The structure and textural properties of the gel depend on the time of its formation. Other processes such as aging, drying, stabilization and thickening depend on the structure of the gel [42]. Achieving the largest silica surface area, pore size and pore volume depend on the removal of surfactant molecules. Calcination of silica nanoparticles at high temperature is a fast and efficient way to remove surfactant from pores. However, the calcination conditions must be chosen carefully because this process can perforate the silica pores. The as-synthesized silica contains not only surfactant molecules but also molecules of solvent in the pores. With rapid calcination, there is a risk that the solvent molecules will start to evaporate quickly and perforate the pores. Surfactant extraction is a slower but more gentle removal process that, on the other hand, does not lead to the complete removal of surfactant molecules. For the mentioned reasons, we have chosen in our work a combination of extraction and calcination of SBA-15(AS) material. The extraction was performed using solvents with different polarity (toluene, THF and 2M HCl solution), and then the material was calcined by slow heating at 600 • C. For modification with organic compounds, the surface was treated by silylation reaction with alkoxysilanes and similar silane compounds. Mentioned surface grafting is suitable for the subsequent attachment of organic compounds with different functional groups. The surface of SBA-15(C), which contains free hydroxyl groups, was subsequently grafted with 3-(chloropropyl)-trimethoxysilane molecules in dry toluene. This step was necessary for the formation of the covalent bond between the surface of SBA-15(C) and polyethylenimine molecules, via a condensation reaction. PEI molecules with different molecular weights (800, 1300 and 2000 g·mol −1 ) were used for surface modification. The materials prepared by the described processes were subsequently used as carriers for the anti-inflammatory drug diclofenac sodium and tested as thermosensitive DDSs.

SBA-15 and Surface-Modified Samples
Infrared spectra of prepared materials are presented in Figure 3, and the assignment of characteristic bands is summarized in Table 1. The infrared spectrum of SBA-15(C) (see Figure 3a) shows a broad band at 3437 cm −1 corresponding to stretching ν(OH) vibration of silanol groups located on the surface of silica (Si-OH) and physisorbed water molecules. Bending vibrations δ(OH) of mentioned groups was found at 1638 cm −1 . The presence of silica framework is reflected by broad and intense bands at 1088 and 809 cm −1 , which can be attributed to the asymmetric (ν as (SiOSi)) and symmetric (ν s (SiOSi)) stretching vibrations, respectively. Bending vibrations of δ(SiOSi) groups were detected at 462 cm −1 . Table 1. Assignment of vibrations and corresponding wavenumbers to characteristic bands in IR spectra of surface-modified materials.  The IR spectrum of SBA-15(C)-Cl after grafting with 3-(chloropropyl)-trimethoxysilane is shown in Figure 3b. Bands observed at the wavelengths of 2981, 2931 and 2900 cm −1 are attributed to the valence vibrations of ν(CH 2 ) groups of propyl chain. Deformation vibration (δ(CH 2 )) of mentioned groups appeared at 1453 cm −1 .
All IR spectra (see Figure 3c) of surface-modified samples with polyethylenimines (SBA-15(C)-PEI-X (X = 800, 1300 and 2000)) contain characteristic bands in the wavenumber range of 3360-3251 cm −1 due to stretching vibrations of primary/secondary amines (ν(NH 2 )/ν(NH)) of PEI molecules. In the IR spectra, deformation bands δ(NH 2 ) in the range from 1602 to 1603 cm −1 are present, which also confirm the presence of amine groups. Moreover, the presence of PEI molecules is also evidenced by the presence of two weak stretching vibrations of aliphatic ν(CH) groups of ethylene bridges located below 3000 cm −1 (see Table 1, Figure 3c).

