**3D Hierarchical, Nanostructured Chitosan**/**PLA**/**HA Sca**ff**olds Doped with TiO2**/**Au**/**Pt NPs with Tunable Properties for Guided Bone Tissue Engineering**

**Julia Radwan-Pragłowska 1, Łukasz Janus 1, Marek Pi ˛atkowski 1,\*, Dariusz Bogdał <sup>1</sup> and Dalibor Matysek <sup>2</sup>**


Received: 28 December 2019; Accepted: 25 March 2020; Published: 2 April 2020

**Abstract:** Bone tissue is the second tissue to be replaced. Annually, over four million surgical treatments are performed. Tissue engineering constitutes an alternative to autologous grafts. Its application requires three-dimensional scaffolds, which mimic human body environment. Bone tissue has a highly organized structure and contains mostly inorganic components. The scaffolds of the latest generation should not only be biocompatible but also promote osteoconduction. Poly (lactic acid) nanofibers are commonly used for this purpose; however, they lack bioactivity and do not provide good cell adhesion. Chitosan is a commonly used biopolymer which positively affects osteoblasts' behavior. The aim of this article was to prepare novel hybrid 3D scaffolds containing nanohydroxyapatite capable of cell-response stimulation. The matrixes were successfully obtained by PLA electrospinning and microwave-assisted chitosan crosslinking, followed by doping with three types of metallic nanoparticles (Au, Pt, and TiO2). The products and semi-components were characterized over their physicochemical properties, such as chemical structure, crystallinity, and swelling degree. Nanoparticles' and ready biomaterials' morphologies were investigated by SEM and TEM methods. Finally, the scaffolds were studied over bioactivity on MG-63 and effect on current-stimulated biomineralization. Obtained results confirmed preparation of tunable biomimicking matrixes which may be used as a promising tool for bone-tissue engineering.

**Keywords:** smart hybrid materials; properties of nanoparticles–reinforced polymers; biotechnology

#### **1. Introduction**

Bone is one of the tissues with the ability of self-regeneration and constant remodeling [1,2]. Despite this fact, each year, over four million surgeries are performed in order to treat this tissue's defects, using autografts or bone substitutes [1,2]. The history of bone pathologies which cannot be self-healed is very long and concerns both congenital and acquired ones. These include traumas, neoplasm, infection or failed arthroplasty, spine arthrodesis, implant fixation, and others [2]. Because of that, bone is known to be the second most commonly replaced tissue [1,2]. Since the application of a patient's own bone is limited by the size of the defect, bone substitutes are rapidly gaining attention. Grafts coming from other donors, so called allografts, carry the risk of transmitting disease, infection, and its rejection. Alternatively, used metal (stainless steel, pure titanium, or its alloys) and synthetic biomaterials have some major limitations, such as poor biocompatibility, biocorrosion, and consumption over time. Non-invasive methods can only accelerate the healing process [2]. To overcome

this issues, novel types of biomaterials are developed which mimic natural bone composition and structure [1–4].

Tissue engineering is one of the most powerful tools used in medicine which brings hope to the millions of patients suffering from various dysfunctions caused by diseases, traumas, or injuries. Its main objective is to restore, improve, or maintain biological functions of damaged tissue [5]. This strategy involves application of the mechanical support called scaffold, which also provides biological help to the cells, due to its chemical composition. To improve their properties, scaffolds can be prepared from various raw materials and are often functionalized with bioactive molecules, such as growth factors or antibiotics [1–3,5].

The biomaterials for bone regeneration are prepared in the form of three-dimensional spatial structure which mimics natural tissue by various methods. Porous materials can be embedded with various substances which promote osteoconduction. The ideal scaffold for this application should meet five conditions, namely biocompatibility, biodegradability, bioactivity, appropriate architecture, and high durability [4–8].

Scaffold dedicated to bone-tissue engineering should provide a temporary mechanical support for the damaged area and stimulate tissue regeneration [2,9]. Therefore, it must be characterized by highly porous architecture, to enable bone ingrowth, as well as neovascularization. The scaffold should constitute a template which promotes extracellular matrix formation and a place for osteoblasts to adhere and proliferate. Such biomaterials can have a form of nanofibers, hydrogels, metal alloys, β-TCP, HA powders, and granules or bioactive glasses [10,11]. Among them, the most promising ones are nanofibrous scaffolds, which can be prepared through electrospinning of the polymer solution since they biomimic natural ECM architecture [12–16]. Their properties, such as high surface area and porosity make them a highly desired materials in terms of bone-tissue-defect treatment [17–19]. However, due to the hierarchical nature of the bone, nanofibrous/porous composites seem to be the most promising ones [5,9].

Scaffolds for bone-tissue engineering that are prepared from minerals have poor mechanical durability, but at the same time, they exhibit good biocompatibility and the potential for various molecules' release. Natural polymers undergoing enzymatic biodegradation which speed rate may be adjusted are used as scaffold components and drug-delivery and -release systems [20]. The most common ones are collagen, fibrin, gelatin, chitin, and its derivatives, or alginate; additionally, biodegradable polymers, such as PLGA/PLA and PCL are widely used [21–24]. Notably, their low durability and fast degradability means that they must be used in conjunction with other materials [14–16]. Electrospinning enables preparation of micro- and nanofibers, using both raw polymers solution as well as composites and nanocomposites [19]. To enhance NFs properties, ready products may undergo postprocessing functionalization by various substances, such as hydroxyapatite, β-TCP, metal, or metal oxide nanoparticles [5,9,10,25–27]. Such an approach enables preparation of bone-tissue scaffolds with tunable properties, including programmed biodegradability rate, enhanced mechanical durability, crystallinity, antibacterial, bioactivity, and others [25,28]. There are some synthetic polymers approved by the FDA, such as polycaprolactone (PCL) and polylactic acid PLA, which can be successfully used for nanofibers' preparation. Other polymers include polyglycolic Acid (PGA) and polyethylene glycol (PEG). Nevertheless, their degradation products cause a local pH decrease due to their acidic nature, which negatively affects proteins and trigger inflammation [1–5]. Chitosan is a poly(saccharide) which is used for porous sponges and nanofibers preparation, yet due to its problems associated with solubility in different solvents, this biopolymer application requires chemical or physical modification [20–24].

The aim of the following research was to develop a novel hybrid material of hierarchical structure with enhanced biological properties. Prepared three-dimensional biomaterials for bone-tissue engineering were developed based on the application of Au [29,30], Pt [31], and TiO2 [32–35] nanoparticles embedded on PLA nanofibers functionalized with hydroxyapatite and highly porous chitosan aerogel matrix. The addition of the nanoparticles has a positive impact on the mechanical properties of the scaffolds, and due to their ability of current conductivity, they enable cell-proliferation stimulation by direct current [25–28]. Moreover, the presence of the NPs has a positive impact on various cell responses by itself [25–27]. The nanoparticles can also help to reduce the risk of bacterial infections, since they exhibit antibacterial activity against various bacterial strains, such as *S. aureus* or *E. coli*. The bactericidal effect occurs due to genetic material damage or disruption of cell membrane [25–28]. The composites were prepared by using polymers of known applicability in tissue engineering [36–38]. Ready scaffolds with tunable properties were characterized over their physicochemical properties, such as chemical structure, crystallinity, morphology, effect on biomineralization, and biocompatibility with human osteosarcoma (MG-63) cell lines [39–41]. Obtained results showed that proposed biomaterials display extraordinary properties and may significantly contribute to bone-tissue-engineering development.

