*Article* **Structural, Optical, and Sensing Properties of Nb-Doped ITO Thin Films Deposited by the Sol–Gel Method**

**Madalina Nicolescu <sup>1</sup> , Daiana Mitrea 1,\*, Cristian Hornoiu <sup>1</sup> , Silviu Preda <sup>1</sup> , Hermine Stroescu 1,\*, Mihai Anastasescu <sup>1</sup> , Jose Maria Calderon-Moreno 1, Luminita Predoana 1, Valentin Serban Teodorescu 2,3, Valentin-Adrian Maraloiu <sup>2</sup> , Maria Zaharescu <sup>1</sup> and Mariuca Gartner 1,\***


**Abstract:** The aim of the present study was the development of Nb-doped ITO thin films for carbon monoxide (CO) sensing applications. The detection of CO is imperious because of its high toxicity, with long-term exposure having a negative impact on human health. Using a feasible sol–gel method, the doped ITO thin films were prepared at room temperature and deposited onto various substrates (Si, SiO2/glass, and glass). The structural, morphological, and optical characterization was performed by the following techniques: X-ray diffractometry (XRD), atomic force microscopy (AFM), scanning electron microscopy (SEM), transmission electron microscopy (TEM), and UV/Vis/NIR spectroscopic ellipsometry (SE). The analysis revealed a crystalline structure and a low surface roughness of the doped ITO-based thin films. XTEM analysis (cross-sectional transmission electron microscopy) showed that the film has crystallites of the order of 5–10 nm and relatively large pores (around 3–5 nm in diameter). A transmittance value of 80% in the visible region and an optical band-gap energy of around 3.7 eV were found for dip-coated ITO/Nb films on SiO2/glass and glass supports. The EDX measurements proved the presence of Nb in the ITO film in a molar ratio of 3.7%, close to the intended one (4%). Gas testing measurements were carried out on the ITO undoped and doped thin films deposited on glass substrate. The presence of Nb in the ITO matrix increases the electrical signal and the sensitivity to CO detection, leading to the highest response for 2000 ppm CO concentration at working temperature of 300 ◦C.

**Keywords:** Nb-doped ITO thin films; Sol–gel; Optical properties; CO detection

#### **1. Introduction**

Indium tin oxide (ITO) is an n-type semiconductor with a wide energy band gap (3.7 eV), low electrical resistance, and high optical transparency in the visible domain. The development of ITO thin films is of a great interest in the scientific community as a result of their interesting properties, which make them possible candidates for different applications, such as optoelectronic devices [1–3], transparent conductive oxides [4], solar cells [5–7], gas sensors [8,9], biosensors [10–12], thermoelectric applications [13,14], and so on. ITO thin films can be prepared by various physical (magnetron sputtering [15–19], pulsed lased deposition [20], ion beam sputtering [21], and electron beam evaporation [22]) and chemical methods (sol–gel method [23,24], spray pyrolysis [25], and low-temperature combustion synthesis method [26]).

In terms of the gas sensing properties, these materials based on doped or undoped ITO thin films have been proven to detect formaldehyde [27], CO2 [28–30], CO [31–33], NO2 [34,35], chlorine [35], benzene [36], toluene [37], and ammonia gases [38–40]. Additionally, ITO thin films can sense ethanol [41,42] and water vapors [43]. Furthermore,

**Citation:** Nicolescu, M.; Mitrea, D.; Hornoiu, C.; Preda, S.; Stroescu, H.; Anastasescu, M.; Calderon-Moreno, J.M.; Predoana, L.; Teodorescu, V.S.; Maraloiu, V.-A.; et al. Structural, Optical, and Sensing Properties of Nb-Doped ITO Thin Films Deposited by the Sol–Gel Method. *Gels* **2022**, *8*, 717. https://doi.org/10.3390/ gels8110717

Academic Editor: Viorel-Puiu Paun

Received: 19 October 2022 Accepted: 3 November 2022 Published: 7 November 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

ITO-coated glass substrates were used in liquid crystals display (LCD) [44–47] and polymer dispersed liquid crystals device fabrication [48–50]. The suitable material for a certain application can be achieved by changing several parameters involved in the preparation process as follows: the deposition technique [16], the metal doping level [51], the pre- and final annealing temperatures [19], and the film thickness [22].

Carbon monoxide (CO) is among the most harmful gases, being associated with several health problems, even death depending on exposure time. CO is released into the environment because of the partial combustion of fuels from cars and domestic or industrial activities. Consequently, CO detection is necessary because of its odorless and colorless properties [52,53]. Over time, researchers have studied metal-oxides based systems, such as SnO2, In2O3, ZnO [8,54], and indium-tin oxide (ITO) [55,56], with the aim to develop CO sensors. According to scientific reports, the properties of ITO thin films can be tailored by doping with various metals: Ag [57], Ga [58], Cr [59], Zn [30,60], Ti [61], Nb [62], and so on. In contrast with metal-doped ITO films, there are few reports concerning Nb-doped ITO thin films [51,62–64]. Therefore, these films were successfully obtained by radio frequency (RF) sputtering, pulsed laser deposition (PLD), and sol–gel methods to investigate the transparent conductive oxide, optoelectronic, and electrochromic properties [51,62–64]. The sol–gel method is a versatile and efficient procedure for the preparation of pure and doped metal oxide films or powders [65,66], showing some advantages such as purity, homogeneity, possibility to introduce dopants in large quantities, low processing temperature, ease of manufacturing, control over the stoichiometry, composition, and viscosity [67].

In this work, we explored the structural, morphological, and optical properties of multilayer Nb-doped ITO thin films prepared by the sol–gel method, deposited onto different substrates (glass, SiO2/glass, and Si). The effects of the dopant (4% Nb) and of the type of substrate were examined by X-ray diffraction (XRD), atomic force microscopy (AFM), scanning electron microscopy (SEM), transmission electron microscopy (TEM), and spectroscopic ellipsometry (SE). The Nb doping level was chosen based on our previous work regarding the Zn-doped ITO as a CO2 sensor.

As aforementioned, our aim was also to examine the influence of Nb doping on the detection properties of ITO thin films. Although this type of material was previously studied for certain applications, as far as the authors are aware, there are no literature data regarding its sensing properties. Therefore, the novelty of this work was the improvement in the electrical properties of ITO film through Nb doping, making it a more sensitive material for CO detection. Accordingly, the sample with the most suitable properties (thickness and porosity) was used for gas measurements of CO.

