*Article* **Effective Carbon/TiO<sup>2</sup> Gel for Enhanced Adsorption and Demonstrable Visible Light Driven Photocatalytic Performance**

**Anam Safri and Ashleigh Jane Fletcher \***

Department of Chemical and Process Engineering, University of Strathclyde, Glasgow G1 1XJ, UK; anam.safri@strath.ac.uk

**\*** Correspondence: ashleigh.fletcher@strath.ac.uk; Tel.: +44-141-5482-431

**Abstract:** A new strategy to synthesise carbon/TiO<sup>2</sup> gel by a sol–gel method is proposed. Textural, morphological, and chemical properties were characterised in detail and the synthesised material was proven to be an active adsorbent, as well as a visible light photocatalyst. Homogenously distributed TiO<sup>2</sup> is mesoporous with high surface area and, hence, exhibited a high adsorption capacity. The adsorption equilibrium experimental data were well explained by the Sips isotherm model. Kinetic experiments demonstrated that experimental data fitted a pseudo second order model. The modification in electronic structure of TiO<sup>2</sup> resulted in a reduced bandgap compared to commercial P25. The absorption edge studied through UV-Vis shifted to the visible region, hence, daylight photocatalytic activity was efficient against degradation of MB dye, as an example pollutant molecule. The material was easily removed post treatment, demonstrating potential for employment in industrial water treatment processes.

**Keywords:** adsorption; carbon/TiO<sup>2</sup> gels; resorcinol formaldehyde RF/TiO<sup>2</sup> gels; photocatalysis; adsorption kinetics; methylene blue dye degradation

#### **1. Introduction**

Adsorption of carbon is perhaps the most widely used water treatment technique. However, there is an ongoing effort to develop efficient adsorbents with reduced regeneration costs. Currently, the combination of carbon and titanium dioxide (TiO2) appears to offer a promising route to obtain an adsorbent with self-regeneration properties. Additionally, the synergistic effect of both carbon and TiO<sup>2</sup> enhances the degradation process due to respective adsorptive and photocatalytic properties. Literature reports several studies to address the synergy of adsorption and photodegradation by experimental demonstration of various carbon/TiO<sup>2</sup> composite materials [1,2]. However, there is still a need to better understand the phenomenon of pollutant-adsorbent interactions in order to have a good knowledge to design an efficient water treatment process. Additionally, the improvement in design involves the type of materials and synthesis process employed to attain maximum efficiency of the system.

Previously, carbon has been combined with TiO<sup>2</sup> through various approaches, in different forms, such as carbon nanotubes [3–5], graphene [6–8], and activated carbon [9,10]. Lately, focus has been shifted to highly porous carbon materials as support matrix for industrial applications, due to the high surface area and tuneable porosity. Ideally, welldeveloped mesoporous structures with large pore volumes and uniform pore size distributions are preferred, due to enhanced accessible surface sites contributing to superior adsorption capacity of pollutants from the aqueous phase. However, the preparation process of these mesoporous carbons is costly and complicated, usually resulting in materials with moderate or low surface area. The efficiency of the material is also limited, since most TiO<sup>2</sup> nanoparticles incorporated in the pores of the carbon are unavailable for photocatalysis [11].

**Citation:** Safri, A.; Fletcher, A.J. Effective Carbon/TiO<sup>2</sup> Gel for Enhanced Adsorption and Demonstrable Visible Light Driven Photocatalytic Performance. *Gels* **2022**, *8*, 215. https://doi.org/ 10.3390/gels8040215

Academic Editors: Hiroyuki Takeno and Avinash J. Patil

Received: 11 February 2022 Accepted: 29 March 2022 Published: 1 April 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/).

Amongst mesoporous carbon materials, carbon gels are a new type of nanocarbon with potential applications in photocatalysis [2,12,13]. Carbon gels produced by polycondensation of resorcinol (R) with formaldehyde (F) are highly porous and have flexible properties. A comprehensive review of sol–gel synthesis of RF gel reveals that the material can be easily tailored to attain desired properties, mainly tuneable porosity, and acts as a support for metals [14]. Hence, RF gels can be promising materials for water treatment applications, mainly due to their stability, owing to aromatic resorcinol rings and their overall interconnected mesoporous carbon structure. For industrial applications where a continuous process system is often required, carbon derived from RF gels can be more efficient and cost-effective than commercial adsorbents, which are in the form of granules or powders and are unsuitable for use in continuous systems.

The aim of this study is to synthesize an adsorbent with visible light driven photocatalytic activity by incorporating TiO<sup>2</sup> nanoparticles into RF gels. A typical synthesis route of an RF gel [15] is modulated in this study to integrate TiO<sup>2</sup> nanoparticles by formulating a twostep synthesis scheme. In addition to enhancement in textural properties of this newly synthesised adsorbent, improvement in photocatalytic properties is expected by (i) modification in electronic structure of TiO2, due to the presence of RF gel as a carbon source, shifting the absorption edge to the visible light region, hence, enabling TiO<sup>2</sup> to activate under visible light irradiation; (ii) the carbon phase can entrap the photogenerated electron and hole pairs, which would otherwise recombine and dissipate heat energy; and (iii) the porous RF gel helps facilitate dispersion of TiO<sup>2</sup> and easy post treatment removal of the adsorbent/photocatalyst.

Here, we report a study of the textural and optical characteristics of the adsorbent/photocatalyst. Detailed adsorption experiments were carried out to study the effect of several parameters on adsorption capacity. Additionally, the interaction behaviour between potential pollutants and the material were investigated, using methylene blue (MB) as a model adsorptive. Equilibrium sorption data were modelled using Langmuir, Freundlich, Sips, and Toth isotherm models. Kinetic analyses were carried out by comparing the experimental data with pseudo first order and pseudo second order expressions, as well as a diffusion model to better understand the transfer behaviour of the adsorbate species. Further, photocatalytic application tests were performed under visible light irradiation and the data were modelled to study the kinetics of photocatalysis.

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

#### *2.1. Morphology*

The morphology of sample, studied using FESEM, is shown in Figure 1. Figure 1a shows a heterogenous nature of synthesised RF/TiO<sup>2</sup> with homogenously distributed TiO2, as represented in Figure 1b. The overall structure shows the nanospheres connected to form a three-dimensional porous network, as represented in Figure 1c [14]. The heterogenous surface is more evident in Figure 1d where organic and inorganic phases can be differentiated. The diameter of microspheres ranged around 0.76–1.66 µm, indicating that the size of the primary particles was slightly larger than pristine RF, which generally is in the nanometre range [16]. Energy dispersive X-ray (EDX) spectra of the microspheres is shown in Figure 1e, (EDX zone shown in supplementary information, Figure S1) which evidently corresponds to the recorded spectra.

#### *2.2. FTIR Analysis*

The IR absorption bands of RF/TiO<sup>2</sup> overall resembled those of the pristine RF gel, as also observed through FESEM images with clear uniform spheres illustrating a porous network and the retention of the gel structure even after addition of TiO2. Typical characteristic peaks, such as the previously reported C=C stretching, CH2, and C-O-C of aromatic rings, methylene bridges, and methylene ether bridges [17,18], were observed. The broad peak at 3300 cm−<sup>1</sup> is characteristic of stretching vibrations associated with phenolic OH groups. Weak vibrations in the range of 2000–1700 cm−<sup>1</sup> are attributed to CH bending of

aromatic compounds. The absorption bands at 1605 and 1473 cm−<sup>1</sup> correspond to aromatic ether bridges, attributed to condensation of resorcinol to form the RF gel network. A strong IR peak, expected in the range 1740–1700 cm−<sup>1</sup> , associated with C=O stretching of aldehyde, was not observed, which confirms that the sol–gel reaction was complete. In comparison to a spectrum of pristine RF, a few additional peaks were observed that verify the chemical linkages between RF and TiO2, as marked in Figure 2. It has been established that the oxygenated surface groups of carbon materials support the attachment of TiO2. [19]. Here, crosslinking of TiO<sup>2</sup> with RF, via the hydroxyl groups, can be observed through the peaks in the vicinity of 1400 cm−<sup>1</sup> , attributed to OH groups of RF, which appeared weak in the spectrum of RF/TiO2, signifying the reaction of OH and TiO2. Meanwhile, new signals observed at 1200 and 1084 cm−<sup>1</sup> suggest formation of Ti-O-C functionalities. Similar crosslinking has previously been reported in TiO2/phenol resol hybrid structures, where chemical interactions between TiO<sup>2</sup> and phenol resol form Ti-O-C complexes. This heterojunction is responsive to visible light due to formation of a charge complex between the interface of TiO<sup>2</sup> and mesoporous phenol resol producing new electronic interactions [20]. Hence, it can be concluded that the interactions between RF and TiO<sup>2</sup> are chemical in nature. Additional signals below 1000 cm−<sup>1</sup> , such as bands at 963 and 880 cm−<sup>1</sup> , are associated with titanium ethoxide functional groups. Additionally, the broad band observed in the range of 600 cm−<sup>1</sup> corresponds to the vibration of Ti-O-Ti bonds [21].

#### *2.3. Surface Area Analysis*

A nitrogen sorption isotherm was measured to determine the specific surface area and pore volume of RF/TiO2. Figure 3 shows N<sup>2</sup> sorption isotherm and pore size distribution (inset Figure 3). As can be seen, the isotherm of RF/TiO<sup>2</sup> is of Type IV classification [22] with a sharp capillary condensation at P/P<sup>o</sup> = 0.4–0.9 and a well-defined hysteresis loop of Type H1, associated with open ended pores whilst suggesting a mesoporous structure [16]. Pore filling occurs at low relative pressure and the calculated mesoporosity in the structure was ~94%. The SBET, corresponding pore size and total pore volume of as prepared RF/TiO<sup>2</sup> is 439 m<sup>2</sup> g −1 , 9.4 nm and 0.71 cm<sup>3</sup> g −1 , respectively. The SBET value of pristine RF gel obtained in this study is 588 m<sup>2</sup> g −1 . The reason in reduced SBET value for RF/TiO<sup>2</sup> is attributed to blockage of pores of RF gel matrix with inclusion of TiO<sup>2</sup> nanoparticles. Meanwhile, in comparison with pristine TiO2, the SBET value is significantly higher for the synthesised RF/TiO2. Additionally, noteworthy SBET value for pristine TiO<sup>2</sup> (i.e., 111 m<sup>2</sup> g −1 ) is obtained in this study, contrary to commercial P25 with SBET value of 57 m<sup>2</sup> g −1 .

#### *2.4. Effect of pH*

The influence of MB sorption was studied by varying the solution pH from 2–12 (25 mL, 100 mg L−<sup>1</sup> , 0.01 g of adsorbent). The adsorption capacities at different pH values are shown in Figure 4. The efficiency of uptake increases from 47.24 to 65.96 mg g−<sup>1</sup> when the pH increases from 2–5. Thereafter, a sharp increase in adsorption capacity is observed at pH ≥ 6. The variation in adsorption behaviour of MB on RF/TiO<sup>2</sup> can be explained by considering the structure of MB and evaluated point of zero charge (pzc). The pHpzc value for RF/TiO<sup>2</sup> is determined to be 7.2 (Figure S2).