#### **2. Materials and Methods**

#### *2.1. Materials*

Fungal chitosan (300,000 g/mol) and 85% deacetylation degree were purchased from PolAura (Dywity, Poland). Poly(lactic acid) (1,500,000 g/mol, melting point = 210 ± 10 ◦C), acetone (analytical grade), NaCl, NH3, H3PO4, Ca(NO3)2·6H2O, NaHCO3, KCl, K2HPO4·3H2O, MgCl2·6H2O, KOH, HCl, CaCl2, chloroauric acid, Na2SO4, (CH2OH)3CNH2, methanol, ethanol, XTT assay, human osteoscarcoma MG-63 cell line, titanium (IV butoxide), nitric acid, chloroplatinic acid, Dulbecco's Modified Eagle Medium (DMEM) with high glucose content, phosphate buffer solution (PBS), streptomycin/penicillin, and trypsin were purchased from Sigma-Aldrich, Pozna ´n, Poland. All other reagents were of analytical grade, purchased from Sigma-Aldrich, Pozna ´n, Poland.

#### *2.2. Methods*

#### 2.2.1. Nanoparticles' Preparation and Characterization

Briefly, gold nanoparticles were prepared via reduction reaction of chloroauric acid, using sodium citrate at 90 ◦C and mixed for 1 h. Platinum nanoparticles were obtained by using chloroplatinic acid as a platinum source, which was instilled inside heated up ethylene glycol (150 ◦C) with constant mixing for two hours, until black color was reached. For the titanium (IV) oxide NPs obtainment, as a precursor, titanium (IV butoxide) was used. To obtain the nanoparticles, the precursor was mixed with ethanol for 30 min, followed by water ethanol solution addition. The ready white precipitate was washed out with distilled water and left to dry. The ready nanoparticles were purified by membrane dialysis, using MWCO 10,000–12,000 Da (Bionovo, Legnica, Poland). For the preparation of hydroxyapatite nanoparticles Ca (NO3)2·6H2O, ammonia and H3PO4 were used. The ready precipitate was washed out with methanol and calcinated at 250 ◦C for one hour. The resulted nanoparticles' polydispersity index (PDI) was calculated, using the Equation (1), given below, based on the TEM images:

$$\text{PDI} = \text{SD/mean} \tag{1}$$

where PDI is polydispersity index, SD is standard deviation, and mean is the arithmetic mean.

#### 2.2.2. Nanofibers Preparation

To obtain a homogenous solution of poly (lactic acid), previously purified polymer was dissolved in the acetone of analytical-grade purity. The solution concentration was 10%. To prepare nanofibers, Industrial Electrospinning System RT Advanced was applied (Linari NanoTech, Pisa, Italy). Nanofibers were prepared at two different potentials, namely 30 and 35 kV. The distance from the spindle was 10. The polymer solution was dosed with the constant speed of 10 mL per hour, with the rotation speed of 1000 RPM. The collector was coated, using aluminum foil. The needle diameter was 500 μm. Obtained

nanofibers were left under room temperature, until complete solvent evaporation, and peeled of the foil. Dried nanofibrous mats were used for further investigations.

#### 2.2.3. Chitosan Aerogels Preparation

To prepare an aerogel, for each sample, 0.5 of fungal chitosan was dissolved in aspartic acid water solution, until homogenous solution obtainment. Then, 5 mL of 1,2-propanodiol was added, and the vessel containing polymer was placed inside of the Prolabo Synthewave microwave reactor (Wrocław, Poland). The crosslinking reaction was carried out for 10 min. During the first step, water evaporation was noticed, followed by a crosslinking process. The ready product was washed out from unreacted amino acid residues, until pH = 7. Then, it was mixed with the HA NPs (10 wt %). The scaffolds were freeze-dried.

#### 2.2.4. Three-Dimensional Scaffolds' Preparation

The hybrid scaffolds were prepared by placing nanofibrous 3D mat onto chitosan aerogel which contained previously absorbed water (5 mL), followed by lyophilization. The samples were frozen at −20 ◦C and lyophilized at 0.30 mPa, using ALFA lyophilizator (Donserv, Warszawa, Poland). Then, to embed metallic nanoparticles, 1% methanol solutions containing TiO2, Au, and Pt NPs were instilled onto chitosan covered with PLA NFs. The modified biomaterials were left to dry. The chemical composition of the ach sample is given in the Table 1. Each sample contained 0.5 g of the chitosan and 10% *w*/*w* of hydroxyapatite. The concentration of the PLA solution for the fibers' preparation was 10%. As a solvent, acetone was used.


**Table 1.** Three-Dimensional biomaterials' composition.

#### 2.2.5. FT-IR Analysis

Fourier-Transform Infrared Spectroscopy (FT-IR) analysis was performed, using FT-IR Nexus 470 Thermo Nicolet spectrometer obtained from Thermo Fisher Scientific (Waltham, MA, USA). For the measurements, an ATR diamond adapter was used. For FT-IR spectra collecting, each sample was completely dried.

CS-PLA-35-HA-Pt Pt, 1.0 35

#### 2.2.6. Liquid-Uptake Ability and Swelling Degree

Liquid-uptake studies were performed in simulated body fluid (SBF). To determine swelling capability, the weighed samples were immersed in SBF for 24 h and weighed again. Based on the weight change, swelling capacity (SC) was determined. To verify the swelling degree, the change in the dimensions of the scaffolds was studied. For this purpose, the samples were cut into cubes (1 cm × 1 cm × 1 cm) and swollen with SBF. After 5 min and 24, their dimensions were measured. Statistical analysis was performed by Excel software, and a *p* < 0.05 value was found to be statistically significant.

#### 2.2.7. XRD Analysis

X-ray powder diffraction (XRD) was carried out, using BRUKER Advanced D8 (Zastávka, Czech Republic). For each analysis, a fully dried sample was used in the amount of 50 mg.

#### 2.2.8. SEM and TEM Analysis

Nanoparticles morphology and diameter were evaluated by Transmission Electron Microscope (TEM), Jeol (Peabody, MA, USA). For the analysis, nanoparticles were dissolved in the methanol of analytical-grade purity, followed by placing onto formvar-coated Cu mesh and left to evaporate under room temperature. The analysis was carried out under HT = 80,000 V, exposure time of 800 ms, and electron (e) dose of 2771.9 e/nm2. The nanoparticles' size was determined by Jeol software. The scaffolds' components were investigated via an FEI Quanta 650 FEG Scanning Electron Microscope purchased from FEI (ThermoFisher Scientific, Oregon, USA) with HV = 10 kV. The elemental analysis was performed by using an EDAX® adapter (X-ray fluorescence method).

#### 2.2.9. Inorganic Matrix Formation and DC-Induced Inorganic Matrix Formation Study

To determine prepared scaffolds in terms of their positive effect on the inorganic matrix formation, which is an important part of the biomineralization, the scaffolds were placed inside sealed vials containing SBF and stored inside of a CO2 incubator, at 37 ◦C, for 7 days. To evaluate the scaffolds' potential to electrostimulation of biomineralization process, samples were placed inside cubic glass vessels containing two platinum electrodes 5V (direct current) immersed in SBF. Then, the biomaterials were investigated over coverage with HA crystals, using Atomic Absorption Spectroscopy (AAS) PU-9100x (Philips), as well as SEM and XRF methods.