Diclofenac Sodium and Drug-Loaded Samples
The structure of diclofenac sodium is presented in Figure 1, and the following bands were observed in its IR spectrum (see Figure S3 in Supplementary Materials): a weak band at 3080 cm −1 belonging to the stretching vibration of aromatic CH groups of phenyl scaffold; sharp, medium-intensity bands at 1496, 1573 and 1604 cm −1 corresponding to the valence vibrations of the conjugated system of C=C bonds in the phenyl ring; a low-intensity, sharp band at about 717 cm −1 related to the deformation vibrations for monosubstituted phenyl ring; and three low-intensity, sharp bands at 767, 951 and 1170 cm −1 corresponding to the deformation vibrations for 1,2,3-trisubstituted phenyl ring. Moreover, in the IR spectrum of diclofenac sodium salt, a strong and sharp band was found at 747 cm −1 , which could be assigned to the vibration of the C-Cl bond. The presence of a secondary amine group is obvious from the weak and broad band at 3386 cm −1 (ν(NH)), and its deformation vibration δ(NH) was detected at 1556 cm −1 . Other characteristic vibrations in the IR spectrum of diclofenac sodium are symmetric and asymmetric vibrations of the carboxylate group (ν(COO − )), located at 1400 and 1550 cm −1 , respectively.
The IR spectra of samples with encapsulated DIC are shown in Figure 3d, and characteristic bands are listed in Table 2. The presence of diclofenac sodium salt encapsulated in prepared materials is evidenced by the presence of two sharp, medium-intensity bands at 1555 and 1580 cm −1 , which correspond to the valence vibrations of the aromatic ring (ν(C=C) ar ). Typical vibrations of carboxylate group ν(COO − ) s and ν(COO − ) as were found at about 1450 and 1510 cm −1 . Another low-intensity, sharp band found at 747 cm −1 corresponds to the valence vibration of the C-Cl bond. Table 2. Assignment of vibrations and corresponding wavenumbers to characteristic bands in IR spectra of prepared materials with loaded drug.

Nitrogen Adsorption/Desorption Measurements
The specific surface area (S BET ), pore size (d) and pore volume (V p ) of prepared materials were determined by nitrogen adsorption/desorption measurement at −196 • C. Obtained adsorption/desorption isotherms are depicted in Figure 4, and calculated textural properties are listed in Table 3.
The extraction/calcination process of as-synthesized SBA-15 was monitored by nitrogen adsorption. The stepwise removal of surfactant from SBA-15(AS) by extraction with toluene (pink curve), hydrochloric acid (black curve), tetrahydrofuran (blue curve) and final calcination (red curve) is depicted in Figure 4a, and calculated textural parameters are summarized in Table 3. A detailed synthetic procedure for surfactant removal is given in Section 2.3. As can be seen from Figure 4a, the use of different polar solvents for the extraction procedure led to the gradual removal of surfactant, resulting, for samples, in an increase in both adsorbed nitrogen volume and associated surface area. The tolueneextracted sample showed a very low surface area of 14 m 2 g −1 . After extraction of the material with hydrochloric acid, there was a high increase in the S BET value to 547 m 2 g −1 . Subsequent extraction with tetrahydrofuran allowed the pore opening of SBA-15 to continue (744 m 2 ·g −1 ), and complete removal of the surfactant occurred after calcination of the material at 600 • C. The specific surface area of calcined material SBA-15(C) was 814 m 2 g −1 , which was further used for the PEI grafting process.
The opposite trend, i.e., the decrease in specific surface area, was observed for materials after the grafting process using 3-(chloropropyl)-trimethoxysilane and PEI molecules with different molecular weights. After grafting the surface with chloropropyl groups, a decrease in S BET was observed from 814 m 2 ·g −1 (SBA-15(C)) to 470 m 2 ·g −1 (SBA-15(C)-Cl). The textural properties of materials modified with PEI molecules showed that the polymers were predominantly grafted to the external surface of grains and at the entrances of the pores (and not in pores) as the specific surface areas were further reduced with PEI molecules, reaching 206 m 2 ·g −1 (SBA-15(C)-800), 184 m 2 ·g −1 (SBA-15(C)-1300) and 144 m 2 ·g −1 (SBA-15(C)-2000), and because the pore size decreased only minimally compared to the sample SBA-15(C)-Cl (see Table 3). Thus, PEI molecules predominantly occupy the particles' outer surface and partially prevent the effective diffusion of nitrogen into the internal pores. This fact was also confirmed by AFM spectroscopy, which clearly showed the PEI overlapping the entrances to the pores (see Section 3.4).