RF/TiO<sup>2</sup> can be amphoteric having both positively and negatively charged surface sites in aqueous solution due to the varying amount and nature of surface oxygen [23]. At pH lower than the pHpzc, the surface of RF/TiO<sup>2</sup> is positively charged, which repels the cationic dye (MB), and resultant interactions are hindered in acidic media due to electrostatic repulsion between the competing H<sup>+</sup> ions on the surface of adsorbent and MB dye molecules. As the pH increases, the surface of RF/TiO<sup>2</sup> becomes deprotonated and the adsorption sites available for interaction with cationic species increase, therefore, increased adsorption capacity is observed. This suggests that the electrostatic forces of attraction between MB and the surface of RF/TiO<sup>2</sup> increases due to increased ion density and positive

charges on the surface. Further, the OH groups on the surface of RF/TiO<sup>2</sup> can also attract MB dye molecules under higher pH conditions. *Gels* **2022**, *8*, x FOR PEER REVIEW 3 of 19

**Figure 1.** Morphology of RF/TiO2 sample (**a**) FESEM image of RF/TiO2, (**b**) TiO2 distribution determined by EDX on the sample, (**c**) distinct appearance of micro/nanospheres, (**d**) isolated microsphere with differentiation between organic–inorganic phase, and (**e**) corresponding EDX spectra. **Figure 1.** Morphology of RF/TiO<sup>2</sup> sample (**a**) FESEM image of RF/TiO<sup>2</sup> , (**b**) TiO<sup>2</sup> distribution determined by EDX on the sample, (**c**) distinct appearance of micro/nanospheres, (**d**) isolated microsphere with differentiation between organic–inorganic phase, and (**e**) corresponding EDX spectra.

The IR absorption bands of RF/TiO2 overall resembled those of the pristine RF gel, as also observed through FESEM images with clear uniform spheres illustrating a porous

*2.2. FTIR Analysis* 

range of 600 cm−1 corresponds to the vibration of Ti-O-Ti bonds [21].

*Gels* **2022**, *8*, x FOR PEER REVIEW 4 of 19

network and the retention of the gel structure even after addition of TiO2. Typical characteristic peaks, such as the previously reported C=C stretching, CH2, and C-O-C of aromatic rings, methylene bridges, and methylene ether bridges [17,18], were observed. The broad peak at 3300 cm−1 is characteristic of stretching vibrations associated with phenolic OH groups. Weak vibrations in the range of 2000–1700 cm−1 are attributed to CH bending of aromatic compounds. The absorption bands at 1605 and 1473 cm−1 correspond to aromatic ether bridges, attributed to condensation of resorcinol to form the RF gel network. A strong IR peak, expected in the range 1740–1700 cm−1, associated with C=O stretching of aldehyde, was not observed, which confirms that the sol–gel reaction was complete. In comparison to a spectrum of pristine RF, a few additional peaks were observed that verify the chemical linkages between RF and TiO2, as marked in Figure 2. It has been established that the oxygenated surface groups of carbon materials support the attachment of TiO2. [19]. Here, crosslinking of TiO2 with RF, via the hydroxyl groups, can be observed through the peaks in the vicinity of 1400 cm−1, attributed to OH groups of RF, which appeared weak in the spectrum of RF/TiO2, signifying the reaction of OH and TiO2. Meanwhile, new signals observed at 1200 and 1084 cm−1 suggest formation of Ti-O-C functionalities. Similar crosslinking has previously been reported in TiO2/phenol resol hybrid structures, where chemical interactions between TiO2 and phenol resol form Ti-O-C complexes. This heterojunction is responsive to visible light due to formation of a charge complex between the interface of TiO2 and mesoporous phenol resol producing new electronic interactions [20]. Hence, it can be concluded that the interactions between RF and TiO2 are chemical in nature. Additional signals below 1000 cm−1, such as bands at 963 and 880 cm−1, are associated with titanium ethoxide functional groups. Additionally, the broad band observed in the

**Figure 2.** FTIR spectra of synthesised RF/TiO2 gel compared to pristine RF gel. **Figure 2.** FTIR spectra of synthesised RF/TiO<sup>2</sup> gel compared to pristine RF gel. m<sup>2</sup> g−1 .

**Figure 3.** Nitrogen sorption isotherms and pore size distribution of synthesised RF/TiO2 gel. **Figure 3.** Nitrogen sorption isotherms and pore size distribution of synthesised RF/TiO<sup>2</sup> gel.

*2.4. Effect of pH*  The influence of MB sorption was studied by varying the solution pH from 2–12 (25 mL, 100 mg L−1, 0.01 g of adsorbent). The adsorption capacities at different pH values are shown in Figure 4. The efficiency of uptake increases from 47.24 to 65.96 mg g−1 when the pH increases from 2–5. Thereafter, a sharp increase in adsorption capacity is observed at pH ≥ 6. The variation in adsorption behaviour of MB on RF/TiO2 can be explained by con-Overall, a good adsorption capacity for MB is observed at pH higher than the pHpzc due to an increased number of negative sites in the higher pH range. This is in good agreement with the fact that, due to the presence of COO− and OH- functional groups, MB dye adsorption is favoured at pH > pHpzc [24]. The same trend has been observed in previous studies with activated carbon and TiO<sup>2</sup> composites where reduced activity was observed at acidic pH and maximum activity was observed in the pH range 6–10 [25–27].

#### sidering the structure of MB and evaluated point of zero charge (pzc). The pHpzc value *2.5. Effect of Contact Time*

qe (mg g-1

)

for RF/TiO2 is determined to be 7.2 (Figure S2). 140 160 180 200 220 Figure 5 shows the effect of contact time on the amount of MB molecules adsorbed by RF/TiO<sup>2</sup> gel under different initial MB concentrations. As shown, the adsorption capacity increases with increase in initial concentration. The equilibrium adsorption capacity increases from 102 mg g−<sup>1</sup> to 207 mg g−<sup>1</sup> by increasing the initial concentration of MB from 50 mg L−<sup>1</sup> to 200 mg L−<sup>1</sup> . Initially, the adsorption capacity overall is rapid for timeframes up to 30 min. This trend is expected, due to the greater driving force of MB dye molecules and immediate availability of vacant adsorption sites, hence resulting

0 2 4 6 8 10 12

pH

**Figure 4.** Effect of pH on the adsorption of MB dye by RF/TiO2 gel.

in increased in frequency of collisions between MB dye molecules and the RF/TiO<sup>2</sup> gel. Additionally, mesoporosity throughout the RF/TiO<sup>2</sup> gel structure provides a high surface area for greater adsorption of MB molecules. It is noteworthy that at higher MB concentration the adsorption rate is greater and adsorption capacity attains equilibrium faster than at low concentration. The reason is attributed to immediate occupancy of available active sites by a large amount of adsorbate molecules. This rapid occurrence of sorption is due to the presence of mesoporosity within the RF/TiO<sup>2</sup> gel, which corresponds to a large portion of the adsorption sites. In this case, the mesoporous structure provides a large surface area to solution volume within the porous network of the adsorbent gel. Additionally, within the mesopores, MB dye molecules are confined to be in close proximity to the surface. Such observations have been reported in previous research, particularly for activated carbons [28]. Over time, saturation of active sites occurs, and adsorption becomes difficult on the fewer available active sites due to repulsive forces between the MB molecules and the RF/TiO<sup>2</sup> gel surface. Additionally, the blockage of pores and charge repulsion of MB dye species may decelerate the adsorption progress. Similar phenomena have been explained for porous TiO<sup>2</sup> and other carbon/TiO<sup>2</sup> porous composite materials, where it may have taken longer for the adsorbate to diffuse deeper in the fine pores [29]. Thereafter, the adsorption capacity increases gradually until 90 min, and equilibrium is attained for the entire concentration range. Thus, equilibrium time was considered as 90 min which was considered sufficient for removal of MB ions by RF/TiO<sup>2</sup> gel. Hence, the contact time was set to 90 min in the remaining experiments to ensure equilibrium was achieved. 0.0 0.2 0.4 0.6 0.8 1.0 0 100 200 300 10 20 30 40 50 60 70 80 90 100 0.0 0.5 dV/dlog(w) Pore Volume (cm³g-1Pore width (nm) Relative Pressure (p/p<sup>0</sup> ) Quantity Adsorbed (cm³g-1**Figure 3.** Nitrogen sorption isotherms and pore size distribution of synthesised RF/TiO2 gel. *2.4. Effect of pH*  The influence of MB sorption was studied by varying the solution pH from 2–12 (25 mL, 100 mg L−1, 0.01 g of adsorbent). The adsorption capacities at different pH values are shown in Figure 4. The efficiency of uptake increases from 47.24 to 65.96 mg g−1 when the pH increases from 2–5. Thereafter, a sharp increase in adsorption capacity is observed at pH ≥ 6. The variation in adsorption behaviour of MB on RF/TiO2 can be explained by considering the structure of MB and evaluated point of zero charge (pzc). The pHpzc value for RF/TiO2 is determined to be 7.2 (Figure S2).

*Gels* **2022**, *8*, x FOR PEER REVIEW 5 of 19

pristine RF gel obtained in this study is 588 m<sup>2</sup>

g−1, 9.4 nm and 0.71 cm<sup>3</sup>

RF/TiO2 is attributed to blockage of pores of RF gel matrix with inclusion of TiO2 nanoparticles. Meanwhile, in comparison with pristine TiO2, the SBET value is significantly higher for the synthesised RF/TiO2. Additionally, noteworthy SBET value for pristine TiO<sup>2</sup>

g−1) is obtained in this study, contrary to commercial P25 with SBET value of 57

 Adsorption Desorption

g−1, respectively. The SBET value of

g−1. The reason in reduced SBET value for

prepared RF/TiO2 is 439 m<sup>2</sup>

(i.e., 111 m<sup>2</sup>

400

1.0

1.5

)

2.0

500

)

600

m<sup>2</sup> g−1 .

**Figure 4.** Effect of pH on the adsorption of MB dye by RF/TiO2 gel. **Figure 4.** Effect of pH on the adsorption of MB dye by RF/TiO<sup>2</sup> gel.

#### *2.6. Effect of Sorbent Dose*

The percentage removal of MB dye increased with increase in the adsorbent dose from 0.005 to 0.01 g but remained almost constant with further increase in the dose range 0.01 to 0.1 g, as represented in Figure 6. Percentage removal was calculated using Equation (2), and showed an increase with increase in adsorbent dose, due to greater availability of vacant active sites, a large surface area, and a greater number of adsorptive sites present on the surface of RF/TiO2. With further increase in adsorbent dose (>0.01 g), the rate of MB removal becomes low, as the concentrations at the surface and solution reach equilibrium. The resultant reduction in adsorption rate is attributed to unoccupied adsorbent sites, as well as overcrowding or aggregation of adsorbent particles [30]. Hence, the surface area available for MB adsorption per unit mass of the adsorbent reduces, whereby percentage removal was not significantly enhanced with further increase adsorbent dose.

50

100

qe (mg g-1

)

150

200

*Gels* **2022**, *8*, x FOR PEER REVIEW 7 of 19

**Figure 5.** Effect of adsorption on contact time and initial concentration of MB dye by RF/TiO2 gel. **Figure 5.** Effect of adsorption on contact time and initial concentration of MB dye by RF/TiO<sup>2</sup> gel. removal was not significantly enhanced with further increase adsorbent dose.

 qt (50 mg/L) qt (100 mg/L) qt (150 mg/L) qt (200 mg/L)

70 **Figure 6. Figure 6.**Effect of adsorbent dose on the removal and adsorption of MB dye Effect of adsorbent dose on the removal and adsorption of MB dye by RF/TiO by RF/TiO2 gel. <sup>2</sup> gel.

#### 200 *2.7. Adsorption Kinetics*

100 150 qe (mg g-1 ) 55 60 65 % Removal The adsorption kinetics were studied using a contact time of 240 min in the concentration range 50–200 mg L−<sup>1</sup> . The experimental data obtained for MB dye adsorption capacity vs. time (t) were fitted with PFO and PSO, as presented in Figure 7a–d. The parameters determined, including measured equilibrium adsorption capacity qe (experimental), theoretical equilibrium adsorption capacity qe (calculated), first order rate constant K1, second order rate constant K2, and regression coefficient R<sup>2</sup> , are presented in Table 1.