#### 2.2.10. Cytotoxicity Study

Cytotoxicity study on a human osteosarcoma cell line MG-63 was carried out, using 2,3-Bis-(2 methoxy-4-nitro5-sulfophenyl)-2H-tetrazolium-5-carboxanilide salt assay (XTT). The salt underwent reduction reaction due to the presence of enzymes coming from metabolically active cells. The cytotoxicity study was carried out according to norm ISO 10993-5 Biological evaluation of medical devices, tests for in vitro cytotoxicity. For the experiments, an MG-63 cell line was used. Bone cells culture was conducted for 7 days, under typical conditions (95% CO2, high humidity and 37 ◦C). The cell culture medium (DMEM) with high glucose content, 2 mM Glutamine supplemented with 10% fetal bovine serum was changed every two days. To perform XTT assay, UV-Vis spectrophotometer Agilent 8453 was applied (Santa Clara, CA, USA).

Statistical analysis was performed by Excel software, and a *p* < 0.05 value was found to be statistically significant.

#### **3. Results and Discussion**

#### *3.1. FT-IR Analysis*

The 3D hybrid scaffolds were prepared via combination of PLA nanofibers and chitosan crosslinked aerogel containing hydroxyapatite nanoparticles. The biomaterials were further modified with three types of nanoparticles: TiO2, Au, and Pt. The general principle of the supporting matrixes is given in Figure 1.

The hybrid scaffolds' components were characterized by FT-IR method (crosslinked chitosan bottom layer and nanofibrous PLA top layer). Figure 2 shows that, during the chitosan crosslinking process, using L-aspartic acid in the 1,2-propanodiol environment, new bonds were formed between free amino groups and carboxyl groups of the amino acid, since the intensity of amide bonds (1658 cm−<sup>1</sup> for pure chitosan) has significantly increased (1658 cm−<sup>1</sup> for crosslinked chitosan).

The incorporation of the aspartic acid also proves the increased intensity of free amino groups in the crosslinked polymer (1574 and 1148 cm<sup>−</sup>1), compared to pure chitosan (1593 and 1149 cm−1). Such a modification pathway results in the maintenance of free amino groups which are responsible for many positive features and induce cellular responses [20–22]. FT-IR spectrum of the nanofibers prepared from the poly (lactic acid) exhibit typical for this polymer band of high intensity at 1751 cm<sup>−</sup>1, and no bands coming from impurities or not-evaporated solvent can be spotted. The other bands come from C–H

bending (1382 cm<sup>−</sup>1) and C–O stretching (1182–1044 cm−1) [37,38]. What is important is that raw PLA is known for its low bioactivity, as well as hydrophobicity. Thus, application of crosslinked chitosan containing hydrophilic groups can increase cells' adhesion and cellular responses [15]. Moreover, crosslinked polymer spectrum exhibits a band coming from carboxyl groups at 3217 cm−<sup>1</sup> that may be present due to the surface degradation as a result of microwave-irradiation, as well as the aspartic acid molecules, which are not fully incorporated into the polymeric matrix only partially grafted leaving one-COOH group free. Superficial degradation of the chitosan causes very slight changes in its chemical structure, as a result of the oxidation of the hydroxyl groups, namely CH2OH, which turns into COOH. Due to the presence of the oxygen in the reaction atmosphere, carboxyl groups are being formed, thus replacing hydroxyl ones. No other degradation products are observed. Such chemical structure will provide appropriate chemical conditions for bone-forming hydroxyapatite crystallization and stimulate its nucleation initiation, since polymeric layers will act as an organic template [40,41].

**Figure 1.** The general 3D scaffolds' obtainment strategy and application.

**Figure 2.** FT-IR spectra of the pure chitosan, crosslinked chitosan, and poly (lactic acid) nanofibers.

Based on the FT-IR data, the proposed chemical structure is given in Figure 3. Such hierarchical composition should well mimic natural bone. Acidic groups provide local charge accumulation, which helps calcium and phosphorous ions' binding [41]. The chitosan layer mimics the naturally occurring organic bone-forming components on which minerals deposition occurs [40]. Introduced molecules of aspartic acid, as well as poly (lactic acid), enable biological system imitation. Importantly, the chemical structure is rich in both anions and cations, which should help to maintain salts present in the simulated body fluids at elevated level at all times [40].

**Figure 3.** (**a**) 3D hybrid scaffolds' chemical structure doped with TiO2 NPs; (**b**) 3D hybrid scaffolds chemical structure doped with Au NPs; and (**c**) 3D hybrid scaffolds doped with Pt NPs.

#### *3.2. Swelling Properties of the 3D Sca*ff*olds*

Hydrogels, which are three-dimensional polymeric structures capable of water solutions sorption, are highly applied in the field of tissue engineering due to their abilities to mimic natural environments. They also have an ability to react on conditions' changes, such as temperature, pH, or electric field [15]. One of their important parameters is porosity and presence of hydrophilic groups such as amino, hydroxyl, and carboxyl [15], which enables swelling with water and provides space for adhesion and proliferation. Figure 4 presents the results of the swelling-capacity investigations. It can be noticed that all of the prepared samples have excellent swelling capacity.

**Figure 4.** Swelling capacity of the prepared samples.

The best results were obtained for samples CS-PLA-30-HA (232%) and CS-PLA-35-HA (227%), which contained only HA NPs. The scaffolds containing metallic NPs were characterized by slightly worse swelling capacity: 221% and 220% for CS-PLA-30-HA-Pt and CS-PLA-35-HA-Pt, respectively, while for CS-PLA-30-HA-Au, it was 213%, and for CS-PLA-35-HA-Au, it was 211%. The significantly lower sorption efficiency was obtained for the samples containing TiO2 nanoparticles, and this can be caused by pores' clogging by agglomerated NPs which hampered water molecules' migration inside the

3D matrix. What is important is that the studies were performed in the simulated body solution, which contained various ions. Their presence in many cases decreases sorption abilities of the hydrogels due to their interactions with functional groups. In the case of proposed supporting tissue-regeneration materials, all of them exhibited SC above 200%, meaning that they are suitable for bone recovery [1,5,6]. High sorption ability is especially important during the biomineralization process, which significantly hampers water absorption by hydrogels [41]. An excellent swelling capacity also gives them the possibility of incorporation growth factors or genes. Additionally, swelling degree in terms of a scaffold volume change after liquid uptake was investigated (Figure 5). One may observe that all samples, after 5 min, changed their volume, as a result of the soaking up with simulated body fluid. However, the increase was not very high, and this can be explained by the fact that the pores, after contact with water molecules, decreased due to the hydrophilic interactions with free hydroxyl, carboxyl, and amino groups of the scaffold and electrostatic interactions. The swelling degree is correlated with the swelling capacity. Notably, a small change (decrease) of the scaffolds' volume was observed in time, and the volume stabilized after one hour. Taking everything under consideration, although a small change in the samples' dimensions was observed, it can be omitted due to the fact that no post-soaking volume increase was observed. The scaffolds are immersed in aquatic solutions (SBF, phosphate buffer, etc.) before implementation inside a patient's body. If it does not increase its volume in time, it may be assumed that it will not cause any damage to the tissues and nerves placed nearby [15].

