Deposition of CAP / Antioxidants Systems on Silica Particles Using the Supercritical Antisolvent Process

: Supercritical carbon dioxide has been used to deposit co-precipitates of natural antioxidants with a polymer onto silica microparticles. The supercritical antisolvent process (SAS) was carried out with the antioxidants by introducing the silica microparticles into the precipitator vessel. Two di ﬀ erent conﬁgurations were employed to pump the solution. In one conﬁguration, the antioxidant and the polymer were dissolved and injected together through a nozzle. In the second conﬁguration, the antioxidant and the polymer were dissolved in di ﬀ erent solutions and sprayed through di ﬀ erent nozzles. The use of operating conditions signiﬁcantly above the critical point (180 bar and 323 K) led to the formation of composites made up of co-precipitates and silica. Delivery proﬁles showed that the presence of the polymer and the silica delayed release of the antioxidant into gastric media, thus protecting it and allowing its full delivery to the intestinal ﬂuids to improve the e ﬀ ectiveness of the antioxidant.


Introduction
The development of new scientific techniques to improve aspects of life in an environmentally friendly way is demanded by society in general and is a principle that is addressed by the scientific community. In recent years, the use of CO 2 under supercritical conditions (i.e., above 304.1 K and 73.8 bar) has been considered as a green and environmentally friendly evolution when compared to more conventional methods [1,2]. The effectiveness of this approach in processes involving extraction [3,4], micronization [5,6], encapsulation [7,8] or impregnation [9,10], while removing or reducing the use of organic solvents that were widely used decades ago in industry, represents a great advance in meeting the current demands of society.
In accordance with the above principles, the supercritical antisolvent process (SAS) exploits the ability to precipitate compounds that are not soluble in supercritical CO 2 . A solution of an active substance is sprayed through a nozzle into a precipitator that contains CO 2 under supercritical conditions. Once the substances are in contact, scCO 2 diffuses into the organic solvent, thus leading to supersaturation of the solution, which in turn results in precipitation of the active substance. A large number of authors have used this method [11,12] to micronize or encapsulate various substances.
In the work described here, the matrix selected to deposit the antioxidants as precipitates was mesoporous SB-300 silica, mainly due to its large specific surface area and high porosity [13]. Some authors have already studied supercritical impregnation onto porous matrixes such as activated In this process, quercetin or mangiferin and CAP were dissolved together in 50 mL of a 2:1 mixture of ethanol or acetone/DMSO, respectively, with a concentration of 20 mg/mL. The mixtures were stirred magnetically for 30 min. CO 2 was blown into a precipitator vessel to reach the pressure and temperature required. The liquid mixture was then injected into the vessel and sprayed through a nozzle with a diameter of 100 µm. The scCO 2 rapidly dissolved in the organic solution during the mixture and this led to the precipitation of the solutes due to the anti-solvent effect ( Figure 1). Appl. Sci. 2020, 10, x FOR PEER REVIEW 4 of 13

Supercritical antisolvent process using two nozzles (SAS2)
In this configuration (Figure 2), CAP and the antioxidant (Q and MNG) were dissolved as separate solutions in ethanol for CAP and Q, and in acetone/DMSO for CAP and MNG experiments. These solutions were pumped into the precipitator vessel through two different nozzles, each 100 µm in diameter. In this case, each 50 mL of solution had a concentration of 20 mg/mL (Q or MNG) and 40 mg/mL (CAP) in order to maintain a 1:2 ratio and the dissolution flow rate was the same for both solutions.

Scanning Electron Microscopy
The morphologies of the samples were analyzed with a Nova NanoSEM 450 field-emission scanning electron microscope (SEM) with 1 nm ultimate resolution. High-vacuum operations were carried out during sample analyses. It is possible to observe secondary and backscattered electrons under various conditions thanks to several detectors and to obtain STEM images. The images were

Supercritical Antisolvent Process Using Two Nozzles (SAS2)
In this configuration (Figure 2), CAP and the antioxidant (Q and MNG) were dissolved as separate solutions in ethanol for CAP and Q, and in acetone/DMSO for CAP and MNG experiments. These solutions were pumped into the precipitator vessel through two different nozzles, each 100 µm in diameter. In this case, each 50 mL of solution had a concentration of 20 mg/mL (Q or MNG) and 40 mg/mL (CAP) in order to maintain a 1:2 ratio and the dissolution flow rate was the same for both solutions.