0.00 0.02 0.04 0.06 0.08 0.10 0 50 Dose (g) 50 **Figure 6.** Effect of adsorbent dose on the removal and adsorption of MB dye by RF/TiO2 gel. As observed from the data, the correlation factor R<sup>2</sup> deviates significantly from 1 for PFO and, therefore, pseudo first order model does not exhibit good compliance with the experimental data for the entire concentration range. This implies that the adsorption reaction is not inclined towards physisorption, and the MB dye molecules adsorb to specific sites on the surface of RF/TiO<sup>2</sup> gel. The argument regarding the failure of the pseudo first order model suggests that several other interactions are responsible for the sorption mechanism. Hence, the correlation coefficients R<sup>2</sup> of the pseudo second order model were compared with pseudo first order parameters. R<sup>2</sup> values for pseudo second order behaviour are approximately 0.99 for the entire concentration range, indicating that the system is more appropriately described by the pseudo second order equation. The dependence on

initial concentration of MB dye is verified by good compliance of the experimental data with the pseudo second order equation, where the adsorption capacity is affected by the initial MB dye concentration, subsequent surface-active sites, and adsorption rate (Other error analyses are represented in Table S1). − − −

**− − − −**

−

−

− −

−

− − − − **Figure 7.** MB uptakes on RF/TiO<sup>2</sup> gel at (**a**) 50 mg L−<sup>1</sup> , (**b**) 100 mg L−<sup>1</sup> , (**c**) 150 mg L−<sup>1</sup> , (**d**) 200 mg L−<sup>1</sup> , and fitted data for pseudo first order and pseudo second order kinetic models.


**Table 1.** Kinetic parameters obtained by fitting kinetic data for MB adsorption to RF/TiO<sup>2</sup> .

The equilibrium sorption capacity increased from 116.97 to 217.59 mg g−<sup>1</sup> when initial dye concentration was increased from 50 to 200 mg g−<sup>1</sup> confirming that MB dye removal is dependent on initial concentration, where the rate limiting step is determined by both adsorbate (MB) and adsorbent (RF/TiO2) concentration. This signifies that the sorption mechanism is chemisorption. Previous studies have explained theoretically that if

diffusion is not the rate limiting factor, then higher adsorbate concentrations would give a good pseudo first order fit whereas, for low concentrations, pseudo second order better represents the kinetics of sorption, analogous to the observations made here [31]. Previously, the adsorption processes of MB on TiO2/carbon composites have also exhibited strong dependencies of pseudo second order fitting parameters on initial concentrations [27]. sion is not the rate limiting factor, then higher adsorbate concentrations would give a good pseudo first order fit whereas, for low concentrations, pseudo second order better represents the kinetics of sorption, analogous to the observations made here [31]. Previously, the adsorption processes of MB on TiO2/carbon composites have also exhibited strong dependencies of pseudo second order fitting parameters on initial concentrations [27].

PFO and, therefore, pseudo first order model does not exhibit good compliance with the experimental data for the entire concentration range. This implies that the adsorption reaction is not inclined towards physisorption, and the MB dye molecules adsorb to specific sites on the surface of RF/TiO2 gel. The argument regarding the failure of the pseudo first order model suggests that several other interactions are responsible for the sorption mech-

are approximately 0.99 for the entire concentration range, indicating that the system is more appropriately described by the pseudo second order equation. The dependence on initial concentration of MB dye is verified by good compliance of the experimental data with the pseudo second order equation, where the adsorption capacity is affected by the initial MB dye concentration, subsequent surface-active sites, and adsorption rate. (Other

The equilibrium sorption capacity increased from 116.97 to 217.59 mg g−1 when initial dye concentration was increased from 50 to 200 mg g−1 confirming that MB dye removal is dependent on initial concentration, where the rate limiting step is determined by both adsorbate (MB) and adsorbent (RF/TiO2) concentration. This signifies that the sorption mechanism is chemisorption. Previous studies have explained theoretically that if diffu-

deviates significantly from 1 for

of the pseudo second order model were com-

values for pseudo second order behaviour

*Gels* **2022**, *8*, x FOR PEER REVIEW 9 of 19

As observed from the data, the correlation factor R<sup>2</sup>

anism. Hence, the correlation coefficients R<sup>2</sup>

error analyses are represented in Table S1)

pared with pseudo first order parameters. R<sup>2</sup>

Figure 8 shows a plot of MB dye uptake (qe) on synthesised RF/TiO<sup>2</sup> against (time)0.5 . The plots exhibit multi-linearity, rather than two straight lines, indicating that the adsorption process is influenced by several steps. The initial segment of the plots shows that diffusion across the boundary of the adsorbent only lasts for a short time in comparison to the whole adsorption process. This second section is attributed to diffusion into the mesopores of the adsorbent, i.e., the MB dye molecules enter less accessible pore sites. Resultantly, the diffusion resistance increases, and the diffusion rate decreases. This stage is a slow and gradual stage of the adsorption process. The third segment represents the final equilibrium stage where intra-particle diffusion slows down to an extremely low rate due to the remaining concentration of the MB dye molecules in the solution. This implies a slow transport rate of MB dye molecules from the solution (through the gel–dye solution interface) to available sites. Here, the surface of the RF/TiO<sup>2</sup> gel, and micropores, may be responsible for the uptake of MB dye molecules. Figure 8 shows a plot of MB dye uptake (qe) on synthesised RF/TiO2 against (time)0.5 . The plots exhibit multi-linearity, rather than two straight lines, indicating that the adsorption process is influenced by several steps. The initial segment of the plots shows that diffusion across the boundary of the adsorbent only lasts for a short time in comparison to the whole adsorption process. This second section is attributed to diffusion into the mesopores of the adsorbent, i.e., the MB dye molecules enter less accessible pore sites. Resultantly, the diffusion resistance increases, and the diffusion rate decreases. This stage is a slow and gradual stage of the adsorption process. The third segment represents the final equilibrium stage where intra-particle diffusion slows down to an extremely low rate due to the remaining concentration of the MB dye molecules in the solution. This implies a slow transport rate of MB dye molecules from the solution (through the gel–dye solution interface) to available sites. Here, the surface of the RF/TiO2 gel, and micropores, may be responsible for the uptake of MB dye molecules.

**Figure 8. Figure 8.** Intra-particle diffusion kinetics of MB dye adsorption on RF/Ti Intra-particle diffusion kinetics of MB dye adsorption on RF/TiOO2. <sup>2</sup> .

#### *2.8. Adsorption Isotherms*

The equilibrium data were analysed using Langmuir, Freundlich, Sips, and Toth isotherm equations to obtain the best fit. The isotherm data plots, and fitting model parameters are shown in Figure 9 and Table 2, respectively. Comparison of the correlation factor R<sup>2</sup> indicates that qe,exp fitted well to the Sips model with the lowest χ <sup>2</sup> value. The qe,cal value, calculated using the Sips model, is closest to qe,exp with R<sup>2</sup> closest to 1. The Sips model is a combination of the Langmuir and Freundlich adsorption isotherms, hence, the model suggests both monolayer and multilayer adsorption. At low MB dye concentrations, the model predicts Freundlich adsorption isotherms as a heterogenous adsorption system and localised adsorption without adsorbate–adsorbate interactions, whereas at high concentrations the model predicts monolayer adsorption as in Langmuir isotherm [32,33]. In the present study, the value of constant ns from Equation (11), the heterogeneity factor, is greater than 1 (i.e., n<sup>s</sup> = 1.91), hence, the adsorption system is predicted to be heterogenous [33]. Further, the Toth isotherm model validates multilayer and heterogeneous adsorption, where the factor n<sup>T</sup> determines heterogeneity. Here, again the value of n<sup>T</sup> is greater than 1, and, therefore, the system confirms heterogeneity. It is evident that the equilibrium uptakes follow the Sips model according to the correlation factor R<sup>2</sup> (other error analyses are represented in Table S2) and the isotherm models fit the data in the order Sips > Toth > Langmuir > Freundlich.

Langmuir

Freundlich

Sips

Toth

order Sips > Toth > Langmuir > Freundlich.

*Gels* **2022**, *8*, x FOR PEER REVIEW 10 of 19

indicates that qe,exp fitted well to the Sips model with the lowest χ<sup>2</sup>

value, calculated using the Sips model, is closest to qe,exp with R<sup>2</sup>

The equilibrium data were analysed using Langmuir, Freundlich, Sips, and Toth isotherm equations to obtain the best fit. The isotherm data plots, and fitting model parameters are shown in Figure 9 and Table 2, respectively. Comparison of the correlation factor

model is a combination of the Langmuir and Freundlich adsorption isotherms, hence, the model suggests both monolayer and multilayer adsorption. At low MB dye concentrations, the model predicts Freundlich adsorption isotherms as a heterogenous adsorption system and localised adsorption without adsorbate–adsorbate interactions, whereas at high concentrations the model predicts monolayer adsorption as in Langmuir isotherm [32,33]. In the present study, the value of constant ns from Equation (11), the heterogeneity factor, is greater than 1 (i.e., ns = 1.91), hence, the adsorption system is predicted to be heterogenous [33]. Further, the Toth isotherm model validates multilayer and heterogeneous adsorption, where the factor nT determines heterogeneity. Here, again the value of nT is greater than 1, and, therefore, the system confirms heterogeneity. It is evident that the

equilibrium uptakes follow the Sips model according to the correlation factor R<sup>2</sup>

error analyses are represented in Table S2) and the isotherm models fit the data in the

value. The qe,cal

(other

closest to 1. The Sips

*2.8. Adsorption Isotherms* 

R2

**Figure 9.** Adsorption data for RF/TiO2 onto MB dye corresponding fits to Langmuir, Freundlich, Sips, and Toth equation. **Figure 9.** Adsorption data for RF/TiO<sup>2</sup> onto MB dye corresponding fits to Langmuir, Freundlich, Sips, and Toth equation.


**Table 2.** Isotherm parameters obtained by fitting MB adsorption data for RF/TiO2 to the Langmuir, Freundlich, Sips, and Toth equations*.*  **Table 2.** Isotherm parameters obtained by fitting MB adsorption data for RF/TiO<sup>2</sup> to the Langmuir, Freundlich, Sips, and Toth equations.

#### *2.9. Thermodynamic Study*

Thermodynamic parameters for the adsorption system are recorded in Table 3. Negative values of free energy changes are evident from the data, which signifies the spontaneous adsorption of MB dye molecules on the sample for the studied temperature range. Adsorption capacity increases with an increase in temperature and a positive ∆H<sup>0</sup> (Table 3) suggests that the adsorption is endothermic in nature. Positive ∆S 0 indicates some structural changes in the MB dye and RF/TiO<sup>2</sup> gel causing an increase in the degree of freedom of the MB dye species and consequently increased randomness at the adsorbent–adsorbate interface. At high temperature, the release of high-energy desolvated water molecules from the MB dye molecules and/or aggregates arise after adsorption on RF/TiO<sup>2</sup> gel, which relates to a positive ∆S 0 [34]. Before sorption begins, the MB ions are surrounded by highly ordered water clusters strongly bound via hydrogen bonding. Once MB ions come in close contact with the surface of RF/TiO2, the interaction results in agitation of the ordered water molecules, subsequently increasing the randomness of the system. Although, the adsorption of MB dye onto RF/TiO<sup>2</sup> gel may reduce the freedom of the system, the entropy increase in water molecules is much higher than the entropy decrease in MB ions. Therefore, the driving force for the adsorption of MB on RF/TiO<sup>2</sup> is controlled by an entropic effect rather than an enthalpic change. Similar phenomena have previously been reported

in order to explain the fact that thermodynamic parameters are not only related to the properties of the adsorbate but also to the properties of other solid particles [35,36].


**Table 3.** Thermodynamic data for MB adsorption onto RF/TiO<sup>2</sup> at various temperatures.

#### **3. Photocatalytic Tests**

Photocatalytic activity was determined by testing the efficiency of RF/TiO<sup>2</sup> against degradation of methylene blue (MB) under visible light irradiation. The maximum absorbance vs. wavelength spectra (in the range of 550–700 nm) were collected and subsequent activity, after 30 min, intervals was recorded, as shown in Figure 10. *Gels* **2022**, *8*, x FOR PEER REVIEW 12 of 19

**Figure 10.** UV-Vis spectra of MB dye degradation using RF/TiO2 gel. **Figure 10.** UV-Vis spectra of MB dye degradation using RF/TiO<sup>2</sup> gel.