**Figure 5.** Swelling degree of the prepared samples.

#### *3.3. XRD Analysis*

The scaffolds were prepared by using both organic and inorganic components. Apart from chitosan and PLA, they contained inorganic phase, namely hydroxyapatite. The biomaterials were additionally functionalized by other nanoparticles types, such as titanium dioxide, gold, and platinum. To confirm the crystallinity with ICDD 9–432 of the inorganic components, XRD analysis was carried out for the obtained compounds. The results presented in Figure 6a prove preparation of crystalline hydroxyapatite, whereas Figure 6b shows that prepared metal oxide is TiO2 nanoparticles.

**Figure 6.** XRD analysis of the scaffolds components: (**a**) hydroxyapatite—the matching patterns are marked with red diamond; (**b**) titanium dioxide—the matching patterns are marked with blue stars.

#### *3.4. Morphology Study of the 3D Sca*ff*olds, and Its Components*

To determine morphological properties of the scaffolds, as well as their semi-components, their structure was analyzed by SEM and TEM microscopy. Figure 7a shows pure crosslinked chitosan aerogel, which has a highly porous structure, with pores dimension in the range between 100 and 500 μm, thus providing appropriate conditions for osteoblasts growth, proliferation, and osteogenesis process meeting requirements for bone-tissue engineering [1,5,6]. The scaffold was further modified with synthetic hydroxyapatite, to increase its affinity to this type of tissue and bioactivity (Figure 7b). It can be noticed that HA is well-dispersed on the scaffold surface and does not block the pores. The HA coverage is uniform. Figure 7c shows XRF analysis of the scaffold elemental composition. It can be noticed that the Ca/P ratio is typical for HA (1.6 approximately) and no contaminants such as heavy metal ions are present [40]. The chitosan scaffolds were further modified with PLA nanofibers, to improve its surface properties [2–5,40]. Figure 7d,e presents the biomaterials surface with incorporated HA nanoneedles, which have a length below 100 nm and width around 20 nm (Figure 7f). The average diameter is 72 nm (length), while the polydispersity index is 0.31. The PLA nanofibers are homogenous and continue. They have coaxial morphology, which means that the fibers have a core–shell structure. The nanofibers obtained under 30 kV are slightly thicker than those prepared under 35 kV. Their density is low enough to provide free diffusion of the gases and nutrients. Their surface is regular. The fiber dimensions are in the range of 0.61–1.56 μm. The average diameter is below 1 μm [2–4]. It can be noticed that the nanohydroxyapatite is well dispersed inside the polymeric matrix, in both cases, and are visible in deeper layers. The scaffolds were also functionalized with other nanoparticles, to improve their bioactivity. Figure 7g,h shows biomaterials doped with TiO2 NPs, which are known for their biosafety and applicability in bone-tissue regeneration [32–35]. They have semi-conductive properties. The titanium dioxide particles are present on the nanofibers surface. However, the SEM images show the spot coverage, not a uniform one, and the NPs are partially aggregated due to their hydrophilic nature in contrast to PLA [11]. The nanoparticles are visible only at the top layer of the biomaterial. There are no significant differences between CS-PLA-30-HA-TiO2 and CS-PLA-35-HA-TiO2 samples. TEM photographs (Figure 7i) present TiO2 nanoparticles that seem

to have an amorphous structure, round shape, and size below 100 nm. The average diameter is 95 nm, while the PDI is 0.18.

**Figure 7.** (**a**) SEM microphotograph of the pure chitosan scaffold (scale bar 500 μm); (**b**) SEM microphotograph of the chitosan scaffolds modified with HA nanoparticles (scale bar 200 μm); (**c**) elemental analysis of the scaffold composition; (**d**) SEM microphotograph of the PLA nanofibers obtained under 30kV covered with HA particles (scale bar 50 μm); (**e**) SEM microphotograph of the PLA nanofibers obtained under 35kV covered with HA particles (scale bar 50 μm); (**f**) TEM microphotograph

of the prepared HA nanoparticles; (**g**) SEM microphotograph of the PLA nanofibers obtained under 30 kV covered with TiO2 nanoparticles (scale bar 50 μm); (**h**) SEM microphotograph of the PLA nanofibers obtained under 35kV covered with TiO2 nanoparticles (scale bar 50 μm); (**i**) TEM microphotograph of the prepared TiO2 nanoparticles; (**j**) SEM microphotograph of the PLA nanofibers obtained under 30 kV covered with Pt nanoparticles (scale bar 50 μm); (**k**) SEM microphotograph of the PLA nanofibers obtained under 35 kV covered with Pt nanoparticles (scale bar 50 μm); (**l**) TEM microphotograph of the prepared Pt nanoparticles; (**m**) SEM microphotograph of the PLA nanofibers obtained under 30 kV covered with Au nanoparticles (scale bar 30 μm); (**n**) SEM microphotograph of the PLA nanofibers obtained under 35 kV covered with Au nanoparticles (scale bar 30 μm); (**o**) TEM microphotograph of the Au nanoparticles.

Next, samples were modified with metallic conductive particles (gold and platinum). Figure 7j,k shows CS-PLA-30-HA-Pt and CS-PLA-35-HA-Pt samples doped with Pt nanoparticles. It can be noticed that the NPs are excellently dispersed and exhibit high affinity to PLA fibers. Interconnected fibers covered with conductive platinum particles may provide good current flow. Again, the NPs are present on the superficial fibers. The morphology of the modified fibers (30 and 35 kV) is almost identical. Interestingly, the Pt NPs created a "spiderweb-like" structure. Figure 7l shows platinum nanoparticles. It can be observed that they are of the size between 10 and 20 nm and of various typical for Pt shapes, like rounded, squared, and triangular. The average diameter is 18 nm, while the PDI is 0.28. Finally, Figure 7m,n provides microphotograph of the nanofibers modified with gold nanoparticles. The Au NPs are well dispersed in the polymeric matrix and exhibit very good affinity to poly (lactic acid) fibers. Most of the superficial fibers are almost fully covered with metallic NPs. What is important, on the contrary to both TiO2 and Pt NPs, these nanoparticles are present both at the surface of the fibrous layer, as well as in the deeper situated ones. However, in the case of CS-PLA-30-HA-Au sample, the Au NPs are mostly visible at external fibers on the opposite to CS-PLA-35-HA-Au, where gold nanoparticles are located more uniformly. This can be explained by the difference in the fibers' diameter compared to PLA NFs prepared under 30 kV (wider fibers) and 35 kV (narrower fibers). Such NPs arrangement should provide good current flow and fiber conductivity. Figure 7o shows gold nanoparticles size and morphology. The particles are of a round, uniform shape, 20–40 nm. The average diameter is 35 nm, while the PDI is 0.14. All of the prepared NPs exhibit typical morphology and should not exhibit cell toxicity during direct contact, as they cannot penetrate the cell membrane nor damage it, due to their lack of sharp edges [25–27]. Their incorporation should provide extraordinary properties to the hierarchized scaffolds and promote biomineralization process [27,32–35]. The proposed structure should enable cell adhesion and proliferation, along with osteogenic differentiation, since they meet architectural requirements for bone-tissue regeneration [40,42–46].