Surface Morphology and Particle Size
The atomic force microscopy (AFM) scan images and selected TEM micrographs of calcined SBA-15(C) material, propylchloride-grafted sample SBA-15(C)-Cl and polyethyleniminemodified material SBA-15(C)-800 are depicted in Figure 5. AFM was used to determine the surface morphology, shape and size of the pores in prepared materials. The AFM images (Figure 5a,b) and TEM micrographs (Figure 5c,d) display the uniform tubular character of pores with a size of 6.5 × 7 nm 2 , which are characteristic of mesoporous silica of the SBA-15 type. As can be seen from Figure 5a-d, the pores have a regular hexagonal structure, are placed next to each other and have an almost uniform structure. Figure 5e shows an AFM scan of sample SBA-15(C)-Cl, which indicates a porous surface with a "rougher" structure compared to the surface of calcined SBA-15(C) material (Figure 5a). The surface roughness confirms the presence of grafted chloropropyl units on the surface. The AFM scan images of SBA-15(C)-800 are presented in Figure 5f. The sample has the same roughened surface as in the case of SBA-15(C)-Cl, with the difference that the relief of the surface is more articulated than that seen in the image in Figure 5e. Taken together, these results show that the surface is coated with polyethylenimine, which forms a thin film.
Small-angle X-ray diffraction measurement based on scattered X-rays is an important technique in the analysis of condensate matter. In the case of porous silica materials, SA-XRD measurement provides important information about the pore organization in terms of symmetry and the unit cell size.
Measured SA-XRD patterns for prepared materials studied in this work are displayed in Figure 6. The SA-XRD patterns contain typical diffraction peaks associated with the presence of a regular hexagonal arrangement structure in the materials. The SA-XRD pattern of the calcined sample SBA-15(C) (see black line in Figure 6) shows the characteristic three well-defined reflections located at 1.019, 1.728 and 2.094 • that could be indexed as (100), (110) and (200) reflections, respectively. In functionalized materials with PEI molecules and samples after DIC encapsulation, the decrease in intensity of (110) and (200) reflections can be seen, while the first intense peak corresponding to (100) reflection is visible in all SA-XRD patterns. In general, the distance between the specific planes and lattice constants can be calculated based on peak position. Based on 2θ angle values, hkl indexes and the quadratic form of the Bragg equation for the p6mm hexagonal symmetry (Equation (1)), the unit cell parameter a for all prepared materials was calculated. Calculated values revealed that unit cell parameter (a) is about 100.5 Å for all samples (see Table 4). Findings from SA-XRD measurements confirm a hexagonal pore structure with a p6mm point group symmetry of synthesized carriers and confirm the porous framework's stability after the surface grafting process and drug loading.
Particle size distribution of surface-modified and drug-loaded samples was determined by a photon cross-correlation spectroscopy (PCCS), and obtained results are presented in Figure 7. The first view of particle size distribution curves shown in Figure 7 shows that the vast majority of the particles are in the submicron range. The particle size is influenced by both the presence of diclofenac sodium and by the molecular weight of the PEI polymers used in the grafting process.   In diclofenac-free samples (Figure 7a), the shift of particle size to the smaller one can be observed with the increase in the number of monomeric units/molecular weight of the polymer. This might be caused by the repulsion of its hydrophobic groups, as larger functional groups are associated with greater repulsion forces, which lead to smaller particles at the end. Moreover, a change in modality can be observed. Initially, the bimodal particle size distribution with one maximum located at around 370 nm and the absolute maximum at around 550 nm can be observed. With increasing the PEI molecular weight, the contribution of smaller particles increases. Namely, for the SBA-15(C)-800 sample, the bimodal size distribution with the smaller fraction being slightly more abundant was evidenced. Further increasing the PEI molecular weight leads to almost complete eradication of the larger fraction of particles and reduction of particle size. Namely, for the SBA-15(C)-2000 sample, almost unimodal distribution with an absolute maximum at the particle size of 287 nm was evidenced.
For the samples containing the therapeutics, the trend seems to be the other way around; i.e., with the increase in PEI molecular weight, the particle size distribution changes from unimodal to polymodal. Namely, for the PEI-free sample, the unimodal distribution with the maximum around 240 nm was evidenced. For this sample, the polydispersity index was calculated to be 0.096, which means that the presented data are relevant [43]. Introducing PEI and increasing its molecular weight leads to a change into bimodal (SBA-15(C)-800), trimodal (SBA-15(C)-1300) and polymodal (SBA-15(C)-2000) particle size distribution. The observed effect might be connected with the interaction of PEI with the therapeutic or with the efficient elimination of diclofenac sodium from the structure of the porous material.