The dye degradation data obtained after treatment with RF/TiO2 showed ~73% MB dye removal after 90 min. This is attributed to the synergy of RF and TiO2, enabling an absorption shift to a higher wavelength, as λmax is detected at 410 nm (Figure 11). Further analysis indicates modification in the electronic structure and a subsequent reduction in bandgap occurs due to doping of TiO2 similar to when combined with carbon [39]. The calculated band gap energy is 2.97 eV, as shown in Figure 11 (inset). The value achieved is significantly lower than pristine TiO2 (i.e., 3.2 eV [21]), indicating photodegradation of MB dye under visible light irradiation. The RF matrix enables entrapment of a photogenerated electron and hole pairs and, therefore, rapid generation of ROS is possible for efficient degradation of the MB dye. These findings are comparable to other carbon/TiO2 systems where synergistic effects have substantially enhanced the performance of the system due to improved optical properties of the material [1,40,41]. Within the studied systems, no photodegradation activity (reduction in concentration and decolourisation of MB dye) was observed in the absence of adsorbent/catalyst, as well as in the presence of pristine RF, indicating that the properties of MB are more stable. Additionally, RF solely may not be recommended for photocatalysis due to slow charge transfer properties, which has also been proven by the study carried out by Zang, Ni, and Liu, where the researchers employed pristine RF resins for visible light photocatalysis [37]. Slight photodegradation is observed in the presence of pristine TiO2, which may be attributed to the potential absorbance of UV-Vis light from the surroundings confirming that the process of MB degradation is light driven. Although the TiO<sup>2</sup> obtained for use in this study has a high surface area, which may possess good adsorption properties to exhibit efficient adsorption of MB dye, since TiO<sup>2</sup> only activates upon UV light irradiation (~280 nm), it does not produce enough reactive oxide species (ROS) to be an effective photodegradation system [38].

(eVcm)1/2 Absorbance (a.u.)max =410 nm The dye degradation data obtained after treatment with RF/TiO<sup>2</sup> showed ~73% MB dye removal after 90 min. This is attributed to the synergy of RF and TiO2, enabling an absorption shift to a higher wavelength, as λmax is detected at 410 nm (Figure 11). Further analysis indicates modification in the electronic structure and a subsequent reduction in bandgap occurs due to doping of TiO<sup>2</sup> similar to when combined with carbon [39]. The calculated band gap energy is 2.97 eV, as shown in Figure 11 (inset). The value achieved is

Eg =2.97

0 1 2 3 4 5 6 7 8 9 10

h

**Figure 11.** Absorption vs. wavelength spectrum of RF/TiO2 dispersed in ethanol measured through

The photodegradation of MB dye, that is, the reduction in concentration with time is recorded in Figure 12 and the recorded data is modelled using pseudo first order kinetics, shown in Figure 12a,b. The data are fitted to the first order kinetic equation (ln (Co/Ce) =

(h)

400 450 500 550 600 650 700 750 800

Wavelength (nm)

UV-Vis spectrophotometer, inset shows calculated band gap of synthesised RF/TiO2.

Absorbance (a.u.)

 0 min 30 min 60 min 90 min

significantly lower than pristine TiO<sup>2</sup> (i.e., 3.2 eV [21]), indicating photodegradation of MB dye under visible light irradiation. The RF matrix enables entrapment of a photogenerated electron and hole pairs and, therefore, rapid generation of ROS is possible for efficient degradation of the MB dye. These findings are comparable to other carbon/TiO<sup>2</sup> systems where synergistic effects have substantially enhanced the performance of the system due to improved optical properties of the material [1,40,41]. is significantly lower than pristine TiO2 (i.e., 3.2 eV [21]), indicating photodegradation of MB dye under visible light irradiation. The RF matrix enables entrapment of a photogenerated electron and hole pairs and, therefore, rapid generation of ROS is possible for efficient degradation of the MB dye. These findings are comparable to other carbon/TiO2 systems where synergistic effects have substantially enhanced the performance of the system due to improved optical properties of the material [1,40,41].

The dye degradation data obtained after treatment with RF/TiO2 showed ~73% MB dye removal after 90 min. This is attributed to the synergy of RF and TiO2, enabling an absorption shift to a higher wavelength, as λmax is detected at 410 nm (Figure 11). Further analysis indicates modification in the electronic structure and a subsequent reduction in bandgap occurs due to doping of TiO2 similar to when combined with carbon [39]. The calculated band gap energy is 2.97 eV, as shown in Figure 11 (inset). The value achieved

*Gels* **2022**, *8*, x FOR PEER REVIEW 12 of 19

500 600 700 800

Wavelength (nm)

**Figure 10.** UV-Vis spectra of MB dye degradation using RF/TiO2 gel.

**Figure 11.** Absorption vs. wavelength spectrum of RF/TiO2 dispersed in ethanol measured through UV-Vis spectrophotometer, inset shows calculated band gap of synthesised RF/TiO2. **Figure 11.** Absorption vs. wavelength spectrum of RF/TiO<sup>2</sup> dispersed in ethanol measured through UV-Vis spectrophotometer, inset shows calculated band gap of synthesised RF/TiO<sup>2</sup> .

The photodegradation of MB dye, that is, the reduction in concentration with time is recorded in Figure 12 and the recorded data is modelled using pseudo first order kinetics, shown in Figure 12a,b. The data are fitted to the first order kinetic equation (ln (Co/Ce) = The photodegradation of MB dye, that is, the reduction in concentration with time is recorded in Figure 12 and the recorded data is modelled using pseudo first order kinetics, shown in Figure 12a,b. The data are fitted to the first order kinetic equation (ln (Co/Ce) = kt) to evaluate the value of the rate constant by slope of plot ln(Co/Ce) vs. time (t) in minutes, where C<sup>o</sup> and Ce is the initial at t = 0 and final concentration at given time of MB concentration, respectively. The value of k here is the measure of photocatalytic performance, as it defines the concentration of reacting substances, that is, photogenerated reactive oxide species, therefore, a higher value of k signifies higher photocatalytic efficiency. As compared to no catalyst (k = 2.43 <sup>×</sup> <sup>10</sup>−<sup>6</sup> min−<sup>1</sup> ) pristine TiO<sup>2</sup> (k = 1.74 <sup>×</sup> <sup>10</sup>−<sup>3</sup> min−<sup>1</sup> ) and pristine RF (k = 6.89 <sup>×</sup> <sup>10</sup>−<sup>4</sup> min−<sup>1</sup> ), the rate of RF/TiO<sup>2</sup> was the highest (k = 1.27 <sup>×</sup> <sup>10</sup>−<sup>2</sup> min−<sup>1</sup> ). Clearly, the rate constant obtained for photodegradation of MB using RF/TiO<sup>2</sup> was the highest. Mainly, improved optical property was the most important advancement in forming RF/TiO<sup>2</sup> gel which is photocatalytically active under visible light (410 nm) irradiation.

The RF/TiO<sup>2</sup> material created in this study exhibits excellent photoactivity under visible light, which can further be explained by the mechanism of MB photodegradation represented in Equations (1)–(8). The system activates when RF/TiO<sup>2</sup> absorbs light with photon energy (hν) and generates conduction band (CB) electron (e−) and valence band (VB) hole (h<sup>+</sup> ) pairs upon under visible light irradiation. The holes interact with moisture on the surface of the adsorbent gel yielding hydroxy free radicals or reactive oxide species (H<sup>+</sup> or OH• ), which are oxidation agents that can mineralise a wide range of organic pollutants, ultimately producing CO<sup>2</sup> and H2O as end products. The reaction sequence below represents the photodegradation of MB, showing a simplified mechanism of photoactivation by a photocatalyst (Equations (1)–(4)) [19]. For the mechanism of photoinactivation of MB in the presence of RF/TiO2, hydroxy free radicals or reactive oxide species (H<sup>+</sup> or OH• ) attack the aromatic ring of the MB structure, degrading it into a single ring structure product, which then finally degrades to CO<sup>2</sup> and H2O (Equations (5)–(8)) [42,43].

*Gels* **2022**, *8*, x FOR PEER REVIEW 13 of 19

**Figure 12.** (**a**) Photocatalytic performance regarding MB dye degradation without catalyst, and with pristine RF, pristine TiO2 and RF/TiO2 gel (**b**) First-order kinetics of photoactivity without catalyst, and with pristine RF, pristine TiO2, and RF/TiO2 gel*.*  **Figure 12.** (**a**) Photocatalytic performance regarding MB dye degradation without catalyst, and with pristine RF, pristine TiO<sup>2</sup> and RF/TiO<sup>2</sup> gel (**b**) First-order kinetics of photoactivity without catalyst, and with pristine RF, pristine TiO<sup>2</sup> , and RF/TiO<sup>2</sup> gel.

kt) to evaluate the value of the rate constant by slope of plot ln(Co/Ce) vs. time (t) in minutes, where Co and Ce is the initial at t = 0 and final concentration at given time of MB concentration, respectively. The value of k here is the measure of photocatalytic performance, as it defines the concentration of reacting substances, that is, photogenerated reactive oxide species, therefore, a higher value of k signifies higher photocatalytic efficiency. As compared to no catalyst (k = 2.43 × 10−6 min−1) pristine TiO2 (k = 1.74 × 10−3 min−1) and pristine RF (k = 6.89 × 10−4 min−1), the rate of RF/TiO2 was the highest (k = 1.27 × 10−2 min−1). Clearly, the rate constant obtained for photodegradation of MB using RF/TiO2 was the highest. Mainly, improved optical property was the most important advancement in forming RF/TiO2 gel which is photocatalytically active under visible light (410 nm) irradi-

$$\text{RF/TiO}\_2 + \text{hv} = \text{e}\_{\text{CB}}^- + \text{h}\_{\text{vB}}^+ \tag{1}$$

$$\rm{h}\_{\rm{CB}}^{-} + \rm{H}\_{2}\rm{O} = \rm{H}^{+} + \rm{OH}^{\cdot} \tag{2}$$

$$\mathbf{e}\_{\mathbf{CB}}^{-} + \mathbf{O}\_{2} = \mathbf{O}\_{2}^{-} \tag{3}$$

$$\rm O\_2^{-} + H^{+} = HO\_2^{-} \tag{4}$$

$$\text{MB} + \text{RF/TiO}\_2 = \text{MB}^+ + \text{e}\_{\text{CB}}^-(\text{RF/TiO}\_2) \tag{5}$$

$$\text{O}\_2 + \text{e}^- = \text{O}\_2^- \tag{6}$$

$$\mathrm{MB^{+}} + \mathrm{OH^{-}} = \mathrm{MB} + \mathrm{OH^{-}} \tag{7}$$

$$\text{MB}^{\cdot+} + \text{OH}^{\cdot} = \text{H}\_2\text{O} + \text{CO}\_2 + \text{other products} \tag{8}$$

#### uct, which then finally degrades to CO2 and H2O (Equations (5)–(8)) [42,43]. **4. Conclusions**

(h<sup>+</sup>

ation.