#### *3.5. Biomineralization Study*

The biomineralization process is crucial during bone-tissue regeneration. Thus, scaffolds should not only stimulate osteoblasts' proliferation but also apatite formation. Mineralization may be induced by the presence of functional groups, such as amino, carboxyl, and hydroxyl, which can bind Ca and P cations [40–42]. The results of biomineralization carried out for seven days, under simulated in vitro conditions with and without DC stimulation, are given in Table 2. After one week of bioactivity study, in all samples, mineral sediment was observed. To determine differences between samples, CS-PLA-30-HA (100%) was used as a reference. Standard biomineralization process, which was carried out in simulated body fluid, showed that the addition of TiO2 nanoparticles positively affects HA formation, imitating the in vivo process, since the biomineralization of CS-PLA-30-HA-TiO2 is higher by 17% compared to CS-PLA-30-HA and by 18% in the case of CS-PLA-35-HA-TiO2 compared to CS-PLA-35-HA. In the case of other samples, no superior bioactivity was achieved. There are various methods applied so to accelerate biomineralization, like increased SBF concentration, agitation, or temperature [40]. In this study, an alternative method to enhance biomineralization was proposed based on DC stimulation. The results given in Table 2 show that the application of direct current

causes improvement of biomineralization in the case of samples doped with conductive metallic nanoparticles (Pt and Au), while in the case of samples functionalized with TiO2 NPs (semiconductor) and HA, no effect was observed. Interestingly, the highest increased of the biomineralization occurred for CS-PLA-30-HA-Au (122%) and CS-PLA-35-HA-Au samples (124%). This can be attributed to the highest conductivity of the gold NPs, as well as the highest coverage of the PLA nanofibers not only on the surface but also on the deeper situated fibers. The improvement in the apatite formation can be explained by the forced ions migration present in SBF (calcium, phosphorous) due to the current flow. Moreover, Au NPs are very well-dispersed, which can lead to the increased number of crystallization nuclei. The fibrous/porous membrane-like structure of the scaffolds causes electroosmosis process, which additionally affects the mineralization process. The results obtained for CS-PLA-30-HA-Pt (116%) and CS-PLA-35-HA-Pt (119%) are slightly worse, and they can be attributed to the lower conductivity of the Pt NPs, as well as their presence only on the external fibers. Finally, it may be noticed that, almost in all cases, the samples prepared using PLA nanofibers obtained under 35 kV, exhibiting better properties in terms of biomineralization than those prepared under 30 kV, which was up to 5% for CS-PLA-30-HA and CS-PLA-35-HA samples.


**Table 2.** Biomineralization efficiency after seven days of incubation in SBF.

In both cases, the main process involves nucleation, crystallization, and finally crystals' growth, which corresponds to in vivo mineralization [40]. The proposed mechanism of the biomineralization on the scaffolds' surface involves the presence of hydroxyl and carboxyl groups at the mineralization sites coming from *N*-grafted aspartic, as well as degraded chitosan mers inducing apatite nucleation. The first stage occurs due to the calcium ions binding by anionic groups present on the scaffolds, a finding which corresponds to other researchers' data [36,40–42].

Figure 8 presents the scaffold surface after seven days of in vitro biomineralization. Sedimented ions, which formed mineral precipitate, can be noticed. The coating is quite smooth and uniform, as is desired [40]. XRF analysis confirmed hydroxyapatite formation.

#### *3.6. Cytotoxicity Study of the Prepared Sca*ff*olds*

The newly developed scaffolds are dedicated to bone-tissue engineering, so one of the principal properties for these biomaterials is biocompatibility. The cytotoxicity study was carried out on an MG-63 human osteosarcoma cell line, which is typically used for this purpose [39]. Figure 9 shows results of cell culture in the presence of scaffolds. After seven days of culture, MG-63 cells' growth was not disturbed. It can be noticed that, depending on the chemical composition, the number of viable cells is different. Although in all cases the % of living cells is above 100% (control—cells cultured without scaffolds), the presence of various nanoparticles affects osteoblasts' proliferation activity. Two types of biomaterials display significant bioactivity CS-PLA-30-HA-TiO2 (118%) and CS-PLA-35-HA-TiO2 (117%). This result corresponds with other researchers' data since titanium dioxide nanomaterials are known for their biosafety with bone-tissue cells [32–35]. Slightly worse results were obtained for CS-PLA-30-HA (112%) and CS-PLA-35-HA (113%), and this is not surprising, since the presence of the hydroxyapatite nanoparticles is known to have a positive effect on bone cells [9,27]. Notably, the scaffolds doped with metallic nanoparticles also exhibited good biocompatibility. Gold NPs are well-studied nanomaterials; however, it is known that their in vitro behavior depends on particle size and shape, as well as the presence of stabilizing agents [25–27]. Au NPs prepared for this study due to the size above 10 nm do not penetrate nor damage cell membrane and do not affect cell cycles, a finding which corresponds to other researchers' results [27,29,30]. The results show that, after seven days, the cell number for sample CS-PLA-30-HA-Au was 110% and CS-PLA-35-HA-Au was 108%; thus, their addition did not cause cytotoxic effect. Finally, samples modified with Pt nanoparticles showed 105% and 102% of cells comparing to the reference. Lack of cytotoxicity can be assigned to their size higher than 10 nm and lack of sharp edges. Additionally, in contrast to silver nanoparticles [28], they are characterized by very good chemical stability and are known to disarm reactive oxygen species (ROS), which may have a positive effect on cells' proliferation, since they prevent their apoptosis triggered by oxygen radicals [25–27,31]. Lack of scaffolds' toxicity can be also assigned to the well-known biosafety of semi-components, such as chitosan [34–36] and PLA [2–7], which are widely used for bone-tissue regeneration [38]. Moreover, the scaffolds' architecture seems to provide good conditions for cells' growth and do not hamper their natural behaviors [38]. The findings show that all of the prepared scaffolds can be considered to be non-toxic and can be used for further studies, which should be focused on in vivo experiments [42–46].

**Figure 8.** Biomineralization process on the scaffold CS-PLA-30-HA-Au (**a**) SEM microphotograph of the sample; (**b**) elemental composition of the sample.

**Figure 9.** Metabolic activity of osteosarcoma MG-63 cells after seven days of in vitro cell culture determined by XTT assay, (\*\* *p* < 0.01).

#### **4. Conclusions**

The aim of the following study was to develop novel 3D bioactive scaffolds with hierarchical structure, using a combination of crosslinked chitosan, electrospinning, and conductive nanoparticles. The innovative scaffolds exhibited extraordinary properties, such as high porosity, excellent swelling properties, and ability of biomineralization electrostimulation. Moreover, the new biomaterials met the requirements for bone-tissue engineering and had a positive impact on MG-63 cells' proliferation activity. The findings showed that the highest bioactivity in contact with cells exhibited samples modified with nanohydroxyapatite and amorphous titanium dioxide NPs, while scaffolds containing nanogold showed highest positive impact on DC-stimulated in vitro biomineralization. Owing to nanostructured architecture, physicochemical properties, and biocompatibility, the proposed scaffold may play an important role in bone-tissue engineering development.

**Author Contributions:** Conceptualization, M.P., J.R.-P., and Ł.J.; methodology, M.P.; investigation, D.M., M.P., J.R.-P., and Ł.J.; resources, D.M.; M.P., and Ł.J; writing—original draft preparation, J.R.-P. and Ł.J.; supervision, M.P. and D.B.; project administration, M.P.; funding acquisition, M.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by The National Centre for Science and Development, Poland, grant number [LIDER/42/0149/L-9/17/NCBR/2018]; The Foundation for Polish Science, grant number [START 073.2019].