#### **2. Results and Discussion**

#### *2.1. Structural Characterization*

XRD Analysis

Figure 1a–c shows the XRD profiles of multilayer ITO/Nb thin films. Different substrates (Si, SiO2/glass, and glass) were coated by five successive layers using the ITO/Nb sol–gel solution. To highlight the influence of the dopant on the ITO film structure, previously reported data [30] on the undoped ITO thin films are presented in Figure 1. The diffraction lines, corresponding to crystal planes (2 2 2), (4 0 0), (4 4 0), and (6 2 2), were observed for both ITO/Nb and ITO thin films. ITO, which is Sn-doped In2O3, crystalizes in the bixbyite-type cubic structure of In2O3, with Ia-3 space group (ICDD file no. 06-0416). Except for the diffraction line of Si (marked with an asterisk on the Figure 1a) belonging to the substrate, no other phases were detected in the XRD patterns, indicating that the Nb and Sn dopants were incorporated into the In2O3 structure. The doped samples present an improved crystallinity, based on the shape of the diffraction line (higher intensity and narrower width). The lattice constants are slightly larger for the doped samples, most likely owing to the incorporation of the dopants into the cubic bixbyite structure. The crystallite

size was estimated using Scherrer's equation [68] only for the crystal plane (222) and was found to be around 10 nm (Table 1).

$$D = \frac{0.94 \times \lambda}{\beta \times \cos \theta} \,\tag{1}$$

where *D* is the average size of the crystallites, *λ* is the X-ray wavelength, *β* is the full width at half the maximum intensity (FWHM), and *θ* is the location of the diffraction line (Bragg angle).

**Figure 1.** XRD patterns of ITO films undoped and doped with 4% Nb deposited on (**a**) Si, (**b**) SiO2/glass, and (**c**) glass.

**Table 1.** Structural parameters and crystallite size of undoped and Nb-doped ITO films deposited by the sol–gel method on three different substrates.


The ITO/Nb films, as well as undoped ITO films, deposited on the SiO2/glass type of support show a better crystallinity in relation to those deposited on glass or Si substrate.

#### *2.2. Morphological Studies*

#### 2.2.1. AFM Measurements

The surface morphology and the roughness of the ITO/Nb thin films were assessed by AFM. Figure 2 shows the 2D topographic AFM micrographs at the scale of (1 × 1) μm<sup>2</sup> for the ITO/Nb films deposited on three different substrates: glass (Figure 2a), SiO2/glass (Figure 2b), and Si (Figure 2c). As could be seen, the films are compact and exhibit a uniform structure of nanometric-sized particles (Figure 2) with the root-mean-square (RMS) roughness values in the range of 0.85–1.29 nm and average roughness in the range of 0.67–1.02 nm (Figure 3). It is suggested that the sol–gel deposition of a SiO2 layer between the glass substrate and the ITO/Nb film slightly increases the roughness of the ITO film, in comparison with the ITO film deposited directly on glass, related to a better crystallinity as observed in XRD. On the other hand, the deposition of the ITO film on Si leads to the lowest roughness in this series, indicating a denser layer (in agreement with the refractive index curves).

**Figure 2.** Topographic 2D AFM images scanned over an area of (1 μm × 1 μm), showing the morphology of the ITO/Nb films deposited on (**a**) glass, (**b**) SiO2/glass, and (**c**) Si.

**Figure 3.** RMS (solid fill) and average (gradient fill) roughness for the ITO/Nb films deposited on different substrates.

#### 2.2.2. SEM Investigation

SEM was also used for the characterization of ITO/Nb films. As can be seen in the tilted film micrographs of different magnifications (Figure 4), the surface of the film is very smooth. Figure 4a is a low-magnification (20,000×) micrograph showing a scratch on the film surface, where the Si substrate (the darker zone marked with an arrow) is exposed. Figure 4b is a higher-magnification image (100,000×) showing the step at the edge of the film on top of the Si substrate, indicating that the film thickness is above 20 nm. The inset is the magnification of the area marked with a square in Figure 4b.

**Figure 4.** SEM micrographs at different magnification: (**a**) 20,000× and (b) 100,000× of the tilted ITO/Nb film deposited on Si, showing the smooth film surface and the film thickness.

The elemental compositional analysis by energy dispersive spectroscopy (EDX) (Table 2) revealed that niobium is incorporated into the film. A cationic ratio Nb/(In+Sn) of 0.037 was determined from the EDX measurements for the Nb-doped film.

**Table 2.** Cation composition measured by EDX elemental analysis of the ITO/Nb film deposited on Si substrate.


#### 2.2.3. TEM Analysis

As revealed by TEM images, the thickness of the ITO/Nb film deposited on Si is generally between 26 and 29 nm for the main layer, which is in accordance with SEM and SE results.

The low-magnification XTEM images (Figure 5a) show that the film has crystallites of the order of 5–10 nm, as well as relatively large pores, around 3 to 5 nm in diameter. The deposited layers are not clearly distinguishable; instead, there is an apparent morphology in three layers separated by rows of pores (Figure 5a).

The high-resolution transmission electron microscopy (HRTEM) images of the ITO/Nb film on Si (Figure 6a) exhibit that the first layer is denser (1), followed by a pore area layer (2) and then another denser layer (3) of half thickness. In the upper area (4), no layers can be distinguished, but there is a mixture of crystallites and pores. The surface layer (5) looks like a "crust" and is less compact. No pores are present in this upper layer, leading to low roughness values as observed in AFM. The polycrystalline ITO film structure has no texture, as revealed by the SAED pattern exposed in Figure 5b. The Si substrate is oriented along the [1 1 0] zone axis and the Si reflections are connected by the white line in the SAED pattern. The HRTEM obtained in the thinner area of the XTEM specimen (Figure 6b) shows a dense morphology of the layer at the bottom of the film. This layer is from 4 to 5 nm thick and can be probably identified with the real first deposited layer. In the rest of the film, the pores appear in the film volume. These pores are aligned in the bottom part of the film

but are randomly arranged in the rest of the film. The pores are not clearly delimited and can also consist of less dense zones containing some amorphous material. The coherent lattice fringes in the HRTEM images (Figure 6b) revealed that the ITO crystallites are in the range of 5 and 10 nm. The morphology of the layer-by-layer deposited ITO film is strongly affected by the crystal growth process, because the final size of the ITO crystallites is bigger than the initial thickness of each deposited layer (about 5 nm). If we compare these results with the case of Zn-doped ITO films [30], it can be observed that the crystallization process and the total film thickness are influenced by the dopant nature [30].