RF/TiO<sup>ଶ</sup> + hν = eେ ି + h୴ ା (1)

hେ ି + HଶO = Hା + OH˙ (2) eେ ି + O<sup>ଶ</sup> = O<sup>ଶ</sup> .ି (3) Oଶ .ି + Hା = HO<sup>ଶ</sup> . (4) MB + RF/TiO<sup>ଶ</sup> = MB.ା + eେ ି (RF/TiOଶ) (5) O<sup>ଶ</sup> + eି = O<sup>ଶ</sup> .ି (6) MB.ା + OHି = MB + OH˙ (7) An RF/TiO<sup>2</sup> gel was successfully synthesised using sol–gel techniques via a straightforward route. The synergy of RF and TiO<sup>2</sup> exhibited excellent adsorption–photodegradation activity due to the corresponding characteristics, mainly mesoporosity and photocatalysis. The synergy of contributing materials allowed modification in the electronic structure of TiO<sup>2</sup> by formation of Ti-O-C chemical linkages, responsible for a reduction in the band gap of TiO<sup>2</sup> for photodegradation upon visible light irradiation. Kinetic studies revealed a pseudo second order reaction, signifying chemisorption phenomenon is involved in the adsorption mechanism. The adsorption isotherm study showed that the system was heterogeneous following the Sips model equation. The spontaneity of the process was validated via thermodynamic studies, which signified an entropically driven adsorption mechanism. Effective photodegradation results were observed due to the high adsorption capacity and improved optical properties of the material, enabling significant MB dye degradation within 90 min. Overall, the material possesses properties that have potential to effectively reduce/eliminate a wide range of pollutants and, therefore, can be employed as a low-cost photocatalytic adsorbent for water treatment Especially in the industrial applications where post treatment separation and recovery of the adsorbent is difficult, employing this material can reduce the costs since in this case the adsorbent precipitates and easy separation is possible just by filtration or even decantation.

#### **5. Material and Methods**

#### *5.1. Synthesis*

Synthesis of RF and TiO<sup>2</sup> precursors was carried out in two separate systems, which were integrated and processed further in order to obtain the final gel.

System 1: Preparation of Titania Sol

For preparation of the titania sol, 1.78 g of titanium precursor: titanium isopropoxide (TTIP) (98+%, ACROS Organics™, Geel, Belgium) was dissolved in ethanol and stirred for 30 min. A mixture of water and HCl was added dropwise to the titania/EtOH solution under constant stirring, at room temperature, to begin hydrolysis. After 2 h, a homogenous solution was obtained. The molar ratios for these parameters were 1 TTIP:10 EtOH:0.3 HCl:0.1 H2O.

System 2: Preparation of RF sol

In total, 7.74 g of resorcinol (SigmaAldrich, ReagentPlus, 99%, Poole, UK) was added to 50 mL of deionised water until completely dissolved. 0.0249 g of sodium carbonate (Na2CO3, Sigma-Aldrich, anhydrous, ≥99.5%), as a catalyst, and 4.23 g of formaldehyde (37wt%) were added to the dissolved resorcinol under continuous stirring, at room temperature.

Finally, the prepared titania sol (system 1) was gradually transferred to the RF sol (from system 2) under constant stirring, at room temperature. The resulting sol was stirred at room temperature for 2 h after which the sol mixture was aged at 85 ◦C for 72 h.

After aging, the process of solvent exchange and drying the RF/TiO<sup>2</sup> gel, first involved cutting the gel into smaller pieces. These pieces were then immersed in acetone for 72 h to facilitate solvent exchange prior to drying, followed by vacuum drying at 110 ◦C for 48 h to obtain the final RF/TiO<sup>2</sup> adsorbent gel. In this way, the final gel corresponded to 10 wt% TiO<sup>2</sup> (theoretical percentage) incorporated in the RF gel matrix.

#### *5.2. Adsorbent Characterisation*

Morphology of the synthesised sample was studied by field emission electron scanning microscope (FESEM) TESCAN-MIRA. The functional groups on the surface of synthesised RF/TiO2, and the chemical linkages between the constituent RF and TiO<sup>2</sup> components, were verified using Fourier Transform Infrared Spectroscopy (FTIR) (MB3000 series, scan range 4000–400 nm). BET surface area measurements were carried out using a Micromeritics ASAP 2420 to obtain N<sup>2</sup> adsorption isotherm at 77 K and pore size was determined via BJH theory [22]. A UV-Vis Spectrophotometer (Varian Cary 5000 UV-Vis NIR Spectrophotometer Hellma Analytics) was used to collect absorption spectra and the data used to interpret the change in electronic structure of RF/TiO<sup>2</sup> [44]. The data were manipulated to calculate the band gap energy values through the Tauc method described in previous studies [44].

#### **6. Adsorption Experiments**

#### *6.1. Effect of pH*

The effect of pH on the sorption of methylene blue (MB) dye was investigated with 0.01 g of sample by adjusting the pH of solution (25 mL, 100 mg L−<sup>1</sup> MB) between 2 and 12, at 23 ◦C. The pH was adjusted using 0.01 M HCl and 0.01 M NaOH. After 2 h of agitation, the solution was centrifuged for 15 min and the supernatant solution was collected via syringe. The initial and final concentrations were measured using a UV-Vis spectrophotometer (Varian Cary 5000 UV-Vis NIR Spectrophotometer, Agilent, UK) and onward calculations were performed.

#### *6.2. Effect of Sorbent Dose*

The amount of sorbent dose was gradually increased from 0.005 to 0.01 g to study the effect of sorbent dose on the adsorption capacity. pH and temperature were maintained at 7.0 and 23 ◦C, respectively, against 25 mL of 100 mg L−<sup>1</sup> MB concentrated solution. The pH was adjusted using 0.01 M HCl and 0.01 M NaOH. After 2 h of agitation, the solution was centrifuged for 15 min and the supernatant solution was collected via syringe. The initial and final concentrations were measured using a UV-Vis spectrophotometer (Varian

Cary 5000 UV-Vis NIR Spectrophotometer Hellma Analytics) and onward calculations were performed.

#### *6.3. Effect of Initial Concentration*

All adsorption experiments were performed at 23 ◦C in 125 mL conical flasks, using a shaker (VWR 3500 Analog orbital shaker) set to 125 rpm. The first experiment was conducted to study the isothermal equilibrium and the effect of initial MB concentration. Standard solutions of MB were prepared using distilled water, with initial concentrations in the range of 20–200 mg L <sup>−</sup><sup>1</sup> . Then, 25 mL aliquots were distributed into each flask, and 0.01 g of the adsorbent gel was added individually to each flask. The pH values of all solutions were recorded and adjusted to 7.0, if required, using 1 M HCl and 1 M NaOH. After 2 h of agitation, the solution was centrifuged for 15 min and the supernatant solution was collected via syringe. The initial and final concentrations were measured using UV-Vis spectrophotometer (Varian Cary 5000 UV-Vis NIR Spectrophotometer Hellma Analytics).

The equilibrium adsorption capacity, q<sup>e</sup> (mg g−<sup>1</sup> ), was calculated using:

$$\mathbf{q}\_{\mathbf{e}} = \frac{(\mathbf{C}\_{\mathbf{o}} - \mathbf{C}\_{\mathbf{e}}) \times \mathbf{V}(\mathbf{l})}{\mathbf{W}} \tag{9}$$

while the respective percentage removal of MB was calculated as:

$$\text{Removal }\%= \frac{\text{C}\_{\text{o}} - \text{C}\_{\text{e}}}{\text{C}\_{\text{o}}} \times 100\% \tag{10}$$

where C<sup>o</sup> and C<sup>e</sup> are the initial MB and final concentration, respectively. W is the weight (g) of the adsorbent and V is the volume (L) of MB solution.

#### *6.4. Effect of Contact Time*

The effect of contact time was studied by adding MB solution (pH 7.0, 25 mL, 100 mg L−<sup>1</sup> ) and 0.01 g adsorbent gel into a flask, which was then agitated for a predetermined contact time between 5 min and 4 h. The samples were prepared and treated as described in Section 2.5 and the amount of adsorption was calculated using Equation (11):

$$\mathbf{q}\_{\rm t} = \frac{(\mathbf{C}\_{\rm o} - \mathbf{C}\_{\rm t}) \times \mathbf{V}}{\mathbf{W}} \tag{11}$$

where C<sup>t</sup> is the equilibrium MB concentration at a given time, and Co, V, and W are as previously defined. Equilibrium concentration was determined by plotting q<sup>t</sup> versus time of aliquots collected at different time intervals. Adsorption-photodegradation (absorption) changes of MB dye with time were also recorded via UV-Vis spectrophotometry.

#### *6.5. Effect of Temperature*

The effect of temperature on the removal of MB (pH 7.0, 25 mL, 20–200 mg L−<sup>1</sup> ) was investigated by adding a known concentration MB solution and 0.01 g adsorbent gel to a flask. A hot plate with a stirrer (120 rpm) was used to maintain a constant temperature of 8, 23, 32, and 40 ◦C, under stirring, for 120 min after which the absorbance versus wavelength spectra were recorded to measure the final concentration, and subsequent adsorption was calculated using Equation (9).

#### *6.6. Kinetic Models*

The kinetic-based models: pseudo first order (PFO) and pseudo second order (PSO) were applied to study the adsorption kinetics and to explain the mode of sorption of MB onto the synthesised RF/TiO2. The PFO model [33] has been frequently used to describe kinetic processes under non-equilibrium conditions. PFO is based on the assumption that the rate of adsorption is proportional to the driving force, that is, the difference between

the equilibrium concentration and solid phase concentration, presented as a differential Equation (12):

$$\frac{d\mathbf{q}\_t}{dt} = \mathbf{k}\_1(\mathbf{q}\_e - \mathbf{q}\_t) \tag{12}$$

Integrating Equation (13) with the initial condition of q<sup>t</sup> = 0 at t = 0, the PFO model can be rewritten, in a linear form, as:

$$\mathbf{q}\_{\rm t} = \mathbf{q}\_{\rm e} \left( 1 - \mathbf{e}^{-\mathbf{k}\_1 \mathbf{t}} \right) \tag{13}$$

Several studies have also reported the use of PSO [45] to interpret data obtained for the sorption of contaminants from water, including dyes, organic molecules, and metal ions. The PSO model assumes that the overall adsorption rate is proportional to the square of the driving force and can be expressed as Equation (14):

$$\frac{\mathbf{dq\_t}}{\mathbf{dt}} = \mathbf{k\_l} (\mathbf{q\_e} - \mathbf{q\_t})^2 \tag{14}$$

Integrating Equation (14) with the initial condition of q<sup>t</sup> = 0 at t = 0, and q<sup>t</sup> = t at t = t, the PSO model can be rewritten as:

$$\mathbf{q}\_{\rm t} = \frac{\mathbf{k}\_2 \mathbf{t} \mathbf{q}\_{\rm e}^2}{1 + \mathbf{k}\_2 \mathbf{t} \mathbf{q}\_{\rm e}} \tag{15}$$

In Equations (13)–(15), qt (mg g−<sup>1</sup> ) and q<sup>e</sup> (mg g−<sup>1</sup> ) are the adsorption capacities of MB dye molecules at time t and at equilibrium, respectively. k<sup>1</sup> (mg g−<sup>1</sup> min−<sup>1</sup> ) and k<sup>2</sup> (mg g−<sup>1</sup> min−<sup>1</sup> ) are the PFO and PSO rate constants, respectively.

#### *6.7. Sorption Isotherm Models*

The equilibrium data for the sorption of MB on RF/TiO<sup>2</sup> adsorbent gel as a function of equilibrium concentration (C<sup>e</sup> mg L−<sup>1</sup> ) was analysed in terms of Langmuir, Freundlich, Sips, and Toth isotherm models [2]. The nonlinear form of Langmuir's isotherm model is represented as:

$$\mathbf{q}\_{\text{e}} = \frac{\mathbf{q}\_{\text{m}} \mathbf{K}\_{\text{L}} \mathbf{C}\_{\text{e}}}{1 + \mathbf{C}\_{\text{e}} \mathbf{K}\_{\text{L}}} \tag{16}$$

where q<sup>e</sup> (mg g−<sup>1</sup> ) is the MB uptake at equilibrium, C<sup>e</sup> (mg L−<sup>1</sup> ) is the equilibrium concentration, q<sup>m</sup> (mg g−<sup>1</sup> ) is the amount of adsorbate at complete monolayer coverage, and K<sup>L</sup> is the Langmuir constant.