Supercritical antisolvent process using two nozzles (SAS2)
In this configuration (Figure 2), CAP and the antioxidant (Q and MNG) were dissolved as separate solutions in ethanol for CAP and Q, and in acetone/DMSO for CAP and MNG experiments. These solutions were pumped into the precipitator vessel through two different nozzles, each 100 µm in diameter. In this case, each 50 mL of solution had a concentration of 20 mg/mL (Q or MNG) and 40 mg/mL (CAP) in order to maintain a 1:2 ratio and the dissolution flow rate was the same for both solutions.

Scanning Electron Microscopy
The morphologies of the samples were analyzed with a Nova NanoSEM 450 field-emission scanning electron microscope (SEM) with 1 nm ultimate resolution. High-vacuum operations were carried out during sample analyses. It is possible to observe secondary and backscattered electrons under various conditions thanks to several detectors and to obtain STEM images. The images were

Scanning Electron Microscopy
The morphologies of the samples were analyzed with a Nova NanoSEM 450 field-emission scanning electron microscope (SEM) with 1 nm ultimate resolution. High-vacuum operations were carried out during sample analyses. It is possible to observe secondary and backscattered electrons under various conditions thanks to several detectors and to obtain STEM images. The images were taken from powder samples placed on a carbon tape and coated with a 15 nm film of gold under an argon atmosphere prior to analysis.

Release Tests
Solutions of intestinal fluid (SIF) and simulated gastric fluid (SGF) were prepared as follows. In order to make SIF, a solution of 6.8 g/L of monobasic potassium phosphate in distilled water was prepared and adjusted to pH 6.8 ± 0.1 with 0.2 M NaOH. SGF was prepared by dissolving 2 g of sodium chloride in 1 L of distilled water and adjusting the pH to 1.2 ± 0.1 with 0.2 N HCl.
Samples of co-precipitates (5 mg) were added to 40 mL of simulated fluid in each release experiment. The experiments were carried out at 37 • C and 165 rpm. Aliquots (3 mL) of each sample were removed for measurement at certain intervals of time up to 8 h. Prior to analysis, the samples were filtered through a Teflon filter (0.45 µm). Each sample was replaced by an equal amount of fluid. A Shimadzu UV-VIS mini spectrophotometer with λ 365 nm was used to measure the quantity of quercetin and mangiferin present in these aliquots. Calculation of the corrected concentration of released quercetin and mangiferin was carried out as defined by Zhu et al. [38].