**Figure 5.** (**a**) Low-magnification XTEM image of the ITO/Nb film on Si and (**b**) its SAED pattern.

**Figure 6.** HRTEM images of the ITO:Nb film on Si: (**a**) morphology of the layers and (**b**) crystallite size and coherent lattice fringes.

The morphology of the apparent layers is shown in Figure 6a. The ITO crystallites size is revealed in the image by the coherent lattice fringes areas (Figure 6b). At the interface of the ITO film with the Si substrate, a SiO2 layer with a thickness of about 3 nm is formed.

#### *2.3. Optical Characterization*

#### SE in UV/Vis/NIR Domain

In the UV/Vis/NIR ellipsometric data analysis, the "General Oscillator" model [69] was applied to the ITO/Nb structure considering Tauc–Lorentz and Drude oscillators. The surface roughness was considered a mixture of 50% material (film) and 50% voids (air) and was fitted with the Bruggeman's effective medium approximation (B-EMA) [70]. The layer thicknesses (dfilm), the optical constants (refractive index—n and extinction coefficient—k), and the band gap energy (Eg) computed by Tauc formula [71] of the ITO/Nb films evaluated from the best fit are presented in Figure 7 and Table 3. A regression analysis of optical data, based on MSE, was used to evaluate the fit quality [69].

**Figure 7.** Optical constants (**a**) n, (**b**) k, (**c**) optical band gap—Eg, thickness, porosity—P, and (**d**) transmission—T of undoped and doped ITO thin films.

**Table 3.** Parameters determined by SE analysis of ITO/Nb thin films for different substrates.


\* Note: dSiO2 is SiO2 thickness and drough is the thickness of the roughness; \*\* note: n, T, and P are calculated for λ = 500 nm.

The porosity (P) of the films was calculated with the following formula [72]:

$$P = \left[1 - \frac{n^2 - 1}{n\_d^2 - 1}\right] \times 100 \text{ (\%)}\,\text{\AA} \tag{2}$$

where *nd* = 1.92 is the refractive index of the pore-free ITO (at λ = 500 nm) from WASE program and n is the refractive index of the ITO/Nb film at the same wavelength (Figure 7c).

The ITO/Nb films deposited on Si (8.44%) have the lowest porosity compared with those deposited on SiO2/glass (27.09%) and glass (38.32%) (see Table 3). The transmission spectra (T) of the ITO/Nb films measured in the 250–900 nm spectral range are shown in Figure 7 and their values at λ = 500 nm are indicated in Table 3. The ITO/Nb films deposited on glass and SiO2/glass exhibit a good transmittance (~80%) from visible and increase to 85% at 800 nm. It was observed that the Nb-doping of ITO reduces the band gap of thin films and can be attributed to the Burstein–Moss shift [73] in the visible domain (Figure 7c).

#### *2.4. Gas Sensing Measurements*

#### CO Sensing Measurements

Both the comprehension of the gas/solid interaction mechanism and identification of active regions in the films (surface, grain, and grain boundaries), which are implied in analyzed gas sensing, were assessed through complex impedance analysis. Using Nyquist plots (Z" vs. Z'), the impedance measurement results (Z = Z'+jZ, where Z' and Z" were the real and imaginary components, respectively) were represented.

Upon the exposure to a reducing gas, the resistance of the undoped and ITO/Nb films deposited on glass decreased, while the exposure to air led to an increase in this parameter. From our gas measurement results, it was concluded that the investigated films exhibited an n-type conductivity, as a consequence of the changes in terms of resistance of the films, function of the reducing gas, or air exposure.

From the intersection of the semicircle in Nyquist plots (Z" vs. Z'), we can determine DC-resistance for our films. In Figure 8, Nyquist plots are presented for ITO/Nb glass at 300 ◦C for different CO concentrations. In Figure 9, the electrical response of ITO/Nb glass and ITO glass film is plotted for various concentrations of CO function of the working temperature. The difference in sensitivity between the two samples can be associated with the presence of Nb in the ITO film.

**Figure 8.** Nyquist plots for the ITO/Nb glass film at 300 ◦C for different CO concentrations.

As observed, the response of the Nb-doped ITO/glass sample exhibits a maximum at 300 ◦C (Rair/RCO = 5), which will be considered the optimum working temperature of the material. It can also be stated that the films are most sensitive to the 2000 ppm CO concentration. The maximum sensitivity of ITO/Nb glass is approximately four times higher than the data achieved for Nb-doped TiO2 samples deposited in similar conditions through a sol–gel approach [74].

(**b**)

**Figure 9.** Electrical response of (**a**) ITO/Nb glass and (**b**) ITO/glass film for various concentrations of CO versus the working temperature.

#### **3. Conclusions**

ITO/Nb thin films were successfully deposited onto three different substrates through the sol–gel method. XRD analysis proved the polycrystalline nature of the films. AFM measurements indicated that all ITO/Nb films exhibit low surface roughness values, below 1 nm. SEM investigation revealed that the films are continuous, homogeneous, and adherent to the substrate. TEM analysis showed that the ITO/Nb films are very thin (26–29 nm), in agreement with SEM and SE, but with a complex morphology (a detailed study will follow in a next paper). The low-magnification XTEM images show that the film has crystallites of the order of 5–10 nm, as well as relatively large pores, around 3 to 5 nm in diameter, as also seen in the coherent lattice fringes of the HRTEM. The polycrystalline ITO film structure has no texture, as observed in SAED patterns. The morphology of the layer-by-layer deposited ITO film is significantly affected by the crystal growth process, because the final size of the ITO crystallites is bigger than the initial thickness (~5 nm) of each deposited layer. The Nb doping of ITO reduced the band gap of the films and can be attributed to the Burstein–Moss shift in the visible domain. The optical transmittance of the films deposited on transparent substrates (glass and SiO2/glass) was found to exceed 80%. The detection properties were characterized in terms of resistance and gas-sensing response. It was found

that the response of the ITO/Nb glass sample exhibits a maximum at 300 ◦C (Rair/RCO = 5) and that the ITO films are most sensitive to the 2000 ppm CO concentration. The sensitivity data are promising, but still preliminary, and will be expanded in future studies.