The Freundlich equation can be expressed as follows:

$$\mathbf{q}\_{\mathbf{e}} = \mathbf{K}\_{\mathbf{F}} \mathbf{C}\_{\mathbf{e}}^{1/\mathbf{n}} \tag{17}$$

where q<sup>e</sup> and C<sup>e</sup> are as defined in the Langmuir equation, adsorption affinity is related to the adsorption constant KF, and n indicates the magnitude of the adsorption driving force and the distribution of energy sites on the adsorbent surface, if n < 1, then adsorption is a chemical process, whereas if n > 1, then adsorption maybe dependent on distribution of the surface sites. Generally, n values within 1–10 represents good adsorption [46].

The Sips isotherm model is a combination of the Langmuir and Freundlich isotherms and is represented as:

$$\mathbf{q}\_{\rm e} = \frac{\mathbf{q}\_{\rm s} \mathbf{K}\_{\rm s} \mathbf{C}\_{\rm e}^{\rm ns}}{1 + \mathbf{K}\_{\rm s} \mathbf{C}\_{\rm e}^{\rm ns}} \tag{18}$$

where q<sup>e</sup> and C<sup>e</sup> are as defined for Equation (16), K<sup>s</sup> is the Sips isotherm model constant (L g−<sup>1</sup> ), and ns; is the Sips isotherm exponent.

The Toth model also describes heterogeneous systems, considering both low- and high-end concentrations. The Toth expression is as follows:

$$\mathbf{q}\_{\mathbf{e}} = \frac{\mathbf{q}\_{\mathrm{m}} \mathbf{K}\_{\mathrm{T}} \mathbf{C}\_{\mathrm{e}}}{\left[1 + \left(\mathbf{K}\_{\mathrm{T}} \mathbf{C}\_{\mathrm{e}}\right)^{\mathrm{t}}\right]^{1/\mathrm{t}}} \tag{19}$$

where q<sup>e</sup> and C<sup>e</sup> are as defined for Equation (16), q<sup>m</sup> is the maximum adsorption capacity, t is the surface heterogeneity, and K<sup>T</sup> is the surface affinity.

#### *6.8. Photodegradation Procedure*

Photocatalytic performance of as prepared RF/TiO<sup>2</sup> was investigated through MB dye degradation, by recording the dye degradation spectra with time using UV-Vis Spectrophotometry. 0.01 g of the adsorbent dose were used against 25 mL of 100 mg L−<sup>1</sup> dye concentration at pH ~7 and a temperature of 23 ◦C and light intensity of 111 Wm−<sup>2</sup> . For comparison, the measurements were also recorded in the absence of catalyst, as well as pristine RF and TiO2. All suspensions were stirred in the dark for 60 min to establish sorption equilibrium before exposure to visible light.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/gels8040215/s1, Figure S1: Zone of Energy dispersive x-ray (EDX) spectra; Figure S2: Point of zero charge (pHpzc) on the surface of RF/TiO<sup>2</sup> ; Table S1: Kinetic parameters obtained by fitting kinetic data for MB adsorption to RF/TiO<sup>2</sup> .; Table S2: Isotherm parameters obtained by fitting MB adsorption data for RF/TiO<sup>2</sup> to the Langmuir, Freundlich, SIPS and Toth equations.

**Author Contributions:** Methodology, A.S.; formal analysis, A.S. and A.J.F.; resources, A.J.F.; writing original draft preparation, A.S.; writing—review and editing, A.J.F.; supervision, A.J.F.; project administration, A.J.F.; funding acquisition, A.J.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** Anam Safri thanks Ashleigh Fletcher and the Chemical and Process Engineering Department at the University of Strathclyde for funding this work. The authors gratefully acknowledge the Materials Science and Engineering Department at Institute of Space Technology, Islamabad, for providing support and facilities to conduct the morphological analysis.

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

#### **References**


## *Article* **Luminescent Hydrogel Based on Silver Nanocluster/Malic Acid and Its Composite Film for Highly Sensitive Detection of Fe 3+**

**Xiangkai Liu 1,2, Chunhui Li 1,2 , Zhi Wang 1 , Na Zhang 1, \*, Ning Feng 1 , Wenjuan Wang <sup>1</sup> and Xia Xin 1, \***


**Abstract:** Metal nanoclusters (NCs) with excellent photoluminescence properties are an emerging functional material that have rich physical and chemical properties and broad application prospects. However, it is a challenging problem to construct such materials into complex ordered aggregates and cause aggregation-induced emission (AIE). In this article, we use the supramolecular self-assembly strategy to regulate a water-soluble, atomically precise Ag NCs (NH<sup>4</sup> )9 [Ag<sup>9</sup> (C7H4SO<sup>2</sup> )9 ] (Ag<sup>9</sup> -NCs, [Ag<sup>9</sup> (mba)<sup>9</sup> ], H2mba = 2-mercaptobenzoic acid) and L-malic acid (L–MA) to form a phosphorescent hydrogel with stable and bright luminescence, which is ascribed to AIE phenomenon. In this process, the AIE of Ag<sup>9</sup> -NCs could be attributed to the non-covalent interactions between L–MA and Ag<sup>9</sup> -NCs, which restrict the intramolecular vibration and rotation of ligands on the periphery of Ag<sup>9</sup> -NCs, thus inhibiting the ligand-related, non-radiative excited state relaxation and promoting radiation energy transfer. In addition, the fluorescent Ag<sup>9</sup> -NCs/L–MA xerogel was introduced into polymethylmethacrylate (PMMA) to form an excellently fluorescent film for sensing of Fe 3+ . Ag<sup>9</sup> - NCs/L–MA/PMMA film exhibits an excellent ability to recognize Fe 3+ ion with high selectivity and a low detection limit of 0.3 µM. This research enriches self-assembly system for enhancing the AIE of metal NCs, and the prepared hybrid films will become good candidates for optical materials.

**Keywords:** silver nanoclusters; malic acid; self-assembly; AIE; sensor

## **1. Introduction**

Metal nanoclusters (NCs), such as gold, silver, and copper NCs, represent a class of multifunctional materials with attractive optoelectronic and photoluminescence properties [1–5]. It consists of a metal core, composed of several to hundreds of metal atoms and a peripheral organic ligand, forming a unique core-shell structure [6–10]. Due to their large Stokes shift, low toxicity, good biocompatibility, and other excellent characteristics, metal NCs can be used as environmentally friendly and biocompatible color conversion materials, fluorescent probes, and excellent biological probe for protein expression [11–19].

Recently, the aggregation-induced emission (AIE) strategy has been used to enhance the photoluminescence of metal NCs, thereby enhancing its application in fluorescent sensing, light-emitting diodes, and other optoelectronic devices [20–22]. At present, the commonly used methods for AIE are solvent and cation-induced aggregation. However, these two methods cannot obtain ordered aggregates, resulting in instability and poor uniformity of nanocluster aggregation [23–26]. Therefore, an effective strategy for manipulating the spatial arrangement of nanostructured units to form a specific structure: self-assembly, was introduced to solve this problem. By regulating the non-covalent forces between multiple molecules (van der Waals forces, hydrogen bonds, electrostatic interactions, and π-π stacking), metal NCs, and other molecules form ordered aggregates with specific functions, which further improves the photoluminescence performance of the metal NCs [27–29].

**Citation:** Liu, X.; Li, C.; Wang, Z.; Zhang, N.; Feng, N.; Wang, W.; Xin, X. Luminescent Hydrogel Based on Silver Nanocluster/Malic Acid and Its Composite Film for Highly Sensitive Detection of Fe 3+ . *Gels* **2021**, *7*, 192. https://doi.org/10.3390/ gels7040192

Academic Editor: Hiroyuki Takeno

Received: 28 September 2021 Accepted: 26 October 2021 Published: 31 October 2021

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

**Copyright:** © 2021 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/).

For example, Shen et al. used the supramolecular self-assembly strategy to obtain stable colloidal aggregates (nanospheres and nanovesicles) of Ag6-NCs/PEI, through multiple electrostatic interactions between Ag6-NCs and polyethyleneimine (PEI) [30]. During the formation of the order structure, the Ag<sup>6</sup> NCs luminescence in the dilute aqueous solution was turned on. Zhang et al. demonstrated the enhancement of the luminescence of Cu NCs by forming compact and ordered self-assembly architectures by changing the annealing temperature in the formation of Cu NCs [31]. Therefore, it can be seen that the self-assembly of metal NCs is very attractive and worth studying.

Fluorescent film sensors are widely considered, due to their advantages, such as convenient portability, real-time detection, and no pollution to the system, for testing [32–37]. Compared with the solution and powder detection forms, the adjustable shape, size, and flexibility of the fluorescent film makes it have wider application prospects. For example, Li et al. optimized the concentration of octadecylamine with amino groups in n-hexane to make the Tris-base modified silver NCs self-assemble into a single-layer silver nanocomposite film at the n-hexane-water interface and use it as a surface enhancement Raman scattering (SERS) active substrate for ultra-sensitive detection of Hg 2+ ions [38]. The SERS sensing monolayer film also has good reproducibility and recovery rate. Katowah et al. used an in-situ method to prepare a ternary nanocomposite film containing copper oxide/polymethylmethacrylate (PMMA)/various carbon-based nanofillers, as a selective Hg 2+ ion sensor and the mixed nanofiller significantly improved performance of PMMA film [39].

Herein, a water-soluble, atomically precise Ag NCs (NH4)9[Ag9(C7H4SO2)9] (Ag9-NCs, [Ag9(mba)9], H2mba = 2-mercaptobenzoic acid, and the molecular structure of Ag9-NCs is shown in Figure S1) were selected to self-assembled with L-malic acid (L–MA) to construct phosphorescent hydrogels (Scheme 1). The Ag9-NCs/L–MA hydrogel has a hollow-tube structure which is regulated by non-covalent interactions, based on hydrogen bonds, which further promotes the AIE phenomenon. In order to further develop its application prospects as fluorescent sensors, Ag9-NCs/L–MA xerogels are doped into PMMA films to construct fluorescent film sensors for detecting Fe 3+ ions. The value of the corresponding quenching coefficient, KSV, is 5.2 × 10 <sup>4</sup> M−<sup>1</sup> in the low concentration range, and the detection limit of Fe 3+ is down to 0.3 µM.

– **Scheme 1.** Schematic illustration of the hollow-tube structure of Ag<sup>9</sup> -NCs/L–MA hydrogel and its composite film for highly sensitive detection of Fe 3+ .

–

–

–

–

–

π π π –

–

–

– – –

–

– –

–

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

#### *2.1. Self-Assembly of Ag9-NCs/L–MA Hydrogels*

The Ag9-NCs in an aqueous solution is in a nonfluorescent state. In order to regulate the aggregation behavior and AIE of Ag9-NCs, though the non-covalent interaction force dominated by hydrogen bonds, L–MA with hydroxyl and carboxyl functional groups is selected as a hydrogen bond donors, and Ag9-NCs, with carboxylate on the periphery, act as hydrogen bond acceptors. Firstly, phase behavior study on Ag9-NCs/L–MA mixed system was performed. Based on our previous study [40,41], *c*Ag9-NCs was fixed at 8 mM, while *c*L–MA was changed. It can be seen that different phase behaviors of the samples were obtained (Figure 1a). Compared with the solution without fluorescence (0~0.1 M L–MA) and the sample in the two-phase with weaker fluorescence intensity (0.1~0.25 M L–MA), the hydrogel with higher luminous intensity (0.25~0.5 M L–MA) was selected as the main research object. Once the hydrogels formed, the UV absorption exhibits a broad peak from 200 to 550 nm (Figure 1d) and the electrons of the benzene ring system and C=O undergo π-π\* and n-π\* transitions, resulting in ultraviolet absorption bands. Moreover, orange–red fluorescence at 628 nm was excited by blue light at 470 nm (Figure 1d) and it can be observed that as *c*L–MA increases, the fluorescence intensity of the hydrogel first increases and then decreases (Figure 1e,f). This is attributed to the fact that when *c*Ag9-NCs = 8 mM and *c*L–MA = 0.3 M, Ag9-NCs and L–MA self-assemble to form highly ordered aggregates, and L–MA sufficiently limits the ligands of Ag9-NCs, while when *c*L–MA is too high or too low, the uniformity of the formed aggregates is poor, and the self-assembly strategy cannot be effectively implemented, which leads to the weakening of fluorescence intensity. Thus, 8 mM Ag9-NCs/0.3 M L–MA hydrogel, which has the highest fluorescence intensity was selected as the typical sample.