To show the above-discussed results in a simpler way, the relationship between the molecular weight of PEI polymer and average particle size x 50 (50% of the particle diameters are smaller than this value) is shown in Figure 8. The constant decrease in x 50 value from 535 nm for the initial SBA-15(C) to 295 nm detected for SBA-15(C)-2000 sample is in accordance with the discussion above. The addition of diclofenac sodium changed the particle size distribution. The introduction of therapeutics showed an opposite trend; namely, x 50 values equal to 237 and 500 nm for the PEI-free sample and the sample containing PEI with molecular weight 1300 g·mol −1 , respectively, were found. The decrease in x 50 value in the case of SBA-15(C)-2000/DIC sample is not relevant, as the polymodal distribution was evidenced in this case and it is also highly probable that a significant amount of micrometer-scale particles sedimenting before the PCCS measurement was present in this case.

Thermogravimetric Analysis and Calorimetry
The thermal stability of calcined SBA-15(C), samples after surface grafting process and samples after diclofenac sodium encapsulation was studied by thermogravimetric (TG) analysis. The obtained TG curves measured in 150-800 • C temperature range are presented in Figure 9, and calculated parameters are summarized in Tables 5 and 6. Thermoanalytical curves and sample weight loss values were normalized at 150 • C [44], and removed weight losses were redistributed between the weight loss corresponding to the release of organic part and the residual mass. From TG results, the amount of PEI molecules bonded on the surface of supports and amounts of loaded diclofenac sodium in prepared materials were calculated.   The thermogravimetric curve of the material SBA-15(C) is displayed in Figure 9a. The TG curve indicates successful removal of surfactant from SBA-15(AS) pores after extraction and calcination processes as no significant weight losses were observed up to 600 • C. Above 600 • C, 0.25 wt.% weight loss on the TG curve was observed, which was associated with the silanol groups' dehydroxylation and condensation on the sample surface. This value was taken into account as a reference in the calculation of desired parameters for functionalized samples. TG curve of the SBA-15(C)-Cl ( Figure 9a) displayed a total weight loss of 6.01 wt.% corresponding to 58.8 mg of propylchloride groups grafted per 1 g of material (see Table 5 Table 5).
The TG analysis was also used to determine the weight of encapsulated DIC molecules in the PEI-modified/unmodified samples. Before the discussion of obtained results, the thermal behavior of diclofenac sodium is described. From Figure S4, it is evident that diclofenac sodium salt is thermally stable after heating to 250 • C. In the temperature range of 250-680 • C, the organic part's thermolysis takes place in three different steps associated with a total weight loss of 83.7 wt.% (calculated mass loss 83.4 wt.%). The decomposition stops at 680 • C, and as the final decomposition product, Na 2 CO 3 was identified (residual mass: observed 16.3 wt.%, calculated 16.6 wt.%). TG curves of diclofenac sodium loaded samples are depicted in Figure 9c, and calculated amounts of DIC are summarized in Table 6. The weight of DIC molecules was calculated from the difference between weight losses of PEI surface-modified materials and weight losses in samples with loaded drug. Moreover, the TG results obtained for pure diclofenac sodium were incorporated and used to calculate the weight losses corresponding to the DIC and residual masses in drugloaded samples. As can be seen from Table 6, the encapsulated quantity of DIC can be sorted in following order (see Table 6 . It should be noted that the calculated values do not provide realistic information on the effect of PEI surface modification on the affinity of DIC to the surface of carriers. To obtain this information, it is also necessary to take into account the textural properties, especially the surface area of the materials. Materials can store different amounts of the drug in their porous structure because of different surface areas/pore volumes (see Table 3 Table 6). From the obtained values, it can be concluded that the affinity of diclofenac sodium is similar for all PEI-modified materials, due to the formation of electrostatic interactions or intermolecular hydrogen bonds between primary/secondary groups of bonded molecules and diclofenac sodium molecules. PEI-modified samples are able to encapsulate four times more amount of drug compared to SBA-15(C).