#### **4. Materials and Methods**

#### *4.1. Thin Film Deposition*

The Nb-doped ITO (ITO/Nb) films were prepared by the sol–gel method on the investigated substrates (glass, SiO2/glass, and Si) using the following as precursors: indium nitrate and 2-tin-ethyl hexanoate as In2O3 and SnO2 sources, 2,4-pentanedione as chelating agent, and niobium (V) ethoxide as dopant. Figure 10 describes the procedure for ITO thin films preparation: In(NO3)3·H2O and 2-tin-ethyl hexanoate solutions of 0.1 M concentration were homogenized by magnetic stirring at room temperature, obtaining a clear transparent solution. The dopant precursor (niobium (V) ethoxide) was added in the solution after 30 min of homogenization. Acetyl-acetone was added after 30 min and the homogenization continued for 3 h at room temperature. A light-yellow sol was obtained and it was kept at room temperature for 24 h. The as-obtained sol was used for the thin film deposition. The obtaining of ITO/Nb films with five layers was carried out by repetitive depositions (at a 5 cm/min withdrawal rate). The final films were achieved after 2 h of annealing treatment at 400 ◦C, with the heating rate of 5 ◦C/min. For the SiO2-coated glass substrate (SiO2/glass), the SiO2 layer was prepared according to the sol–gel method as presented in our previous work [75], preventing the diffusion of some elements from glass to ITO.

**Figure 10.** The flow chart of the ITO/Nb thin films preparation.

#### *4.2. Thin Film Characterization*

The structure of the ITO/Nb films was evaluated by the X-ray diffraction (XRD) method. XRD patterns were recorded using a Rigaku Ultima IV multifunctional diffraction system (Rigaku Corp., Tokyo, Japan), with Cu Kα (λ = 1.5406 Å) radiation, generated at a voltage of 30 kV and a current of 30 mA. The diffractometer was set in thin film geometry with a fixed incidence angle at α = 0.5◦. The measurements were performed at a scan rate of 5◦ (2θ)/min over a range of 5–90◦. Crystallite size was obtained from the Scherrer's formula only for the crystal plane (222).

AFM measurements were performed with an XE-100 apparatus (Park Systems) selecting the so-called non-contact working mode, in order to decrease the tip–sample interaction. All AFM images were registered using NCLR tips (Nanosensors™), with less than 8 nm radius of curvature. The AFM micrographs were processed with XEI (v.1.8.0) Image Processing Program developed by Park Systems for tilt correction and roughness evaluation.

Microstructural evaluation of the samples was achieved by SEM investigations using a FEI Quanta 3D microscope operating in the range of 5 and 30 kV.

TEM analysis working in low and high resolution as well as selected area electron diffraction (SAED) using a JEOL ARM200F analytical electron microscope operated at 200 kV was performed for systematic morphological investigations of the prepared thin films. The sample was prepared through the classical method of cross section by cutting 2 × 1 mm2 pieces, gluing them face to face, followed by mechanical polishing and final ionic thinning with the help of a Gatan PIPS System.

SE measurements were carried out at room temperature on J.A. Woollam Co. Inc. (Lincoln, NE, USA) equipment composed of a variable angle spectroscopic ellipsometer. The SE spectra were recorded in the 300–1700 nm (UV/Vis/NIR) wavelength range with a 10 nm step, at an incident angle of 70◦. For multi-parameter fitting, WASE program provided by Woollam was used. To minimize the difference (mean square error—MSE) between the experimental and the theoretical data, an iterative least-squares method was used. From the ellipsometric data analysis, the film thickness and the refractive index (n) were obtained with an accuracy of ±0.2 nm and ±0.005, respectively. The optical transmission was measured with the same equipment at a 0◦ incidence angle.

The ITO/Nb films deposited on glass were evaluated for gas sensing performances by impedance measurements. The four-point probe method inside a Probostat standard cell was used for gas sensing measurements. The samples were placed in a controlled atmosphere under a continuous gas flow of 177 mL/min, using a calibrated system of mass-flow controllers. Air and CO were mixed inside a vessel placed before the inlet of the impedance measurement cell. The electrical measurements were performed with a fourprobe method AC impedance spectrometer equipped with a Solartron 1260 electrochemical interface, with an applied AC bias amplitude of 500 mV. Electrochemical impedance spectra (EIS) were recorded in the frequency domain from 3 MHz to 100 Hz at temperatures of 200 to 400 ◦C with a ProboStat cell (NorECs, Oslo, Norway).

**Author Contributions:** Conceptualization, M.G.; methodology, L.P. and M.Z.; formal analysis, M.A., J.M.C.-M., M.N., H.S., C.H., S.P., V.-A.M., and V.S.T.; investigation, M.A., J.M.C.-M., M.N., H.S., C.H., S.P.,L.P., V.-A.M., and V.S.T.; data curation, M.G.; writing—original draft preparation, M.A., J.M.C.-M., M.N., H.S., C.H., S.P., L.P., V.-A.M., V.S.T. and M.Z.; writing—review and editing, M.G., M.A., J.M.C.- M. and D.M.; visualization, L.P. and M.Z.; supervision, M.Z.; project administration, M.G.; funding acquisition, M.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** The support of the grant number PN-III-P2-2.1-PED-2019-2073 (308 PED/2020) is gratefully acknowledged.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

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

**Acknowledgments:** The paper was carried out within the research program "Science of Surfaces and Thin Layers" of the "Ilie Murgulescu" Institute of Physical Chemistry. The support of the Romanian Government that allowed for the acquisition of the research infrastructure under POSCCE O 2.2.1 project INFRANANOCHEM—No. 19/01.03.2009 9 is acknowledged.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Dan Eduard Mihaiescu 1, Daniela Istrati 1, Alina Moros, an 1,\* , Maria Stanca <sup>2</sup> , Bogdan Purcăreanu 3,4,\* , Rodica Cristescu <sup>5</sup> , Bogdan S, tefan Vasile <sup>3</sup> and Roxana Doina Trus, ca <sup>3</sup>**