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

#### **References**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).

### *Article* **3D Ultrasensitive Polymers-Plasmonic Hybrid Flexible Platform for In-Situ Detection**

**Meimei Wu 1, Chao Zhang 1,2, Yihan Ji 1, Yuan Tian 1, Haonan Wei 1, Chonghui Li 1, Zhen Li 1, Tiying Zhu 1, Qianqian Sun 1, Baoyuan Man 1,2 and Mei Liu 1,2,\***


Received: 3 January 2020; Accepted: 30 January 2020; Published: 9 February 2020

**Abstract:** This paper introduces a three-dimensional (3D) pyramid to the polymers-plasmonic hybrid structure of polymethyl methacrylate (PMMA) composite silver nanoparticle (AgNPs) as a higher quality flexible surface-enhanced Raman scattering (SERS) substrate. Benefiting from the effective oscillation of light inside the pyramid valley could provide wide distributions of 3D "hot spots" in a large space. The inclined surface design of the pyramid structure could facilitate the aggregation of probe molecules, which achieves highly sensitive detection of rhodamine 6G (R6G) and crystal violet (CV). In addition, the AgNPs and PMMA composite structures provide uniform space distribution for analyte detection in a designated hot spot zone. The incident light can penetrate the external PMMA film to trigger the localized plasmon resonance of the encapsulated AgNPs, achieving enormous enhancement factor (~6.24 <sup>×</sup> 108). After undergoes mechanical deformation, the flexible SERS substrate still maintains high mechanical stability, which was proved by experiment and theory. For practical applications, the prepared flexible SERS substrate is adapted to the in-situ Raman detection of adenosine aqueous solution and the methylene-blue (MB) molecule detection of the skin of a fish, providing a direct and nondestructive active-platform for the detecting on the surfaces with any arbitrary morphology and aqueous solution.

**Keywords:** SERS; PMMA; AgNPs; in-situ; adenosine; methylene-blue

#### **1. Introduction**

With ultra-sensitive, rapid, and non-destructive properties, surface-enhanced Raman scattering (SERS) is generally considered a valuable analytical technique for achieving label-free detection of biochemical molecules [1–3]. Recent research concerning SERS active substrates focused on designing and fabricating metallic nanostructures with tunable plasmonic features on hard substrates [4–7]. Significant Raman enhancement tends to occur at nanogaps between neighboring metal nanostructures with high local-field intensities (so-called "hot spots") [8–10]. However, in many cases, SERS substrates based on silicon wafers or glass slides are rigid, brittle, and hence unsuitable for directly analyzing target molecules in a solution or gas [11,12]. Compared to traditional rigid substrates, flexible substrates can wrap around complex surfaces and be tailored to the desired size and shape [13]. In addition to achieving a highly active SERS substrate, the effective and rapid collection of probe molecules to the substrate presents a significant challenge [12].

Generally, flexible substrates with simple and periodic two-dimensional (2D) structures are limited by their low density of hot spots distributed at the gaps between horizontal nanoparticals (NPs), which leads to weak Raman enhancement [14]. Thus, to further improve SERS sensitivity, the integration of flexible three-dimensional (3D) structures is imperative. Such 3D flexible SERS substrates can offer large specific surfaces and create high-density hot spots to capture more target molecules [15]. Recently, great efforts were dedicated to fabricating highly sensitive flexible substrates by combining noble plasmonic nanoparticles with polymers. Park et al. have demonstrated a transparent and flexible SERS substrate based on the polydimethylsiloxane (PDMS) film embedded with gold nanostar (GNS) [13]. Kumar et al. fabricated a 3D flexible SERS substrate by depositing AgNPs on structured polydimethylsiloxane using a Taro leaf as the template to detect malachite green [16]. Kumar et al. fabricated a buckled PDMS silver nanorods array as an active 3D SERS sensor for detecting bacteria [17]. Despite demonstrating enormous application prospects for real sample detection, these flexible substrates still involve unresolved issues regarding metal-molecule contact and metal particle oxidation, both of which can cause poor reproducibility and stability in SERS detection [17,18].

In this study, we propose a 3D pyramid-shaped AgNPs@PMMA flexible composite platform for SERS application. The inclined surface of the periodic pyramid-shaped structures could facilitate the aggregation of probe molecules and create high-density 3D "hot spots". In addition, the AgNPs and PMMA composite structures provide uniform space distribution for analyte detection in designated hot spot zone, and the incident light can penetrate the external PMMA film to trigger the localized plasmon resonance of the encapsulated AgNPs, achieving enormous enhancement factor (~6.24 <sup>×</sup> <sup>10</sup>8). The fabricated 3D pyramid-shaped AgNPs@PMMA flexible substrate achieved highly sensitive detection of rhodamine 6G (R6G) and crystal violet (CV), and displayed both high stability and signal reproducibility. Following mechanical stimuli in our experiments, such as stretching and bending, the flexible substrate maintained the stable Raman signal, which corroborates our theoretical simulation. Finally, we successfully achieve the in-situ detection in adenosine aqueous solution and the methylene-blue (MB) molecule detection of the skin of a fish, which demonstrates its great potential for the detecting on the surfaces with any arbitrary morphology and aqueous solution.

#### **2. Materials and Methods**

#### *2.1. Materials*

Acetone (CH3COCH3, 99.5%), alcohol (C2H6O, 99.7%), ethylene glycol (C2H6O2, 99.0%), R6G, crystal violet (CV), MB, silver nitrate (AgNO3), and polymethyl methacrylate (PMMA) were all obtained from Sinopharm Chemical Reagent Co. Ltd. (Shanghai, China). Polyvinylpyrrolidone (PVP, *M*w = 55,000) was obtained from Sigma-Aldrich (St. Louis, MO, USA), and the adenosine sample was purchased from Sangon Biotech (Shanghai) Co. Ltd. (Shanghai, China). The adenosine was dissolved in ultrapure water to make a 0.01M stock solution and then diluted to the final concentration before use.

#### *2.2. Fabrication of 3D Pyramid AgNPs @PMMA Flexible Substrate*

Figure 1 schematically illustrates the fabrication procedure for the 3D-pyramid-shaped AgNPs @PMMA flexible (P–AgNPs@PMMA) substrate. Pyramid Si substrate (P–Si) was pretreated using the method of wet texturing boron-doped monocrystalline silicon in a NaOH solution [19]. AgNPs dispersion was synthesized by the oil bath synthesis, according to previous work [20]. First, 0.075 g of PMMA grains were put into 100 mL of acetone, then this solution was stirred at 80 ◦C for 1 h. High-density and uniform AgNPs dispersion was added into the PMMA solution, and subsequently, this mixed solution was stirred at room temperature and allowed to stand overnight. The mixed solution was then dip-coated on the P-Si substrate, after the mixed layer was dried, the sample was immersed in a NaOH (30%, 100 mL) solution for 1 h to remove the P–Si supporter. The flexible films were rinsed with running deionized water, inverted on a glass substrate, and dried at room temperature to be used as the SERS substrate in our experiments.