Then, the dynamic change of fluorescence intensity with the hydrogel formation was also studied. Once Ag9-NCs and L–MA mixed, under the drive of non-covalent interaction forces, the fluorescent intensity of 8 mM Ag9-NCs/0.3 M L–MA hydrogel increased immediately within 30 min, achieving AIE effect. After about 1 h, the selfassembly process is almost complete, the fluorescence intensity stabilized and no longer changed (Figure 1g,h). Moreover, the peak position of the fluorescence spectrum has a slight blue shift with time. It is speculated that Ag9-NCs, L–MA and water molecules gradually formed the hydrogel through the hydrogen bonds, the vibration of the peripheral ligand of Ag9-NCs is restricted, and ligand-to-metal charge transfer occurred, resulting in a blue shift of the peak position. Thus, the AIE phenomenon of Ag9-NCs/L–MA hydrogel can be attributed to the limited intramolecular vibration and rotation of the ligand of Ag9-NCs, which inhibits the ligand-related non-radiative excited state relaxation and promotes radiation energy transfer. Furthermore, the average of lifetime of Ag9-NCs solution is 3.277 ns, while the average of the lifetime of 8 mM Ag9-NCs/0.3 M L–MA hydrogel increased to 7.4383 µs (Figure 1i, Table S1), which is approximately 2270 times that of Ag9-NCs solution. Therefore, the large Stokes shift (158 nm) and microsecond lifetime (7.4383 µs) indicate that it is essentially phosphors. Besides, the quantum yield of 8 mM Ag9-NCs/0.3 M L–MA hydrogel, measured using the integrating sphere, is 11.20%, which is higher than that of aggregates formed by the assembly of Ag9-NCs with other substances [42]. Longer fluorescence lifetime and higher quantum yield of Ag9-NCs/L–MA hydrogels make it excellent candidates for luminescent sensing materials.

–

fore, the large Stokes shift (158 nm) and microsecond lifetime (7.4383 μs)

–

–

– reased to 7.4383 μs

–

– – – – – – **Figure 1.** (**a**) Phase transition with different concentration of L–MA. (**b**) Photographs of the hydrogels with different L–MA concentrations. (**c**) Fluorescence photographs of the hydrogels with different L–MA concentrations. (**d**) Photoexcitation (red line, emission = 628 nm), photoemission (black line, excitation = 470 nm) spectra of Ag<sup>9</sup> -NCs hydrogels, and UV–vis absorption (blue line). (**e**) Photoluminescence (PL) spectra of hydrogels with different concentration of L–MA for excitation at 470 nm. (**f**) Comparison of the luminescence intensity of hydrogels with different concentration of L–MA at 628 nm. (**g**) PL spectra of Ag<sup>9</sup> -NCs hydrogels with different incubation times for excitation at 470 nm. (**h**) Comparison of the luminescence intensity of hydrogels with different incubation times. (**i**) PL decay profiles of Ag<sup>9</sup> -NCs hydrogel. Inset: the PL decay profiles of the powder of lyophilized Ag<sup>9</sup> -NCs solution.

Scanning electron microscopy (SEM) and transmission electron microscope (TEM) microscopic observations show that the hydrogel is composed of tangled hollow tubes with a very high aspect ratio (20:1), and most of the tubes are between 1–2 µm in length and about 50–100 nm in diameter (Figure 2a–f and Figure S2). The confocal laser scan microscopy (CLSM) image shows that the tubes structure is accompanied by strong fluorescent properties. Further observation by TEM shows that these hollow tubes are composed of a large number of smaller diameter fibers (10–20 nm) (Figure S3). Moreover, it can be seen in SEM that a small amount of hollow tubes are entangled to form a spiral (Figure 2d). It is speculated that the appearance of AIE can be attributed to the appearance of ordered hollow tube structure. The detailed formation mechanism of the hollow-tube structure of Ag9-NCs/L–MA hydrogel can be shown in Scheme 1.

– –

–

– **Figure 2.** (**a**,**e**) TEM, (**c**) CLSM, and (**b**,**d**,**f**) SEM images of the hollow tubes of 8 mM Ag<sup>9</sup> -NCs/0.3 M L–MA hydrogel with different magnifications (the inset of a is a photograph of the hydrogels).

#### *2.2. Structural Analysis of the Hydrogel*

–

– – In order to further analyze the mechanism of supramolecular self-assembly, a series of characterizations were carried out. Thermogravimetric analyses (TGA) can be used to measure the thermal stability of the studied substances (Figure 3a). The decomposition temperature of L–MA is about 145 ◦C and the decomposition temperature of the H2mba ligand of Ag9-NCs is about 200 ◦C. After assembly, the decomposition temperature of the H2mba ligand of Ag9-NCs for Ag9-NCs/L–MA xerogel is about 260 ◦C, indicating that the stability of Ag9-NCs is further improved by supramolecular self-assembly. – – −1 – – – –

–

–2 μm in length

−1

−1

–

–

modulus (G′

– – – – – – – astic modulus (G′) and viscous modulus (G″) as a function of frequency (τ = 10 Pa). ( – – – – – **Figure 3.** (**a**) TGA and (**b**) FT–IR spectra of lyophilized Ag<sup>9</sup> -NCs solution, L–MA and Ag<sup>9</sup> -NCs/L–MA xerogel. (**c**) SAXS results of Ag<sup>9</sup> -NCs/L–MA xerogel. (**d**) XRD results of lyophilized Ag9–NCs solution, L–MA and Ag<sup>9</sup> -NCs/L–MA xerogel. (**e**) Variation of elastic modulus (G′ ) and viscous modulus (G") as a function of frequency (τ = 10 Pa). (**f**) CD spectra of Ag9–NCs solution, L–MA, Ag<sup>9</sup> -NCs/L–MA hydrogels, D–MA and Ag<sup>9</sup> -NCs/D–MA hydrogels.

–

2π/q), which is equivalent to the size of Ag

–

) is much larger than the loss modulus (G″ = 614.6 Pa),

– ffraction peaks at 2θ = 30.72 correspond to Ag

–

G′ is larger than G″ and

–

–

–MA xerogel at 2θ –

1:√2: 2: √5

Fourier transform infrared (FT-IR) is an effective tool for analyzing the supramolecular forces (Figure 3b). The broad peak of Ag9-NCs/L–MA xerogel at 2500 cm−<sup>1</sup> to 3200 cm−<sup>1</sup> represents the stretching vibration of the hydroxyl group, which is stronger and wider than that of Ag9-NCs, indicating the formation of hydrogen bonds in the system. The peak at 1700 cm−<sup>1</sup> representing the stretching vibration of the C=O in the carboxyl group becomes weaker than that of L–MA after the gel is formed, and slightly shifts to a lower wave number, which represents the formation of hydrogen bonds and confirms that L–MA participates in the construction of the hydrogel. The peak at 3540 cm−<sup>1</sup> represents that the hydroxyl groups in L–MA undergo intermolecular association through hydrogen bonds to form L–MA dimers, which disappear after gelation, indicating that L–MA forms hydrogen bonds with the peripheral ligands of Ag9-NCs. The peaks at 1537 cm−<sup>1</sup> and 1377 cm−<sup>1</sup> belong to the antisymmetric and symmetric stretching vibration of C=O in carboxylate from ligand of Ag9-NCs. For Ag9-NCs/L–MA xerogel, the peak of C=O in carboxylate moves, which further confirms the formation of hydrogen bonds between ligand of Ag9-NCs and L–MA. Moreover, it can be seen that the absorption peak at 900 cm−<sup>1</sup> , representing O–H in the carboxyl group of L–MA is strong and after the hydrogel is formed, this peak becomes weaker and narrower, confirming that most of the carboxyl groups of L–MA are involved in the formation of hydrogen bonds, limiting the bending vibration of the O–H bonds in the carboxyl groups.

Small- angle X-ray spectroscopy (SAXS) and X-ray diffraction (XRD) can effectively characterize the deposition pattern and spatial structure of the xerogels. SAXS shows that Ag9-NCs/L–MA xerogel exhibits four scattering peaks, and the scattering factor q ratio is 1 : <sup>√</sup> <sup>2</sup> : 2 : <sup>√</sup> 5, which is a tetragonal stack (Figure 3c). The smallest repeating unit of the aggregate d = 1.36 nm (d = 2π/q), which is equivalent to the size of Ag9-NCs. From the XRD results, it can be observed that L–MA is a triclinic crystal system, while the peak of Ag9-NCs/L–MA xerogel at 2θ = 20–40◦ represents various possible regular arrangements of atomic layers in assembled nanostructures, which is very different from L–MA and Ag9-NCs (Figure 3d), indicating Ag9-NCs and L–MA formed a multi-complex during the self-assembly. According to the Bragg equation, the interplanar distances of the atomic layer are in the range of 2.37–3.81 Å. Diffraction peaks at 2θ = 30.72 correspond to Ag-Ag, indicating that the d10-d10 argentophilic interaction may exist in the Ag9-NCs/L–MA xerogel [40,41,43].

The rheological characteristics are of great significance for supramolecular hydrogels, and the mechanical strength of the Ag9-NCs/L–MA hydrogels can be evaluated by the rheological measurement. In the stress scanning (Figure 3e), the storage modulus (G′ = 16,900 Pa) is much larger than the loss modulus (G" = 614.6 Pa), indicating that the hydrogels exhibit a solid-like nature. The Ag9-NCs/L–MA hydrogels exhibit good viscoelasticity, a wider linear viscoelastic region, and a higher yield stress (892.8 Pa), indicating that the hydrogels constructed by the Ag9-NCs are more rigid and exhibit strong damage resistance. In the frequency sweep test (Figure S4), G′ is larger than G" and the moduli of the hydrogels are almost independent of the applied frequency. The above results indicate that the Ag9-NCs/L–MA hydrogels have high mechanical strength.

The circular dichroism (CD) spectrum has further confirmed that the Ag9-NCs/L– MA hydrogel possesses supramolecular chirality. L–MA has a positive Cotton effect at 212 nm and no obvious CD signal was detected in the Ag9-NCs aqueous solution. But the Ag9-NCs/L–MA hydrogel has a positive Cotton effect at 230 nm, and the original CD signal of L–MA disappears. Moreover, we also used D-MA to construct Ag9-NCs/D-MA hydrogel and it is interesting to find that the CD spectrum of the Ag9-NCs/D-MA hydrogel is exactly opposite of the Ag9-NCs/L–MA hydrogel. It can be concluded that through supramolecular self-assembly, the molecular chirality of L–MA (or D-MA), was successfully transferred to the supramolecular level, which induced the Ag9-NCs/L–MA hydrogel (Ag9-NCs/D-MA hydrogel) has supramolecular chirality (Figure 3f).