**Abstract:** As a third-generation β-lactam antibiotic, cefotaxime shows a broad-spectrum with Grampositive and Gram-negative bacteria activity and is included in WHO's essential drug list. In order to obtain new materials with sustained release properties, the present research focuses on the study of cefotaxime absorption and desorption from different functionalized mesoporous silica supports. The MCM-41-type nanostructured mesoporous silica support was synthesized by sol–gel technique using a tetraethyl orthosilicate (TEOS) route and cetyltrimethylammonium bromide (CTAB) as a surfactant, at room temperature and normal pressure. The obtained mesoporous material (MCM-41 class) was characterized through nuclear magnetic resonance (NMR), scanning electron microscopy (SEM), high-resolution transmission electron microscopy (HR-TEM), N2 absorption–desorption (BET) and Fourier transform infrared spectroscopy (FT-IR), proving a good micro-structured homogeneity (SEM images), a high surface area (BET, 1029 m2/g) correlated with high silanolic activity (Q3/Q4 peak ratio from 29Si MAS-NMR), and an expected uniform hexagonal structure (2–3 nm, HRTEM). In order to non-destructively link the antibiotic compound on the solid phase, MCM-41 was further functionalized in two steps: with aminopropyl trimethoxysilane (APTMS) and glutaraldehyde (GA). Three cefotaxime-loaded materials were comparatively studied for low release capacity: the reference material with adsorbed cefotaxime on MCM-41, MCM-41/APS (aminopropyl silyl surface functionalization) adsorbed cefotaxime material, and APTMS–GA bounded MCM-41—cefotaxime material. The slow-release profiles were obtained by using an on-flow modified HPLC system. A significant improved release capacity was identified in the case of MCM-41/APS/GA—cefotaxime due to the covalent surface grafting of the biological active compound, recommending this class of materials as an effective carrier of bioactive compounds in wound dressing, anti-biofilm coatings, advanced drugs, and other related applications.

**Keywords:** MCM-41; sol–gel; low release; cefotaxime; mesoporous material

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **1. Introduction**

The discovery of MCM-41-type mesoporous silica could be consider a milestone in the field of silica-based materials due to the uniform hexagonal pore structure, the high surface area [1], chemical, thermal stability, biocompatibility, and high loading capacity [2,3]. The abundance of silanol groups on the surface of mesoporous silica nanoparticles (MSNs) allows for enhanced flexibility for surface modification through functionalization; thus, widening their range of application. MSNs are used in various applications such as therapeutic/health, drug delivery, agricultural fields, food industries, optoelectronic sensing [4,5], catalysis [6–9], gas separation [10,11], and lanthanides recovery

**Citation:** Mihaiescu, D.E.; Istrati, D.; Moros,an, A.; Stanca, M.; Purc ˘areanu, B.; Cristescu, R.; Vasile, B.S, .; Trus, ca, R.D. Low Release Study of Cefotaxime by Functionalized Mesoporous Silica Nanomaterials. *Gels* **2022**, *8*, 711. https://doi.org/

Academic Editor: Andrei Jitianu

10.3390/gels8110711

Received: 4 October 2022 Accepted: 2 November 2022 Published: 3 November 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

from waste water [12,13]. In addition, organic/inorganic micro/nanocomposites using oxidic-based microporous structures and different strategies of surfaces functionalization are used as innovative solutions for advanced waste water treatments and pathogen inactivation approaches [14,15]. An important application of mesoporous silica consists of its usage as a carrier for biologically active compounds, such as gemcitabine [16] and quercetin [17], both used in cancer treatment. The main advantages of using mesoporous silica nanocomposites in biomedical research and application fields include their good biocompatibility, increased bio-availability of the target compounds for the tissue of interest (as example tumor tissue), sustained or controlled release of the bioactive compounds, and reduced toxicity of these compounds in healthy tissues [18,19].

V. Pardhi and collaborators improved the dissolution rate of niclosamide in the human body by using mesoporous silica as a delivery system and studied the influence of mesoporous silica morphology on the dissolution rate [20]. The study demonstrated that niclosamide loaded on mesoporous silica has a higher dissolution rate than the drug as such and silica morphology also influences the dissolution rate, this being lower for non-porous silica and higher for non-ordered mesoporous silica.

V. Nairi et al. studied the adsorption and release of ampicillin from mesoporous silica matrices, which differ from each other either by pore size or chemical properties due to the surface functionalization. The compared materials differ in pore size equivalent to electric charge (SBA-15 and MCM-41) and different surface electric charge (SBA-15 and SBA-15- NH2). Both the adsorption and release of ampicillin are influenced by the surface electric charge more than the pore size. The surface negative charge of the two non-functionalized matrices favors rapid release at pH 7.4, being slightly higher for MCM-41. As the MCM-41 particles have smaller dimensions than the SBA-15 ones, a higher amount of drug was adsorbed. In the case of the SBA-15-NH2 matrix, the electrostatic interactions favor an antibiotic sustained release that is slower than that of the nonfunctionalized material [21]. The biological responses could be enhanced by modifying the surface of mesoporous silica, such as the surface functionalization by attaching specific functional groups or molecules. The most common methods of functionalization are co-condensation and postgrafting. The selected approach influences both the physicochemical interaction and surface chemistry modification. The co-condensation or direct synthesis functionalization method can be performed in a single reaction vessel; and thus, has a reduced number of steps and synthesis time, leading to a more homogeneous distribution of functional groups on the surface. On the other hand, the post-grafting method uses simple and mild conditions, and involves the subsequent surface modification by direct grafting or secondary grafting. Since only molecules that are small enough can diffuse into the mesoporous matrix, this functionalization method is limited by both the pore size and architecture [22,23].

The special structure of mesoporous silica offers an efficient protection of biomolecules susceptible to metabolic transformations in the gastrointestinal tract after oral administration [24]. Silicon is a structural component of connective tissues, being essential for bones [25] and skin [26]. Acute and chronic toxicity studies have shown that the mesoporous silica does not have adverse effects for oral administration due to its low water solubility; however, the long-term pulmonary exposure to amorphous or crystalline silica (solid dispersions in the gas phase) has been associated to serious respiratory disorders [27,28].

Cefotaxime is a third-generation cephalosporin antibiotic that when it is intravenously or intramuscularly administered, becomes active against Gram-positive and Gram-negative bacteria, except pseudomonas. It is used to treat meningitis, lower respiratory tract infections, urinary tract, inflammatory pelvic diseases, skin infections, and gonorrhea [29,30]. Cefotaxime has the ability to inhibit bacterial cell wall synthesis compared to penicillin by blocking the transpeptidation step in peptidoglycan biosynthesis [31]. The cell envelope of Gram-positive and Gram-negative bacteria consist of a plasma membrane and cell wall. The difference between those two is that the Gram-negative bacteria poses an additional outer impermeable membrane to large molecules. The cell walls are similar (a single layer of peptidoglycan); however, they differ in thickness: 20–80 nm for Gram-positive bacteria

and less than 10 nm for Gram-negative bacteria [32,33]. A desorption study of cefotaxime from MCM-41 and MCM-41/APS materials proves significant advantages of the amino groups' (-NH2) surface functionalization [34].