The lower critical solution temperature (LCST) of an aqueous solution of PEI-modified samples was investigated by DSC measurements. Figure [45,46], it can be assumed that there is only a partial change in the conformation of the PEI molecules at the observed temperatures.

Drug Adsorption Properties
The drug adsorption properties of prepared materials were monitored in solution by UV-VIS spectroscopy based on a calibration curve of DIC in methanol (see Figure S1 in Supplementary Materials). From obtained UV-VIS spectra, diclofenac sodium contains a characteristic absorption band at 280 nm, which corresponds to an electron transition n → π*. Diclofenac sodium was loaded into the supports by the impregnation method using the methanolic drug solutions with different initial concentrations: 6.25 × 10 −3 , 1.25 × 10 −2 , 2.5 × 10 −2 , 5 × 10 −2 and 10 −1 mol.dm −3 . The determination of the supernatant concentration after adsorption, defined as the equilibrium concentration, allows the calculation of the amount adsorbed, Q ADS , by the following equation: where V is the volume of the drug solution (dm 3 ), C i (mol·dm −3 ) is the initial concentration, C eq (mol·dm −3 ) is the equilibrium concentration and m s is the mass (g) of adsorbent used. Then, the adsorption isotherms of DIC onto SBA-15(C)-800, SBA-15(C)-1300 and SBA-15(C)-2000 porous solids at 42 • C were determined, and obtained results are presented in Figure 10. A similar shape of isotherm was found for the three solids, corresponding to an L-type one [47]. L-type adsorption isotherm is frequently found in the case of energetic adsorption sites with a narrow and homogeneous distribution. Very similar values of DIC adsorbed amounts are found as isotherms are nearly superimposed. Furthermore, at low concentrations the slope of adsorption isotherms is very similar for the three solids and weak, evidencing a weak affinity of DIC for each solid. This is a very good result considering that DIC should be released afterward. Thus, it seems that there is no impact of the polymer chain length on the DIC adsorption properties. As shown in Figure 11, it is apparent that the prepared carriers can store high amounts of the drug: 221 wt.% in SBA-15(C)-2000, 227 wt.% in SBA-15(C)-1300 and 227 wt.% in SBA-15(C)-800 (wt.% after drug adsorption using 10 −1 mol·dm −3 DIC solution).

Drug Release Properties
In vitro drug release measurements were realized in two model media differing in pH, at pH = 7.4 for the simulated intravenous solution/small intestine environment and at pH = 2 for the simulated gastric juice. In addition, drug release was also assessed at two different temperatures: (i) At T = 37 • C, representing a normal human body temperature, the conformation of PEI molecules should partially or totally close the pores. (ii) At T = 42 • C, representing a fever temperature, the conformation of PEI molecules should partially or totally open the pores. The amount of DIC released from unmodified/modified samples was investigated using UV spectroscopy at different time intervals (0.5, 1.5, 3.5, 5.5, 7.5, 9.5 and 24 h). The actual diclofenac sodium concentrations during in vitro drug release experiments were calculated using the absorbance at two wavelengths depending on the pH value at λ = 287 nm at pH = 7.4 and λ = 285 nm at pH = 2. The loaded amounts of DIC (in mg/g od solid) determined by TG measurements (see Section 3.5) were used to calculate the weight percentage (wt.%) of the amount of drug released in selected time intervals: 191.1 mg for SBA-15(C)/DIC, 205.6 mg for SBA-15(C)-800/DIC, 178.7 mg for SBA-15(C)-1300/DIC and 141.8 mg for SBA-15(C)-2000/DIC (see Section 3.6). Figure 12 presents the time-dependencies of diclofenac sodium released quantity from the prepared materials under different temperature and pH conditions. Table 7 summarizes the amounts of released diclofenac sodium after 24 h. have a controlled release of the drug. A qualitative analysis was performed on a sample of the commercial product, which confirmed the presence of the undefined silicate component. The controlled release indicated that the material was unlikely to be porous, and thus, the drug was only granulated. The drug was released in less than 5 h, and the equilibrium drug concentrations equalized almost immediately. No effect of temperature was evidenced, as also found for the SBA-15(C) support.  First, release experiments were performed on unmodified silica material SBA-15(C)/DIC (see Figure 12a), and obtained results were used as "standard" for comparison with modified materials. As can be seen from Figure 12a, almost no release is observed at acidic pH = 2 as DIC release in wt.% is less than 10%. Under acidic conditions (pH = 2), diclofenac sodium salt is protonated and transformed to diclofenac acid, which has extremely low solubility in water, and this was reflected in the amount of drug released. At pH = 2, only a small influence of temperature was observed: approximately 1.5 times more drug was released at T = 42 • C (8.1 wt.%) compared to T = 37 • C (5.3 wt.%). A similar effect of the temperature on the DIC release properties was observed at pH = 7.4. The temperature had only a small effect on the release profiles: the DIC release from the surface of SBA-15(C)/DIC was a little more continuous at T = 37 • C compared to T = 42 • C. The maximal amounts of the drug released after 24 h at pH = 7.4 were similar: 89.4 wt.% at 37 • C and 91.6 wt.% at 42 • C. As shown in Figure 12a, the pH of the medium plays an important role in the drug release process, mainly at low pH.