#### *2.3. Ag9-NCs/L–MA Xerogel/PMMA Film for Sensing*

*–*

–

–

–

–

The composite material obtained by doping the xerogel into the polymer matrix has the advantages of easy processing and good flexibility and can enhance the practical application of the gel [32,33]. PMMA has good UV resistance, chemical durability, and good mechanical properties. Therefore, based on the excellent sensing performance of metal NCs and the advantages of simple preparation of organic films as a matrix material, Ag9-NCs/L– MA xerogel are introduced into the organic glass matrix to form a new type of organic film. Herein, the 8 mM Ag9-NCs/0.3 M L–MA xerogel is introduced into the PMMA, and through the solvent volatilization method to get a hybrid film. Firstly, several films were prepared with different doping concentration of 8 mM Ag9-NCs/0.3 M L–MA xerogel and it can be observed that as the doping concentration increases (from 0.1% to 0.4%), the fluorescence intensity increases but the transmittance decreases (Figure 4a–c). Integrating transmittance and fluorescence spectra, a composite Ag9-NCs/L–MA/PMMA film with a mass concentration of 0.3% xerogel was selected to study its sensing performance. – – – – –

–

–

–

–

– **Figure 4.** (**a**) transmittance of films with different doping concentrations. (**b**) transmittance of films with different doping concentrations (400–700 nm). (**c**) FL of films with different doping concentrations (from 0.1% to 0.4%).

– – – Then, 0.3% Ag9-NCs/L–MA xerogel/PMMA films were placed into different metal cation aqueous solutions for 10 min and then, the luminescence spectra of these films were studied (Figures S5 and S6). It is obviously that the composite film shows an obvious quenching effect only toward Fe 3+ (Figure 5a,b). Furthermore, the 0.3% Ag9-NCs/L–MA xerogel/PMMA films were placed into aqueous solutions containing different concentrations of Fe 3+ ions for 10 min to study the corresponding luminescence intensity (Figure 5c,d). With the gradual increase of *c*Fe3+, the luminescence intensities at 628 nm are significantly weakened. In order to investigate the luminescence quenching efficiency, the Stern–Volmer (S-V) equation was used to quantitatively analyze the quenching curve [32]:

$$\mathbf{I}\_0/I = \mathbf{1} + \mathbf{K}\_{SV}[\mathbf{M}] \tag{1}$$

where *I*<sup>0</sup> and *I* are the luminescence intensity at 640 nm when c = 0 and c = *M*, respectively, *KSV* is the quenching coefficient of Fe 3+ ions, and [*M*] is the molar concentration of Fe 3+ ions. The quenching curve shows a well-fitted linear relationship at low *c*Fe3+ (1–100 µM). The *KSV* value was calculated to be 2.3 × 10 <sup>4</sup> M−<sup>1</sup> . According to the detection limit expression defined by IUPAC, the detection limit of the composite membrane is calculated to be 0.3 µM, which indicates that the as-prepared 0.3% Ag9-NCs/L–MA xerogel/PMMA film material exhibited good sensing performance. The limit of detection of the Ag9-NCs/L– MA xerogel/PMMA film far less than the maximum concentration of Fe(III) in drinking water of 5.36 µM (specified by the Minister of Health of the People's Republic of China).

μM, which indicates that the as –

of 5.36 μM (specified by the Minister of Health of the People's Republic of

– – – – – **Figure 5.** (**a**) Luminescence emission spectra and (**b**) luminescence intensities at 640 nm for 0.3% Ag<sup>9</sup> -NCs/L–MA xerogel/PMMA film in different cations. (**c**) Luminescence emission spectra and (**d**) luminescence intensities at 640 nm for 0.3% Ag<sup>9</sup> -NCs/L–MA xerogel/PMMA film in Fe 3+ with different concentrations. The insert of d shows Stern–Volmer quenching curve of the luminescence intensity of 0.3% Ag<sup>9</sup> -NCs/L–MA xerogel/PMMA film at 640 nm against Fe 3+ concentration; (**e**) luminescent emission spectra and (**f**) luminescence intensities at 640 nm for 0.3% Ag<sup>9</sup> -NCs/L–MA xerogel/PMMA film in solutions of different cations with Fe 3+ .

– High sensitivity and high selectivity are the basic requirements of fluorescence sensors. Various mixed metal cation solutions containing Fe 3+ (1.0 equivalent) and different competing cations (1.0 equivalent) were prepared for competition experiments. By comparing the fluorescence intensity of 0.3% Ag9-NCs/L–MA xerogel/PMMA film in different mixed metal cation solutions (Figure 5e,f), it can be seen that the fluorescence is still quenched in the presence of other competing metal cations, which indicates that the composite film has satisfactory selectivity for Fe 3+ detection. In addition, the used films were picked out and washed with distilled water several times, and the corresponding luminescence intensities cannot be recovered. Therefore, it would be used as a disposable test strip for detection Fe 3+ ions in drinking water in the future.

−1

–100 μM

–

– – – The possible sensing mechanism of Fe 3+ ions quenching the fluorescence of 0.3% Ag9-NCs/L–MA xerogel/PMMA film was further analyzed. It can be seen from UV-Vis (Figure 6a) that the absorption peaks of Fe 3+ and the Ag9-NCs/L–MA hydrogel overlap in the absorption range, indicating that the competitive absorption of Fe 3+ reduces the energy transfer efficiency [42]. Moreover, structural collapse or change maybe also cause the luminescence to be quenched [32]. The XRD test was performed on the Ag9-NCs/L– MA xerogel before and after immersion in Fe 3+ ion aqueous solution (Figure 6b). The XRD spectra of the samples were not consistent, which means that the structure of the Ag9- NCs/L–MA xerogel was destroyed after treated by Fe 3+ ion aqueous solution. Therefore, the collapse of the crystal structure is also one of the reasons for the quenching of luminescence. Thus, it can be concluded that the fluorescence quenching of Ag9-NCs/L–MA xerogel/PMMA film is caused by the competitive absorption of Fe 3+ and the destruction of the crystal structure. The fluorescence quenching phenomenon of the Ag9-NCs/L–MA xerogel/PMMA film is shown in Scheme 1.

–

–

–

**Figure 6.** (**a**) UV-Vis absorption of hydrogel and Fe 3+ . (**b**) XRD of xerogel and xerogel after Fe 3+ soaking 24 h.

#### **3. Conclusions**

– – – – – −1 μM. The composite In summary, through self-assembly, Ag9-NCs, and L–MA formed a supramolecular hydrogel, with highly ordered aggregates, by regulating a variety of non-covalent interactions. The nanostructure (tangled hollow tubes) in the hydrogel indicated that L–MA restricts the intramolecular vibration and rotation of the ligand of the Ag9-NCs, so that it can emit a stable and bright orange–red phosphorescent emission. The excellent photoluminescence properties of Ag9-NCs/L–MA xerogel make it likely to be used as highly sensitive probes for Fe 3+ . The Ag9-NCs/L–MA xerogel/PMMA composite film can selectively identify Fe 3+ , the quenching coefficient *KSV* is 2.3 × 10 <sup>4</sup> M−<sup>1</sup> , and the detection limit is 0.3 µM. The composite film also possesses good recyclability and anti-interference ability, with respect to another ions. The current research aims to achieve AIE, by precisely regulating the formation of ordered aggregates of metal NCs, broadening the research field of metal NCs, and enriching the practical applications of these luminescent materials.

#### **4. Experiment Section**

#### *4.1. Materials*

– in the experiments, with a resistivity of 18.25 MΩ cm<sup>−</sup> Ag9-NCs was synthesized and purified, according to our previous work, which have a crystal structure [44]. L–MA and D-MA were purchased from Sinopharm Chemical Reagent Co (Shanghai, China) and used without further purification. Ultrapure water used in the experiments, with a resistivity of 18.25 MΩ cm−<sup>1</sup> , was obtained using a UPH-IV ultrapure water purifier (Sichuan, China). Polymethylmethacrylate (PMMA, average Mw: ~350,000) was purchased from Sigma-Aldrich (Shanghai, China). Dichloromethane (CH2Cl2) were obtained from local supplier with the quality of analytical grade and used without further purification.

#### *4.2. Synthesis of Ag9-NCs/L–MA Hybrid Nanostructures*

*–* In this experiment, 0.5 mL of Ag9-NCs solution (15.87 mM) was added to 0.5 mL L–MA solution (0.6 M) with stirring. The hydrogel was successfully prepared after 8 h of constant temperature (20 ◦C) in a thermostat. The hydrogels was lyophilized in a vacuum extractor at 60 ◦C for 5 day to collect the orange–yellow powder.

#### *4.3. Fabrication of Ag9-NCs/L–MA/PMMA Composite Thin Film*

The PMMA powder (200 mg) was dissolved in dichloromethane (8 mL), then followed by addition of the corresponding required of orange–yellow Ag9-NCs/L–MA xerogels. After being evenly dispersed, place it in a petri dish with a diameter of 6 cm at room temperature for 5 h. Each set of data has been measured 3 times using different batches of film to reduce the error.

#### *4.4. Characterization*

A copper mesh was inserted into the gel to obtain a sample and, after drying under an IR lamp for 45 min, TEM images were observed under a JCR-100CX II (JEOL) microscope. The gel was placed on a silica wafer, dried for 45 min under an IR lamp, and observed by field-emission SEM and AFM, respectively. UV−vis data were recorded on a Shimadzu UV-2600 spectrophotometer. Fluorescence data were tested on an LS-55 spectrofluorometer (PerkinElmer, Waltham, MA, USA) and an Edinburgh Instruments FLS920 luminescence spectrometer (xenon lamp, 450 W), respectively. SAXS measurements were performed using an Anton-Paar SAX Sess mc<sup>2</sup> system with nickel-filtered Cu Kα radiation (1.54 Å) operating at 50 kV and 40 mA. XRD patterns were taken on a D8 ADVANCE (Germany Bruker) diffractometer, equipped with Cu Kα radiation and a graphite monochromator. FT-IR spectra in KBr wafer were recorded on a VERTEX-70/70v spectrophotometer. CLSM observations were performed using an inverted microscope (model IX81, Olympus, Tokyo, Japan), equipped with a high-numerical-aperture 60 oil-immersed objective lens. The rheological measurements were carried out on an Anton-Paar Physica MCR302 rheometer with a cone–plate system. Before the frequency sweep, an amplitude sweep at a fixed frequency of 1 Hz was carried out to ensure that the selected stress was in the linear viscoelastic region. The frequency sweep was carried out from 0.01 to 100 Hz at a fixed stress of 10 Pa. TGA was performed under a nitrogen atmosphere at 25–700 ◦C, with a heating speed of 10 ◦C min−<sup>1</sup> on a TA SDT Q600 thermal analyzer. CD spectra were obtained using a JASCO J-810 spectropolarimeter, which was flushed with nitrogen during operation. The absolute fluorescence quantum yields were measured with a spectrofluorometer (FLSP920, Edinburgh Instruments Ltd., Livingston, UK), equipped with an integrating sphere.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/ 10.3390/gels7040192/s1, Figure S1: The molecular structure of Ag<sup>9</sup> -NCs, Figure S2: AFM image of fibers, Figure S3: (a–b) TEM image of the fibers. (c–d) SEM image of fibers, Figure S4: Elastic modulus (G′ ) and viscous modulus (G") as a function of the applied stress at a constant frequency (1.0 Hz), Figure S5: Photographs of Ag<sup>9</sup> -NCs/L–MA xerogel/PMMA film under different conditions, Figure S6: Luminescence emission spectra of 0.3% Ag<sup>9</sup> -NCs/L–MA xerogel/PMMA film in 100 µM Fe3+ aqueous solution with different interaction time, Table S1: The average lifetime of Ag<sup>9</sup> -NCs and hydrogel.

**Author Contributions:** X.L.: Investigation, methodology, writing—original draft, formal analysis; C.L.: investigation, methodology, formal analysis; Z.W.: investigation, methodology; N.Z.: investigation, methodology, formal analysis; N.F.: investigation, formal analysis; W.W.: investigation, formal analysis; X.X.: conceptualization, resources, writing—review & editing, funding acquisition, supervision. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We gratefully acknowledge the financial support from the National Natural Science Foundation of China (21972077), and Key Technology Research and Development Program of Shandong (2019GGX102019).

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

#### **References**