The present research focuses on the study of the absorption/desorption of cefotaxime from functionalized mesoporous silica support, specifically focusing on the significant release difference between the non-destructively linked cefotaxime and physically adsorbed compound on the solid phase. Furthermore, a different low-release experimental system was proposed, using a modified HPLC system (avoiding sampling and HPLC injection) that is able to provide on-flow analysis (more than 4000 sampling points for a single experiment).

#### **2. Results and Discussion**

The synthesis of cefotaxime-bonded MCM-41/APS/GA composites involved several reaction steps, followed by advanced investigations of the intermediate and final materials: MCM-41 synthesis by room temperature/pressure sol–gel method, APTMS surface functionalization in order to provide a -NH2 linking center for the glutaraldehyde surface grafting step, followed by the final cefotaxime (non-destructive) linking to the free thiazolyl amino group. Further desorption studies were performed in order to provide the release profiles of three materials: MCM-41—cefotaxime (3:10 (*w/w*)), MCM-41/APS—cefotaxime (3:10 (*w/w*)) loaded materials (physical adsorption), and covalent-linked cefotaxime MCM-41/APS/GA (3:10 (*w/w*)). In order to obtain an effective concentration of the drug in the desorption environment, the 3:10 (*w/w*) loading level of cefotaxime was established by the previous bioactive compound loading studies [35], which provided a significant out-ofpore compound concentration (the quick release of out-of-pore cefotaxime provides a fast response in the first stage of treatment), followed by a sustained release (in-pore desorption, in order to sustain the already established cefotaxime concentration). The cefotaxime GA cross-linked composite provides a significantly lower drug release due to the involvement of the secondary aldiminic bound cleavage from the solid support. This behavior would be a significant advantage in wound dressing, advanced drugs design, and antibiofilm coatings applications because the pH drop from bacterial activity will be followed by the desorption enhancement due to an accelerated aldiminic-bound cleavage.

SEM images (Figure 1) of MCM-41 (a), MCM-41 with particle size details (b), MCM-41/APS (c), MCM-41/APS/GA (d), and cefotaxime-linked MCM-41/APS/GA (e) materials exhibit a relative uniform microgranular structure both in shape and particle size. APTMS surface functionalization, the glutaraldehyde surface-grafting step, and the final cefotaximeloading process do not affect the morphology of the composites; in terms of both shape and particle size, a slight particle aggregation can be observed (e) for the loaded material, as expected from the out-of-pore cefotaxime loading. The microstructured spherical particle distribution was the subject of several MCM-41 synthesis optimization steps, avoiding as possible the low-diameter nanoparticles (almost pharmaceutical release applications avoiding nanometer scale-range particles due to the significant tissue penetration).

Functionalization of MCM-41 with APTMS is demonstrated by the appearance of the peak corresponding to nitrogen; in addition, the peaks corresponding to silicon and oxygen are only revealed by energy dispersive spectroscopy (EDS) (Figure 2).

Figure 3 exhibits HR-TEM images of non-functionalized MCM-41 material (because of the high electron energy, surface grafted materials are less suitable for this analysis). The HR-TEM images confirm the MCM-41 granular mesoporous silica morphology (200–1500 nm particle size) and its mesoporous hexagonal structure. For the MCM-41 synthesis operated at normal pressure and temperature, the obtained structure shows a good hexagonal ordering degree and an average pore size of about 2–3 nm, proved also by the high surface area.

**Figure 1.** SEM images for initial MCM-41 material (**a**), MCM-41 with particle size details (**b**), MCM-41/APS (**c**), MCM-41/APS/GA (**d**) and MCM-41/APS/GA-cefotaxime (**e**).

**Figure 2.** The EDS spectrum for MCM-41/APS.

**Figure 3.** Representative HR-TEM images of non-functionalized MCM-41 material for 20 nm (**a**) and MCM-41/APS for 10 nm (**b**) scale bars.

According to IUPAC, the N2 adsorption/desorption isotherm is type IV, specific to MCM-41 type materials (Figure 4). This feature highlights the characteristics of the MCM-41 material: the specific surface area of 1029.06 m2 g−<sup>1</sup> pore sizes of 2.41 nm and pore volume of 0.62 cm<sup>3</sup> g<sup>−</sup>1, being characteristic of mesoporous materials, which also confirms our HR-TEM data. The pore size distribution (Figure 4) confirms the uniformity of the pore diameter.

**Figure 4.** N2 adsorption/desorption isotherm and pore size distribution for MCM-41.

29Si –NMR MAS analysis of the MCM-41 and MCM-41/APS materials highlight some unique structural aspects of the mesoporous silica silanolic surface, such as the presence of silicon atoms Q4, Q3, and Q2 (single and geminal silanolic groups). Figure 5a shows a small Q2 geminal and high Q3 bands, evidencing the presence of almost single silanolic groups on the silica surface. The APTMS functionalization provides effective binding of MCM-41 single silanolic groups (Figure 5b), Q<sup>3</sup> band significant mitigation correlated with the presence of a T3 band, attesting to the existence of a new covalent silicon—carbon bound [36].

**Figure 5.** 29Si MAS NMR spectrum (8 kHz rotation frequency) of non-functional mesoporous silica (**a**) and APS-grafted (**b**) materials.

In order to correlate the specific vibrational information with the MAS-NMR results, further FT-IR and Raman investigations of the intermediate and final materials were performed.

The presence of the saturated chain of glutaraldehyde is confirmed by the absorption band at 2933 cm<sup>−</sup>1, with a significantly higher intensity compared to the MCM-41/APS band (2941 cm<sup>−</sup>1) correlated with the presence of the propyl chain (Figure 6).

**Figure 6.** Comparative horizontal attenuated total reflectance (HATR)-FT-IR spectra of MCM-41/cefotaxime, MCM-41/APS—cefotaxime, and MCM-41/APS/GA—cefotaxime (the final 3 cefotaxime-loaded materials used for the desorption experiments).

The comparative FT-IR and Raman spectra of MCM-41, MCM-41/APS, and MCM-41/APS/GA confirm the presence of the saturated chain of GA due to the presence of absorption bands centered at 2929 and 2871 cm<sup>−</sup>1, respectively, in Figure 7; and 2925 and 2913 cm<sup>−</sup>1, respectively, in Figure 8. The absorption bands centered at 2935 and 2878 cm<sup>−</sup>1, respectively, in Figure 7; and 2918 and 2892 cm<sup>−</sup>1, respectively, in Figure 8, confirm the presence of APS-saturated chains; while the -NH2 groups can be observed at 3361 and 3290 cm<sup>−</sup>1, respectively, in Figure 7; and 3369 and 3310 cm<sup>−</sup>1, respectively, in Figure 8 (correlated to the presence of the peak corresponding to nitrogen in the EDS spectrum Figure 2 and T<sup>3</sup> peak from MAS-NMR Figure 5). Moreover, the intensity of the characteristic bands of -NH2 groups from the MCM-41/APS/GA significantly decreased in comparison to the mesoporous silica–amino groups.