SAS1 and SAS2 Experiments
This study was focused on the variation of the CO 2 flow rate and how the two configurations (SAS1 and SAS2) affected the deposition (see Supplementary Materials). The results of experiments 1-4, in which the co-precipitates are CAP/Q, are provided in Table 1. In these experiments a pressure of 90 bar, a temperature of 313 K, and a constant dissolution rate of 6 mL/min were fixed for all experiments, the CO 2 flow rate varied from 10 to 30 g/min, and the appropriate SAS process configuration. The results measured at the bottom of the precipitation vessel in the experiments are provided in Figure 3. It can be observed that in experiments 1, 3, and 4 there was precipitation on the bottom of the vessel, on the wall, and on the nozzle. However, in experiment 2 there was only precipitation on the wall of the vessel. It is important to note that the silica particles were located on the bottom of the vessel. In these experiments, the bottom of the vessel appeared wet with residual organic solvent. This observation could be due to the fact that the experiments were carried out very close to the MCP, which for DMSO/acetone/CO 2 are 79.7 bar at 313 K and 89.6 bar at 323 K, [35,36] and for ethanol/CO 2 are 79.4 bar at 313.4 K and 89.8 bar at 322.5 K [37]. This situation would cause a subcritical process that cannot remove all of the organic co-solvent. In the experiments carried out with a higher flow rate of CO 2 (exp. 1 and 3), the bottom of the vessel was covered with a large quantity of particles-more than in exp. 2 and 4. This could be caused by an increase in the turbulence near to the jet generating a higher supersaturation [39] and providing a higher efficiency in the micronization. The SEM images for experiments 1-4 ( Figure 4) did not show the presence of silica particles, which are spherical particles of around 3-8 µm (as provided by the manufacturer, the Miyoshi laboratory) [13]. The spherical particles correspond to the precipitation of CAP and the rectangular sticks were identified as precipitated Q. Appl. Sci. 2020, 10, x FOR PEER REVIEW 6 of 13 The images of the precipitates on the bottom of the vessel in deposition experiments 5 and 6 are shown in Figure 3 and these correspond to CAP/MNG co-precipitates. It can be observed that almost all compounds were deposited on the bottom of the vessel to form a "body-mass". The SEM images of these "body-mass" solids ( Figure 5) for both configurations showed the presence of silica particles and CAP/MNG co-precipitates. In this case, a slight variation in the morphologies of the particles between SAS1 and SAS2, as observed in a previous study [34], was found. SAS2 co-precipitate particles (exp. 5) showed two different morphologies (quasi-spherical and sticks) against one morphology (quasi-spherical) for the particles observed for co-precipitates with SAS1. It is remarkable that the micronization of MNG also led to quasi-spherical particles and the raw MNG particles are sticks [26]. It can be argued that with the double injection the amorphization of the MNG did not occur and particles similar to the raw material, i.e., with some degree of crystallinity, were formed.
In these experiments one could observe the difference between the silica particles, which are spherical particles of 3-8 µm, and the spherical particles and sticks corresponding to co-precipitated CAP/MNG, which were smaller.   The images of the precipitates on the bottom of the vessel in deposition experiments 5 and 6 are shown in Figure 3 and these correspond to CAP/MNG co-precipitates. It can be observed that almost all compounds were deposited on the bottom of the vessel to form a "body-mass". The SEM images of these "body-mass" solids ( Figure 5) for both configurations showed the presence of silica particles and CAP/MNG co-precipitates. In this case, a slight variation in the morphologies of the particles between SAS1 and SAS2, as observed in a previous study [34], was found. SAS2 co-precipitate particles (exp. 5) showed two different morphologies (quasi-spherical and sticks) against one morphology (quasi-spherical) for the particles observed for co-precipitates with SAS1. It is remarkable that the micronization of MNG also led to quasi-spherical particles and the raw MNG particles are sticks [26]. It can be argued that with the double injection the amorphization of the MNG did not occur and particles similar to the raw material, i.e., with some degree of crystallinity, were formed.
In these experiments one could observe the difference between the silica particles, which are spherical particles of 3-8 µm, and the spherical particles and sticks corresponding to co-precipitated CAP/MNG, which were smaller.  The images of the precipitates on the bottom of the vessel in deposition experiments 5 and 6 are shown in Figure 3 and these correspond to CAP/MNG co-precipitates. It can be observed that almost all compounds were deposited on the bottom of the vessel to form a "body-mass". The SEM images of these "body-mass" solids ( Figure 5) for both configurations showed the presence of silica particles and CAP/MNG co-precipitates. In this case, a slight variation in the morphologies of the particles between SAS1 and SAS2, as observed in a previous study [34], was found. SAS2 co-precipitate particles (exp. 5) showed two different morphologies (quasi-spherical and sticks) against one morphology (quasi-spherical) for the particles observed for co-precipitates with SAS1. It is remarkable that the micronization of MNG also led to quasi-spherical particles and the raw MNG particles are sticks [26]. It can be argued that with the double injection the amorphization of the MNG did not occur and particles similar to the raw material, i.e., with some degree of crystallinity, were formed.
In these experiments one could observe the difference between the silica particles, which are spherical particles of 3-8 µm, and the spherical particles and sticks corresponding to co-precipitated CAP/MNG, which were smaller. Appl. Sci. 2020, 10, x FOR PEER REVIEW 7 of 13 Finally, experiments 7 and 8 were carried on the CAP/Q co-precipitates under the same conditions as experiments 5 and 6 due to the successful results obtained with these conditions. In this case, the increase in the pressure from 90-180 bar avoided the possible complications related to being in subcritical conditions and the CO2 flow rate was fixed at 30 g/min. The bottom of the vessel was free of co-solvent for both configurations of the SAS process (see Figure 3) and a similar dry "body-mass" as in experiments 5 and 6 was observed. The SEM images ( Figure 6) verify the presence of silica particles and CAP/Q co-precipitates. Furthermore, a variation between SAS1 and SAS2 processes was not observed, with stick-shaped particles obtained for both SAS1 and SAS2, usually identified as Q particles [30,31], NPs and MPs corresponding to CAP, and larger spheres corresponding to silica.
In these experiments, the amount of co-precipitate with Q was lower than in experiments 5-6 with MNG, and this change was due to the experimental conditions used, which favored a higher precipitation of MNG than CAP, as described previously [31,34]. In order to confirm the presence of quercetin and mangiferin in the silica co-precipitates, 5 mg of each sample were dissolved in 50 mL of acetone/DMSO for MNG co-precipitates and 50 mL of ethanol for Q co-precipitates ( Table 2). The contents in these solutions were measured by UV-VIS spectrophotometry at a wavelength of 365 nm using a previously obtained calibration curve.
It was observed that in experiments 7 and 8-corresponding to the use of quercetin-the concentration was significantly lower than in experiments 5 and 6, where the concentration of mangiferin was six times higher than quercetin for the SAS1 process and three times higher for the SAS2 process. The fact that the amount of MNG was higher than Q is not unexpected since the operating conditions imposed to achieve the "body-mass" are suitable for MNG according to previous studies carried out on CAP/MNG co-precipitation.  Finally, experiments 7 and 8 were carried on the CAP/Q co-precipitates under the same conditions as experiments 5 and 6 due to the successful results obtained with these conditions. In this case, the increase in the pressure from 90-180 bar avoided the possible complications related to being in subcritical conditions and the CO 2 flow rate was fixed at 30 g/min. The bottom of the vessel was free of co-solvent for both configurations of the SAS process (see Figure 3) and a similar dry "body-mass" as in experiments 5 and 6 was observed. The SEM images ( Figure 6) verify the presence of silica particles and CAP/Q co-precipitates. Furthermore, a variation between SAS1 and SAS2 processes was not observed, with stick-shaped particles obtained for both SAS1 and SAS2, usually identified as Q particles [30,31], NPs and MPs corresponding to CAP, and larger spheres corresponding to silica.  Finally, experiments 7 and 8 were carried on the CAP/Q co-precipitates under the same conditions as experiments 5 and 6 due to the successful results obtained with these conditions. In this case, the increase in the pressure from 90-180 bar avoided the possible complications related to being in subcritical conditions and the CO2 flow rate was fixed at 30 g/min. The bottom of the vessel was free of co-solvent for both configurations of the SAS process (see Figure 3) and a similar dry "body-mass" as in experiments 5 and 6 was observed. The SEM images ( Figure 6) verify the presence of silica particles and CAP/Q co-precipitates. Furthermore, a variation between SAS1 and SAS2 processes was not observed, with stick-shaped particles obtained for both SAS1 and SAS2, usually identified as Q particles [30,31], NPs and MPs corresponding to CAP, and larger spheres corresponding to silica.
In these experiments, the amount of co-precipitate with Q was lower than in experiments 5-6 with MNG, and this change was due to the experimental conditions used, which favored a higher precipitation of MNG than CAP, as described previously [31,34]. In order to confirm the presence of quercetin and mangiferin in the silica co-precipitates, 5 mg of each sample were dissolved in 50 mL of acetone/DMSO for MNG co-precipitates and 50 mL of ethanol for Q co-precipitates ( Table 2). The contents in these solutions were measured by UV-VIS spectrophotometry at a wavelength of 365 nm using a previously obtained calibration curve.
It was observed that in experiments 7 and 8-corresponding to the use of quercetin-the concentration was significantly lower than in experiments 5 and 6, where the concentration of mangiferin was six times higher than quercetin for the SAS1 process and three times higher for the SAS2 process. The fact that the amount of MNG was higher than Q is not unexpected since the operating conditions imposed to achieve the "body-mass" are suitable for MNG according to previous studies carried out on CAP/MNG co-precipitation.  In these experiments, the amount of co-precipitate with Q was lower than in experiments 5-6 with MNG, and this change was due to the experimental conditions used, which favored a higher precipitation of MNG than CAP, as described previously [31,34].
In order to confirm the presence of quercetin and mangiferin in the silica co-precipitates, 5 mg of each sample were dissolved in 50 mL of acetone/DMSO for MNG co-precipitates and 50 mL of ethanol for Q co-precipitates ( Table 2). The contents in these solutions were measured by UV-VIS spectrophotometry at a wavelength of 365 nm using a previously obtained calibration curve. It was observed that in experiments 7 and 8-corresponding to the use of quercetin-the concentration was significantly lower than in experiments 5 and 6, where the concentration of mangiferin was six times higher than quercetin for the SAS1 process and three times higher for the SAS2 process. The fact that the amount of MNG was higher than Q is not unexpected since the operating conditions imposed to achieve the "body-mass" are suitable for MNG according to previous studies carried out on CAP/MNG co-precipitation.