The DIC release properties for PEI-modified samples were investigated at both pH levels and both temperatures. Results obtained at pH = 2 are very similar to those obtained for SBA-15(C)/DIC: only a small percentage of adsorbed DIC (<16%) was released at T = 37 • C and T = 42 • C. As can be seen from Figure 12b, at pH = 7.4 and both temperatures, the number of amine groups did not affect the total amount of drug released. The molecular weight/number of monomer units had no impact on the amount of released drug (a similar trend was found in TG analysis, as seen in Section 3.6), as DIC amounts released after 24 h were approximately the same: 51.1 wt.% at 37 • C and 73.5 wt.% at 42 • C for material SBA-15-800(C)/DIC, 48.1 wt.% at 37 • C and 75.1 wt.% at 42 • C for sample SBA-15-1300(C)/DIC and 49.2 wt.% at 37 • C and 72.7 wt.% at 42 • C for support SBA-15-2000(C)/DIC. The lower value of the maximum amounts of drug released found (close to 70 wt.%) compared to SBA-15(C)/DIC (close to 90 wt.%) can be explained by the creation of specific interactions (hydrogen bond formation and electrostatic interactions) between the drug molecules and primary/secondary amine functional groups. Furthermore, as can be seen from the above values, the temperature has an influence on the amount of drug released. Indeed, for all materials containing PEI molecules, the DIC release amount found at T = 42 • C is larger than the one at T = 37 • C, and the difference observed is between 22 and 27 wt.%, depending on materials. At T = 37 • C, the PEI molecules develop hydrogen bonding, and the entrances of the pores are partially blocked, whereas at T = 42 • C, the hydrogen bond system is temporarily disrupted and the pores are opened, favoring an important release of the drug. These results and DSC measurements (see Figure 9d) confirm the thermosensitivity of our carriers, even if the pore blocking is not total.
For a better interpretation of drug release results, the solubility of diclofenac sodium was studied under experimental conditions as described in the present work. The dispersed drug in the media was stirred for 24 h, and the dissolved quantity of drug was measured by UV-VIS spectroscopy. The calculated drug solubilities were 1399 µg·cm −3 at pH = 7.4 and T = 37 • C, 1508 µg·cm −3 at pH = 7.4 and T = 42 • C, 1.76 µg·cm −3 at pH = 2 and T = 37 • C and 1.91 µg·cm −3 at pH = 2 and T = 42 • C, and the determined results are in very good agreement with the published data [48,49]. As can be seen from obtained values, the temperature had a weak effect on the drug solubility. However, with increasing temperature, the amount of dissolved drug increased. ). This experimental result also points to the creation of intermolecular interactions between PEI and the drug molecules and the thermosensitivity of prepared DDSs. Moreover, lower amounts of DIC released from supports at pH = 2 when compared to pure diclofenac sodium could be an advantage of our prepared materials. It is known from clinical studies that one of the side effects of DIC with its long-term use or at high doses is the formation of gastric ulcers. It is estimated that NSAIDs, including DIC, cause around 2000 deaths a year due to the perforation of stomach ulcers [50,51]. SBA-15(C)-2000/DIC material suppresses the solubility of NSAIDs in gastric fluid (pH = 2) by 92% at T = 37 • C or 51% at T = 42 • C in comparison to the pure drug, which may suppress the risk of gastric ulcers.