**Figure 7.** Comparative HATR-FT-IR spectra of MCM-41, MCM-41/APS, and MCM-41/APS/ GA materials.

In order to estimate the release particularities linked to the different interactions between the biologically active compound and the solid mesoporous support, the final

step of this study was related to the comparative low-release capacity evaluation of the three loaded materials: MCM-41—cefotaxime, MCM-41/APS—cefotaxime, and MCM-41/APS/GA—cefotaxime. The experimental system (Figure 9) uses an Agilent 1200 series HPLC system with a manual injector, with several significant modifications: the solvent feed line of the upper mobile phase bottle was inserted in the release vessel, directly passing the liquid phase to the HPLC pump and further through the injection loop; the column was bypassed (the liquid flow from the pump was directly connected with the UV–Vis detector); and the detector exhaust line was redirected to the release vessel for the purpose of closing the whole on-flow loop. Together with the insertion of the desorption bag in the release solution, a formal blank manual injection was performed to start the data acquisition, the acquired "chromatogram" actually containing the desired desorption profile. For a proper decontamination of the whole system and cross-contamination mitigation, several washing steps were performed after each experiment, with different liquid phase pH values and baseline monitoring. For quantitative measurements, the calibration profiles were obtained and a mean value of 50 data points from each calibration profile was used for the linear regression calculations (the data points were extracted from the flat final region of the calibration profiles; the calibration samples were fully dissoluted, yielding a final flat profile of the desorption curve after 10 min of release).

**Figure 9.** The experimental system (modified HPLC) used for the desorption profile acquisition.

In comparison to other on-flow or discontinuous sampling systems, our proposed desorption system provides significant advantages: high sampling rates (5–20 Hz range, feasible values for the proposed experiments are lower than 5 Hz due to the low absorbance changes compared to a normal HPLC analysis) and smooth shape of the desorption profile; low detection limits—correlated with the high HPLC detector performance; a low dead volume due to the HPLC lines; low volume of the UV–Vis detection flow cell; and long-term desorption capability (more than 48 h with a proper experiment setup).

All the desorption experiments were performed in ultrapure water (at 6.5 pH), using a similar experiment setup and acquisition conditions.

A significant release difference between the non-destructively linked cefotaxime and physically adsorbed compound on the APS functionalized MCM-41 solid phase can be established from Figure 10. Significant release differences are observed in the two important regions of the desorption profiles: the first region involves a fast release of cefotaxime in the first 20–30 min, while the final region, with a significant slope difference, involves a low, sustained release of the compound. The GA covalent-linked cefotaxime composite shows an intermediate release concentration (between the other two materials) after 120 min; however, it shows a significant lower release trend. The higher cefotaxime release concentration, compared to the MCM-41/APS adsorbed compound one, could be explained

by a lower in-pore surface access of cefotaxime after the glutaraldehyde link (in the final loading stage of composite synthesis), by a higher amount of the out-of-pore cefotaxime available for a fast release in the first 20–30 min. Following the final slope trend of the two profiles, there is expected to be a higher release concentration of the MCM-41/APS adsorbed cefotaxime at several hours of desorption time (overpassing the GA-bonded cefotaxime profile) due to the differences in the release mechanism (desorption and aldiminic bond cleavage, respectively).

**Figure 10.** Desorption profiles of cefotaxime on the three support types.

The significant slope difference of the two profiles in the first 20 min of desorption can be explained by a lower interaction of the out-of-pore cefotaxime with the GA-grafted surface due to the hydrophobic saturated carbon chain of glutaraldehyde, yielding an increased release of the out-of-pore cefotaxime.

Because of the significant differences of the silanolic access of loaded cefotaxime, the non-functionalized MCM-41 material shows a significant release difference when compared to functionalized materials. In addition, due to the higher access of water at the polar silanolic surface, an enhanced release of cefotaxime is expected (because of the hydrogen bond interaction of cefotaxime with the silanolic groups, expected at -NH-, -NH2, and -OH sites; the water dissolution rate will be controlled by the water molecules access at the silanolic surface. For the APS-grafted surface, the water access at the silanolic groups is significantly reduced, providing a lower release compared to MCM-41. The APS/GA material exposes a significantly lower polarity surface to the water environment; and thus, a faster release of the out-of-pore absorbed cefotaxime is expected). Despite the significant differences, a similarity between the final slopes of the MCM-41 and MCM-41/APS materials must be noticed, proving a similar in-pore diffusion mechanism. Certainly, this result correlated with the significant slope difference of the GA linked-material, which should sustain the desorption mechanism differences.

#### **3. Conclusions**

The present work describes the synthesis, characterization, and comparative low release study of a new mesoporous composite material, with covalent cefotaxime grafting on a MCM-41/propylamino/glutaraldehyde (MCM-41/APS/GA) solid phase.

The obtained materials were characterized by BET, FT-IR, RAMAN, SEM, HR-TEM, and NMR investigation methods. MCM-41 synthesis was conducted to micrometer range particles (avoiding low-diameter silica nanoparticles), with a 1029 m2g<sup>−</sup><sup>1</sup> specific surface area (BET), pore size of 2.41 nm, and pore volume of 0.62 cm<sup>3</sup> g<sup>−</sup>1; and a high silanolic surface trough Q<sup>3</sup> silanolic group peak from the 29Si MAS NMR spectrum. The further APS surface functionalization (performed by MCM-41—APTMS reaction) was proved by the presence of free -NH2 groups (EDS, IR, Raman), and GA surface grafting was obtained by the glutaraldehyde aldiminic link. The last stage involved MCM-41, MCM-41/APS, and MCM-41/APS/GA materials loading with cefotaxime (3:10 (*w/w*).

We proposed a low release experimental system, using a modified HPLC system (avoiding sampling and HPLC injection) that is able to provide on-flow analysis (more than 4000 sampling points for a single experiment). The three mesoporous materials: MCM-41 cefotaxime, MCM-41/APS—cefotaxime, and MCM41/APS/GA-cefotaxime were tested in desorption experiments in water. A significant release difference between the nondestructively linked cefotaxime, physically adsorbed compound on the APS functionalized, MCM-41 solid phase was established. Moreover, significant release differences were observed in the two important regions of the desorption profiles: the first region involving a fast release of cefotaxime in the first 20–30 min and the final region, with a significant slope difference involving a low, sustained release of the compound. The GA covalentlinked cefotaxime composite shows an intermediate release concentration (between the other two materials) after 120 min; however, it shows a significant lower release trend. The low release capacity of the mesoporous glutaraldehyde-linked cefotaxime composite would be a significant advantage in wound dressing applications, advanced drugs design, antibiofilm coatings, and other related applications.