Release Test
The release experiments on both SGF and SIF were performed for experiment 6, which contained mangiferin co-precipitates and experiment 8, which contained quercetin co-precipitates. These two experiments were chosen because the products contained the highest amounts of antioxidant in the analyzed samples ( Table 2). The release tests for the deposition of co-precipitates on silica particles were compared with release tests carried out on the antioxidant processed by SAS, co-precipitate CAP/Q and CAP/MNG in a 2:1 ratio and also with raw quercetin and mangiferin.
The release of composites with quercetin in both SGF and SIF is represented in Figures 7 and 8. It can be observed that in both diagrams the dissolution rate in the gastric fluid is lower than in the intestinal fluid for the deposition tests. This finding is due to the fact that quercetin is more soluble in the intestinal fluid [40]. It can be observed that the deposition of co-precipitate on silica using the SAS2 configuration increased the efficiency of the protection of the quercetin in SGF when compared to the co-precipitate with the same ratio but without silica [31], delaying the release during the first hour. However, it is very interesting to note that the deposition experiment showed a higher and burst release in the SIF when compared with the co-precipitation experiment without silica in the first 15 min. Finally, quercetin had a higher dissolution rate in all experiments, with the amount released reaching 100% in the first 15 min for processed Q and deposited composite, 30 min for the co-precipitate, and almost 4 h for raw Q. This demonstrates the effectiveness of the SAS method for micronizing and increasing the dissolution rate of this type of compound while providing protection for the active principle, in this case, the antioxidant in the gastric fluid.