Finally, the DIC release was also studied using commercial support, named Diclofenac Duo, from the company PharmaSwiss Figure 12. This product was reported to have a controlled release of the drug. A qualitative analysis was performed on a sample of the commercial product, which confirmed the presence of the undefined silicate component. The controlled release indicated that the material was unlikely to be porous, and thus, the drug was only granulated. The drug was released in less than 5 h, and the equilibrium drug concentrations equalized almost immediately. No effect of temperature was evidenced, as also found for the SBA-15(C) support.
To simulate the process of drug delivery through the digestive tract, an experiment with continuously changing pH and temperature was performed. After peroral administration, the carrier material passes through media with different pH values, starting from the stomach (pH = 2), moving through the large intestine (pH = 5.5) and, finally, entering the small intestine (pH = 7.4).
The material SBA-15(C)-1300/DIC was selected and used for the mentioned experiment, and the effect of applied dynamic conditions on the amount of drug released is shown in Figure 13. Figure 13 represents a fragmented graph with different applied conditions. In the first part, at pH = 2 and a normal body temperature of T = 37 • C, a minimum drug release of 1.8 wt.% after 30 min under applied conditions was observed. After this time, the system was heated to T = 42 • C, and a higher amount of drug (2.9 wt.%) was released (second fragment). Applied conditions examined the behavior of the carrier in the stomach at normal and inflammatory temperatures. Subsequently, the pH of the system was increased to 5.5 at a constant temperature (T = 42 • C), which simulated the conditions in the large intestine during fever (the third part of Figure 13). As shown in Figure 13, more than 50% of the drug was released after 2.5 h. The fourth part shows the simulated conditions of the small intestine environment at pH = 7.4 and T = 42 • C, when the maximum release of diclofenac sodium occurs with a value of 73.4 wt.%. In the case of lower temperature (T = 37 • C) at the same pH, DIC is readsorbed and the maximum amount released decreases to 47.5 wt.% (after 1 h). The above-observed results prove that after the peroral administration of our designed DDS, a gradual release of the drug occurs that depends on the pH and temperature in the gastrointestinal tract. Higher drug release occurs with increasing pH and also at higher temperatures, which imitate the inflammatory state of the body.

Drug Release Kinetics
Various kinetic models (zero order, first order, Korsmeyer-Peppas, Hixson-Crowell and Higuchi) were applied to fit the data obtained from drug release studies to determine the DIC release rate and its release mechanism. For this purpose, the data obtained at pH = 7.4 were analyzed, as the amount of drug release at pH = 2 was negligible due to the low solubility of the diclofenac sodium in an acidic environment. Calculated release rate constants with corresponding r 2 values based on the linear regression analysis of selected models are summarized in Table 8.
The release behavior of diclofenac sodium from sample SBA-15(C) at T = 37 • C and T = 42 • C and samples SBA-15(C)-800 and SBA-15(C)-1300 at T = 42 • C can be described by the Higuchi kinetic equation. The Higuchi model has commonly been used to model diffusion-controlled release processes of drug release from porous matrixes [52][53][54]. Higuchi's model describes the drug release from support as the square root of a timedependent process based on the Fickian diffusion process [55].
The Korsmeyer-Peppas model was applied for the samples with drug release of a slower rate, as only the first 60% of drug release data are supposed to be fitted in this model. From the calculated parameters, it was proven that the Korsmeyer-Peppas model is the best model for samples SBA-15(C)-800, SBA-15(C)-1300 and SBA-15(C)-2000 at T = 37 • C with regression coefficients (r 2 ) of 0.9364, 0.9544 and 0.9932, respectively (see Table 8). The obtained diffusion exponent value (n) was >1 for all samples, suggesting that diclofenac sodium release follows "super case-II" transport. The mentioned drug release model is typically used in the analysis and fitting of drug release data related to polymeric systems in which the release mechanism is not well known or when the release involves more than one type of drug release phenomenon [56,57]. This analysis of the release data also indicated the thermosensitivity of the polymers located on the surface of supports. At T = 37 • C, when polymers are in a closed conformation, a slower drug release is observed compared to T = 42 • C (open conformation).