#### **4. Materials and Methods**

#### *4.1. Reagents and Equipment*

Cefotaxime hydrochloride, tetraethylorthosilicate (TEOS), glutaraldehyde (GA), aminopropyl trimethoxysilane (APTMS), methanol, ethanol, acetonitrile (ACN), and ammonium hydroxide solution 25% were purchased from Sigma-Aldrich and were used without further purification. Cetyltrimethylammonium bromide (CTAB) was purchased from Fluka and used as such. All the chemicals were of analytical purity and used as received.

Scanning electron microscope (SEM) QUANTA INSPECT F (Thermo Fisher—formerly FEI, Eindhoven, The Netherlands) field emission gun of resolution 1.2 nm was used to investigate the sample surface morphology, using energy dispersive X-ray (EDX) with the resolution to MnKα 133 eV.

High-resolution electron transmission microscope (HR-TEM) were performed on a Tecnai G2 F30 S-TWIN equipped with energy dispersive spectroscopy (EDS) as well as a selected area electron diffraction detector (SAED) purchased from Thermo Fisher—formerly FEI (Hillsboro, OR, USA). The microscope was operated in transmission mode at 300 kV; the HR-TEM point resolution was 2 Å and line resolution was 1 Å.

Fourier transform infrared (FT-IR) spectra were recorded using a Nicolet iS50FT-IR (Thermo Nicolet, Massachusetts, USA) spectrometer equipped with a DTGS detector and Raman accessory. The measurements were carried out in the range of 4000–400 cm<sup>−</sup>1, using the resolution 4 cm−<sup>1</sup> and 100 scans per spectrum. All the spectra were recorded using horizontal attenuated total reflectance (HATR) with diamond crystal. FT-Raman spectra were collected using the same device, using an InGaAs detector, a CaF2 beamsplitter, 100 scans per spectrum, and the excitation laser power at 0.50 W.

The release profiles were recorded using a modified HPLC system, as previously presented [35].

Brunauer–Emmett–Teller (BET) analysis. The nitrogen adsorption/desorption isotherms were registered at 77 K in the relative pressure range p/po = 0.005–1.0, by a NOVA 800 Gas Sorption Analyzer (Anton Paar QuantaTec, Inc., Boyton Beach, FL, USA). Data processing was performed using Kaomi software. Prior to adsorption measurements, the samples were degassed up to 180 ◦C under vacuum for 4 h. The standard Brunauer–Emmett–Teller (BET) equation was used to determine the specific surface area. The gas volume absorbed at a relative pressure p/po~1 allowed the estimation of the total pore volume. The Barrett–Joyner– Halenda (BJH) model was used to obtain the pore size distribution and mesopore volume from the desorption branch of the isotherm.

NMR was performed using a Bruker Avance III spectrometer (Bruker, MA, USA), with a 14.1 Tesla magnet (600 MHz proton resonance frequency). The samples were analyzed in solid state (29Si) using a 3.2 mm diameter rotor and a rotational frequency of 8 kHz. For 29Si nuclei, the NMR MAS "one pulse" technique was used.

#### *4.2. Synthesis*

The method of mesoporous silica synthesis is similar to that of previously described work [35]. Further, the mesoporous MCM-41 material was functionalized with propylamino groups by reaction with APTMS in acetonitrile.

#### Pressure Vessel Functionalization

Mesoporous silica (100 mg) was mixed with APTMS–ACN solution (1:4). The reaction mixture was placed in a pressure-tight reaction vessel and kept for 12 h at 85 ◦C. After cooling, the mixture was separated by centrifugation, washed with ultrapure water (at least 5×), and dried at 105 ◦C for 7 h.

In addition, a second functionalization method (microwave assisted) was attempted in order to improve the APS surface grafting; however, the results obtained sustained an in-pore polycondensation of the silane with a large surface area loss. This surface functionalization method could still be a good alternative to pressured vessel synthesis, using low microwave (MW) exposure times and reaction conditions optimization.

Further grafting of glutaraldehyde on the amino functionalized silica surface was performed as follows: the amino functionalized material (100 mg) was dispersed in ultrapure water (2 mL) and mixed with 20% solution of glutaraldehyde (2 mL). The obtained mixture was kept for 5 h at 55 ◦C. After cooling, the solid was separated by centrifugation and washed (at least 5×) times with ultrapure water. The final material was dried for 7 h at 55 ◦C.

Cefotaxime loading was performed as follows: the functionalized amino–glutaraldehyde material (100 mg) was mixed with cefotaxime (30 mg) in 10 mL methanol and subsequently, dried under vacuum in a rotary evaporator at 30 ◦C. The solvent evaporation process provides a cefotaxime concentration gradient, respectively, a high pore loading of the final product, and an out-of-pore distribution of the cefotaxime excess. A final drying stage was performed for 12 h at 55 ◦C. This method assures no bioactive compound loss in the loading stage (compared to the liquid phase loading methods that involve a solid–liquid partition).

**Author Contributions:** Conceptualization, D.E.M., D.I., A.M., M.S. and B.P.; methodology D.E.M., D.I., A.M., M.S., B.P., R.C., B.S, .V. and R.D.T.; validation, D.E.M., D.I. and A.M.; investigation, D.E.M., B.P., B.S, .V. and R.D.T.; resources, D.E.M. and R.C.; data curation, D.E.M., D.I., A.M., M.S. and B.P.; writing—original draft preparation, D.E.M., A.M. and M.S.; writing—review and editing, D.E.M., D.I., A.M. and R.C.; visualization, D.E.M., D.I., A.M., M.S., B.P., R.C., B.S, .V. and R.D.T.; supervision, D.E.M. and D.I. All authors have read and agreed to the published version of the manuscript.

**Funding:** The work has been funded by the Romanian National Authority for Scientific Research and Innovation, CNCS UEFISCDI, projects no. PN-III-P2-2.1-PTE-2016-0160, 49-PTE/2016 (PROZECHIMED) and PN-III-P4-ID-PCE-2016-0884, 142/2017 (BIOMATE).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Available on demand.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