Release Test
The release experiments on both SGF and SIF were performed for experiment 6, which contained mangiferin co-precipitates and experiment 8, which contained quercetin co-precipitates. These two experiments were chosen because the products contained the highest amounts of antioxidant in the analyzed samples ( Table 2). The release tests for the deposition of co-precipitates on silica particles were compared with release tests carried out on the antioxidant processed by SAS, co-precipitate CAP/Q and CAP/MNG in a 2:1 ratio and also with raw quercetin and mangiferin.
The release of composites with quercetin in both SGF and SIF is represented in Figures 7 and 8. It can be observed that in both diagrams the dissolution rate in the gastric fluid is lower than in the intestinal fluid for the deposition tests. This finding is due to the fact that quercetin is more soluble in the intestinal fluid [40]. It can be observed that the deposition of co-precipitate on silica using the SAS2 configuration increased the efficiency of the protection of the quercetin in SGF when compared to the co-precipitate with the same ratio but without silica [31], delaying the release during the first hour. However, it is very interesting to note that the deposition experiment showed a higher and burst release in the SIF when compared with the co-precipitation experiment without silica in the first 15 min. Finally, quercetin had a higher dissolution rate in all experiments, with the amount released reaching 100% in the first 15 min for processed Q and deposited composite, 30 min for the co-precipitate, and almost 4 h for raw Q. This demonstrates the effectiveness of the SAS method for micronizing and increasing the dissolution rate of this type of compound while providing protection for the active principle, in this case, the antioxidant in the gastric fluid.   In the case of mangiferin deposition, the same trend as for quercetin was found (Figures 9 and  10). A higher dissolution rate was observed in the intestinal fluid than in the gastric fluids. The protective effect of the polymer and silica with the MNG was evident. In SFG fluid, the antioxidant released after 8 h for the co-precipitate and deposited composites were around 50%, while the raw MNG was 80% and MNG processed by the SAS process was almost completely dissolved. In this case, the deposition experiment carried out in this work did not show any difference when compared with the co-precipitated experiment at the same ratio but without silica, which was carried out in a previous study [34] in SGF fluid. It can be observed, as in the case of Q, that in SIF the deposition experiment showed a burst release in the first 15 min-in contrast to the co-precipitation experiment. This fact could be associated with higher efficiency of the controlled release of the antioxidant in coprecipitate experiments in the absence of silica particles.
If the target is protecting the MNG in the gastric fluids due to its beneficial properties, this must be released and absorbed in the intestinal fluids. The experiments were successful because, in the SIF, the deposition experiment carried out with the SAS1 configuration made it possible to create composites of many compounds with a result as good as with the co-precipitate, i.e., increasing the solubility to almost 100% in the SIF. Finally, SAS1 and SAS2 gave similar results, although the versatility offered by the SAS2 configuration is advantageous in the use of different co-solvents for the polymer and the active principle. In the case of mangiferin deposition, the same trend as for quercetin was found (Figures 9 and 10). A higher dissolution rate was observed in the intestinal fluid than in the gastric fluids. The protective effect of the polymer and silica with the MNG was evident. In SFG fluid, the antioxidant released after 8 h for the co-precipitate and deposited composites were around 50%, while the raw MNG was 80% and MNG processed by the SAS process was almost completely dissolved. In this case, the deposition experiment carried out in this work did not show any difference when compared with the co-precipitated experiment at the same ratio but without silica, which was carried out in a previous study [34] in SGF fluid. It can be observed, as in the case of Q, that in SIF the deposition experiment showed a burst release in the first 15 min-in contrast to the co-precipitation experiment. This fact could be associated with higher efficiency of the controlled release of the antioxidant in co-precipitate experiments in the absence of silica particles.