**Figure 1.** Schematic illustration of the fabrication process for the P–AgNPs @PMMA substrate.

#### *2.3. Characterization*

Scanning electron microscopy (SEM) images of the P–AgNPs@PMMA substrate were characterized using a scanning electron microscope (SEM, Sigma 500, Carl Zeiss, Jena, Germany) with an energy dispersive spectrometer (EDS) at a voltage of 10 kV. The SERS spectra of the P–AgNPs@PMMA substrate were recorded by a Raman spectrometer (LabRAM HR Evolution, Horiba, Kyoto, Japan) with 532 nm laser excitation. The effective power of the laser source was kept at 0.048 mW, and a 50× objective (N.A. = 0.50) was used throughout the test. The diffraction grid was 600 gr/nm, and the integration time was 8 s. All the SERS spectra in the experiment are expressed on the basis of average spectra. The UV-visible absorption spectra of the synthesized silver nanoparticles was collected by a dual beam UV-visible spectrophotometer (TU-1900, Beijing General Analysis General Instrument Co., Ltd, Beijing, China).

#### **3. Results and Discussion**

To investigate the structural properties of the P–AgNPs@PMMA composite substrates, we performed SEM characterization (Figure 2). AgNPs were deposited on the surface of the P-Si substrate via dip-coating, as shown in Figure 2a. The inclined surface of the periodic pyramid-shaped structures could facilitate the aggregation of probe molecules and create high-density 3D "hot spots", which is critical for the final enhancement activity. Figure 2b,c, respectively exhibit lowand high-magnification SEM images of the P–AgNPs@PMMA flexible substrate after removal of the P-Si substrate. As pictured in Figure S1 (from the Supplementary Materials), the average size of these AgNPs is ∼78nm, and the gaps among nanoparticles are very narrow, which supports huge electromagnetic enhancement for absorbed molecules. The composition of the P–AgNPs @ PMMA flexible substrate is also analyzed with high-resolution EDS mapping in Figure 2d,e, indicating that high-density AgNPs were successfully embedded within the PMMA film. Figure 2f shows a strong UV-vis absorption peak at ~476 nm of AgNPs, which is attributed to the strong coupling of the plasma between the AgNPs. Thus, a high signal-to-noise ratio SERS signal can be generated at 532 nm incident excitation light.

**Figure 2.** (**a**) Scanning electron microscopy (SEM) image of AgNPs deposited on the P-Si substrate; (**b**) and (**c**) are SEM images of the P–AgNPs@PMMA substrate in different magnification; (**d**) and (**e**) energy dispersive spectrometer (EDS) mapping of the P–AgNPs@ PMMA flexible substrates; (**f**) UV–vis absorption spectra showing the SPR peak of the AgNPs at ~476 nm.

It has been reported that high-density "hot spots" are always generated at the slits or tips of metal nanostructures and the interspacing gaps between the adjacent metal nanostructures could markedly affect the ability of detection limit and the intensity of SERS signals [15]. To further optimize the Raman enhancement effect of the composite SERS substrate, we adjusted the distribution density of AgNPs in the PMMA solution to find the best PMMA/Ag volume ratio. Figure 3a shows the SERS spectra for R6G (10−<sup>7</sup> M) absorbed on AgNPs@PMMA/P-Si substrates with different PMMA /Ag volume ratio. The characteristic SERS peaks of R6G were observed around 612, 773, 1308, 1358, 1505, and 1645 cm−1. The peak at 612 cm−<sup>1</sup> is attributed to C–C–C bond stretching modes. The peak at 773 cm−<sup>1</sup> is due to the R6G C−H out-of-plane bending mode. The peaks around 1308, 1358, 1505, and 1645 cm−<sup>1</sup> originate from aromatic C–C stretching [21,22]. It is observed that the Raman signal of the R6G molecule shows an apparent trend from the rise to decline with the continuous addition of silver colloid in a volume of PMMA solution. From Figure 3b, we can see that the SERS performance of the substrate is optimal with the PMMA/Ag volume ratio of 2:3. As shown in Figure S2 (from the Supplementary Materials), the SEM images of different volume ratio AgNPs@PMMA/P-Si substrate further illustrate that the number of AgNPs is small and sparse in the initial stage of the reaction. The area where "hot spots" generated between the silver particles is small, and the Raman signal is weak. As silver particles become denser with the continuous addition of silver colloid, the particle spacing decreases, which causes the increases of the density of "hot spots", and further the Raman signal of the molecule gradually increases. When the PMMA/Ag volume ratio was 2:3, the SERS signal of the molecule reached the strongest. As the silver colloid further addition, the AgNPs accumulate to form large clusters, inhibiting the formation of electromagnetic fields between the particles, resulting in a weakening of the Raman signal. Therefore, the AgNPs@PMMA/P-Si substrates with PMMA/Ag volume ratio of 2:3 were determined as optimal.

**Figure 3.** (**a**) The surface-enhanced Raman scattering (SERS) spectra obtained from the P–AgNPs@ PMMA substrate with different PMMA/Ag volume ratio; here, the concentration of rhodamine 6G (R6G) is 10−<sup>7</sup> M. Insert: the structural formula of R6G molecule; (**b**) change in Raman intensity of R6G at 774 and 1365 cm−<sup>1</sup> as a function of volume ratio; (**c**) SERS spectra of 10−<sup>7</sup> M R6G absorbed on the P–AgNPs@PMMA (black line), P–AgNPs/PMMA (red line) and F–AgNPs@PMMA (blue line) substrates.

To further investigate the SERS effect of the AgNPs and PMMA composite structures and demonstrate the advantages of this 3D pyramid structure for SERS activity, we analyzed and compared the Raman spectra of R6G (10−<sup>7</sup> M) molecules adsorbed on pyramid-shaped AgNPs and PMMA composite (P–AgNPs@PMMA), AgNPs attached to the pyramid-shaped PMMA film (P-AgNPs/PMMA) and planar AgNPs and PMMA composite (F–AgNPs@PMMA) flexible SERS substrates in Figure 3c. Compared to F–AgNPs@PMMA, P–AgNPs/PMMA has a higher SERS signal, which indicates that the inclined surface of the 3D pyramid structure could facilitate the aggregation of probe molecules and create high-density 3D "hot spots". In addition, the AgNPs and PMMA composite structures provide uniform space distribution for analyte detection in designated hot spot zone, avoiding the calculation of Raman enhancement factor error caused by different SERS spectra measured in different regions, achieving reliable enhancement factor measurement. The P–AgNPs@PMMA substrate provides a strong enhancement factor equivalent to P–AgNPs/PMMA substrate, and obtains high spatial uniformity and good stable SERS spectrum. The uniform detection geometry can avoid signal distortion caused by direct contact with the probe molecules to cause deformation of the probe molecules, obtaining a more stable SERS signal. As a whole, the high SERS intensity of the composite P–AgNPs@PMMA substrate benefits from the high-density 3D "hot spots". More importantly, the hybrid detection geometry ensures uniform distribution of analyte molecules in the designated hot spot zone, and the incident light can penetrate the external PMMA film to trigger the localized plasmon resonance of the encapsulated AgNPs, achieving strong enhancement factor and stable Raman signal.