The best fit for the commercial product Diclofenac Duo was obtained with the Hixson-Crowell model with regression coefficients 0.9669 and 0.9502. The Hixson-Crowell cube root equation is a mathematical model applied and frequently used to describe powder dissolution or drug release from specific formulations. The shape of the drug dosage form is spherical, and the size will decrease as the system dissolves. For drug dosage forms, this model best describes release from erodible matrixes such as sustained-release tablets/granules [58].

Conclusions
In the present work, we prepared, characterized and described novel thermosensitive materials based on mesoporous silica SBA-15 functionalized with polyethylenimines for controlled drug release of diclofenac sodium salt. The as-synthesized SBA-15 matrix was prepared by the sol-gel method, and the surfactant located in pores was carefully removed by a combination of extraction and calcination processes. Subsequently, the surface of the calcined material was modified with chloropropyl groups using a grafting procedure. The last synthetic step in the preparation of supports consisted of chloropropyl groups' condensation reaction and PEI molecules. In this study, PEI molecules present different molecular weights and, thus, different numbers of monomer units/numbers of amine groups. PEI polymers on the surface serve as thermosensitive molecules, which at lower temperatures form hydrogen bonds with each other and, due to their bulk, should block the entry into the pores. At higher temperatures, there is an initial disruption of intermolecular interactions, which should cause drug release. The mentioned conformational change was confirmed by DSC measurements. The successful synthesis of the prepared supports was confirmed by the results obtained using different analytical techniques such as infrared spectroscopy (IR), atomic force microscopy (AFM), small-angle X-ray diffraction (SXRD), nitrogen adsorption, transmission electron microscopy (TEM) and photon cross-correlation spectroscopy (PCCS). The nonsteroidal anti-inflammatory drug diclofenac sodium was encapsulated in the prepared materials. The amount of incorporated drug in the solid matrixes, PEI-modified/unmodified samples, was determined by thermogravimetric analysis. The drug adsorption determined by UV-VIS spectroscopy using a solution of diclofenac sodium at the maximum concentration shows that the prepared materials are able to store up to 23 times more drug than the weight of the carrier used. Subsequently, experiments were performed to assess drug release from the prepared materials into media with different pH values (simulated gastric fluid (pH = 2) and small intestine environment/simulated body fluid (pH = 7.4)) and temperatures (T = 37 • C representing normal body temperature and T = 42 • C simulating inflammatory temperature). From the obtained release curves, it could be concluded that in the case of pH, a higher amount of drug was liberated at pH = 7.4 compared to pH = 2 due to the drug solubility. In the case of temperature, a higher amount of DIC released was observed at a higher temperature, and the observed results confirmed thermosensitivity of the PEI-modified materials. Moreover, the drug release properties of prepared compounds were compared to a commercial product under the same experimental conditions. Various kinetic models were applied to fit the drug release data to study the mechanism of diclofenac sodium release. It was demonstrated that drug release at pH = 7.4 could be described by applying Higuchi and Korsmeyer-Peppas models.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/ma14081880/s1: Figure S1: (a) UV spectra of diclofenac sodium methanolic solutions and (b) corresponding calibration curve. Figure S2: Calibration curve of diclofenac sodium in saline solutions at (a) pH = 2 and (b) pH = 7.4. Figure S3: IR spectrum of diclofenac sodium salt. Figure S4: Thermogravimetric curve of diclofenac sodium salt. Funding: This work was supported by the Slovak Research and Development Agency project "Intelligent nanoporous systems as carriers of drug" under contract No. APVV-15-0520 andby the Development Operational Programme Integrated Infrastructure for the project "Nanoparticles for solving diagnostic-therapeutic problems with COVID-19 (NANOVIR)", ITMS 2014+:313011AUW7, co-founded by the European Regional Development Fund (ERDF). This research was also supported by Slovak Grant Agency VEGA, grant No. 2/0044/18 "High-energy milling for the synthesis of nanomaterials by a bio-approach and selected environmental applications".

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.