Conclusions
Pressure and temperature conditions affect the success of the deposition process of the coprecipitates. The use of conditions well above the critical point of the mixture favors the deposition of the precipitates on silica. The low CO2/dissolution flow rate ratio leads to precipitates with solvent residues and in less appreciable quantities. The deposition of the co-precipitates on the silica occurs mainly at the bottom of the precipitator under conditions far above of the MCP, with the presence of silica in the rest of the precipitator very limited. Both the SAS1 and SAS2 configurations are appropriate to deposit co-precipitates onto silica particles, with a slight difference observed in the morphology of the co-precipitate in the case of the MNG. The release of co-precipitate deposited on silica presents higher dissolution rates in SGF than in SIF for both antioxidants. The presence of silica inside the vessel during the SAS process provides mixed co-precipitate and matrix without an adverse effect on the benefits in terms of improving the protection of the active substance during the release test in SFG when compared to co-precipitate releases in the absence of silica. This approach also facilitated total and burst dissolution in SIF-in contrast to co-precipitate releases without the presence of silica. If the target is protecting the MNG in the gastric fluids due to its beneficial properties, this must be released and absorbed in the intestinal fluids. The experiments were successful because, in the SIF, the deposition experiment carried out with the SAS1 configuration made it possible to create composites of many compounds with a result as good as with the co-precipitate, i.e., increasing the solubility to almost 100% in the SIF. Finally, SAS1 and SAS2 gave similar results, although the versatility offered by the SAS2 configuration is advantageous in the use of different co-solvents for the polymer and the active principle.

Conclusions
Pressure and temperature conditions affect the success of the deposition process of the co-precipitates. The use of conditions well above the critical point of the mixture favors the deposition of the precipitates on silica. The low CO 2 /dissolution flow rate ratio leads to precipitates with solvent residues and in less appreciable quantities. The deposition of the co-precipitates on the silica occurs mainly at the bottom of the precipitator under conditions far above of the MCP, with the presence of silica in the rest of the precipitator very limited. Both the SAS1 and SAS2 configurations are appropriate to deposit co-precipitates onto silica particles, with a slight difference observed in the morphology of the co-precipitate in the case of the MNG. The release of co-precipitate deposited on silica presents higher dissolution rates in SGF than in SIF for both antioxidants. The presence of silica inside the vessel during the SAS process provides mixed co-precipitate and matrix without an adverse effect on the benefits in terms of improving the protection of the active substance during the release test in SFG when compared to co-precipitate releases in the absence of silica. This approach also facilitated total and burst dissolution in SIF-in contrast to co-precipitate releases without the presence of silica.