R6G and CV, most commonly used for SERS molecules, were chosen to test the SERS performance of the P–AgNPs@PMMA substrate (Figure 4). Figure 4a illustrates the Raman intensities of R6G deposited on the P–AgNPs@PMMA substrate at various molecular concentrations. It can be seen that the intensity of SERS peaks decreased with the decreasing concentration of R6G solution. Obviously, the 614 and 774 cm−<sup>1</sup> peaks of R6G were still observed on the substrate even when the concentration was decreased to 10−<sup>13</sup> M, indicating that the substrate exhibits strong Raman effects. Here, we selected the 614 cm−<sup>1</sup> peak to further investigate the relationship between the Raman intensity and R6G concentration. The reasonable linear response of the P–AgNPs@PMMA substrate is displayed in Figure 4b. The correlation coefficient (R2) values of the peaks at 614 and 774 cm−<sup>1</sup> can reach 0.988 and 0.989, respectively. In addition, we have calculated the number of adsorbed molecules at different concentrations. The number of adsorbed molecules are estimated by

$$N = \frac{\mathcal{C}\_{\rm m} \ V\_{m} \mathcal{S}\_{\rm laser}}{\mathcal{S}\_{\rm s}} N\_{A\prime} \tag{1}$$

where *C*m is the concentration of molecular solution, *V*m is the volume of molecular aqueous solution dipped onto the whole substrate, *S*<sup>S</sup> is the area of the whole substrate, *S*laser is the area of the focal spot of laser, and *N*<sup>A</sup> is the Avogadro's constant. In the experiments, 20 μL different concentration R6G and CV aqueous solutions were respectively dipped in the substrate. The diameter of the focused laser spot is ~2 μm. The size of the substrate is 5 mm × 5 mm [23]. We obtained the number of adsorbed molecules in the range of 1.51 <sup>×</sup> 106 <sup>−</sup> 0.15, depending on different molecules concentration (10−6–10−<sup>13</sup> M). The number of adsorbed molecules gradually increases with the molecular concentration increases. Linear correlation between concentration and intensity can be supported by the number of adsorbed molecules. These results demonstrate the well linear dependence of the substrate. We calculate the enhancement factors (*EF*) for R6G molecule absorbed on the P–AgNPs@PMMA substrate, using the following standard Equation:

$$EF = \frac{I\_{SERS} / N\_{SERS}}{I\_{RS} / N\_{RS}}.\tag{2}$$

*I*SERS and IRS are the peak intensity of the SERS spectra and the Raman intensity obtained from the SERS sample and pure PMMA film, respectively. *N*SERS and *N*RS are the numbers of target molecules illuminated by the laser spot on the sample and pure PMMA film respectively [24]. The enhancement factor (**EF**) for 614 cm−<sup>1</sup> peak of this hybrid structure is calculated as 6.24 <sup>×</sup> 108. The high EF of the P–AgNPs@PMMA substrate is more superior than the previous metal nanoparticles-based SERS substrates (Table 1), which can be attributed to the combined effect of the 3D-pyramid structure and PMMA film. Pyramid structure demonstrates large specific surfaces, which provide many activity sites, and the inclined surface design of this structure can capture more target molecules. Moreover, the PMMA film acts as a high-dielectric layer and enables the strong surface plasmon coupling between AgNPs embedded in PMMA. The uniform hybrid detection geometry ensures uniform distribution of analyte molecules in the designated hot spot zone, providing reliable enhancement factor measurement, resulting in highly sensitive detection of target molecules.

**Table 1.** Comparing the performance of various flexible SERS sensors.


To further investigate SERS acitivity of P–AgNPs@PMMA substrate, we also collected the Raman signal of the CV molecule with different concentrations (Figure 4c). It is well-known that the Raman cross-section of CV molecules is significantly different from R6G at the same excitation wavelength [29]. The molecule's diameter of CV is 15.10 Å [30]. It can be seen that the Raman main peaks located at ~914, 1178, 1533, 1587, and 1620 cm−<sup>1</sup> are present in the SERS spectra of CV. The two peaks at ~914 and ~1178cm−<sup>1</sup> are attributed to the ring skeletal vibration of radical orientation. Meanwhile, the peaks around ~1533, ~1587, and ~1620 cm−<sup>1</sup> are due to the ring C–C stretching modes. The Raman peaks of CV at 914 cm−<sup>1</sup> can be observed even though the concentration of CV decreased to 10−<sup>12</sup> M, reaching the minimum detection limit, which can be attributed to the 3D high-density SERS "hot spot", and the excitation laser can easily penetrate the external PMMA film, causing strong surface plasmon resonance of the embedded AgNPs. Figure 4d shows the Raman intensity of CV changed linearly with the molecule concentration, and the R2 value of the peaks at 914 and 1587 cm−<sup>1</sup> can reach 0.990 and 0.993, respectively. We obtained the number of adsorbed molecules in the range of 1.51 <sup>×</sup> <sup>10</sup><sup>7</sup> <sup>−</sup> 1.51, depending on different molecule concentrations (10−5–10−<sup>12</sup> M), which present a

good linear fit relationship. These results adequately demonstrate that the P–AgNPs@PMMA substrate possesses perfect SERS performance and good quantitative detection capability.

**Figure 4.** (**a**) Raman spectra of R6G from 10−<sup>6</sup> to 10−<sup>13</sup> M on the P–AgNPs@PMMA substrate. Insert: the structural formula of R6G molecule; (**b**) the Raman spectra of R6G at 614 and 774 cm−<sup>1</sup> as a function of the molecular concentration on the P–AgNPs@PMMA substrate in log scale; (**c**) Raman spectra of CV with concentration from 10−<sup>5</sup> to 10−<sup>12</sup> M on the P–AgNPs@PMMA substrate. Insert: the structural formula of CV molecule; (**d**) the Raman spectra of CV at 914 cm−<sup>1</sup> and 1567 cm−<sup>1</sup> as a function of the molecular concentration on the P–AgNPs@PMMA substrate in log scale.

As a highly sensitive analysis tool, the uniformity and reproducibility of Raman signals are important influence factors for practical applications. We randomly selected 15 spots for statistics on the P–AgNPs@PMMA substrate. Figure 5a shows the SERS signals of the R6G molecules (10−<sup>7</sup> M) collected on the P–AgNPs@PMMA substrate. It can be seen that the SERS spectra of R6G (10−<sup>7</sup> M) from different points did not exhibit obvious changes in intensities or shapes. To more intuitively compare the peak fluctuation, the R6G Raman signal intensities at 614, 774, and 1362 cm−<sup>1</sup> are shown in Figure 5b. These results demonstrate that the intensities of these peaks almost make a horizontal line, and the relative standard deviation of the peaks at 614, 774, and 1362 cm−<sup>1</sup> are 6.913%, 3.926%, and 2.318%, respectively. These percentages are much lower than the scientific standard (20%), which indicates the high homogeneity and reproducibility of the P–AgNPs@PMMA substrate [31]. This is mainly because the AgNPs and PMMA composite structures provide uniform space distribution for analyte detection in designated hot spot zone, resulting in more intense electromagnetic field coupling, and a more uniform SERS signal. Besides, SERS active substrates not only need high sensitivity, high uniformity, but also good signal stability. Thus, time stability is another important parameter in the SERS detection.

We exposed the P–AgNPs@PMMA substrate to the air and performed a SERS test every five days to investigate the intensity change of the peak, as shown in Figure 5c. The Raman signal on the P–AgNPs@PMMA substrate drops very little. Figure 5d shows the corresponding Raman intensities of R6G molecule at 614 cm−1. These experimental results demonstrate the P–AgNPs@PMMA hybrid detection geometry not only ensures uniform distribution of analyte molecules in the designated "hot spot" zone, but effectively prevents the embedded AgNPs from being oxidized by reacting with components in the air, as well as protect the target molecules from deformation and signal distortion caused by the direct interaction between AgNPs and molecules.

**Figure 5.** (**a**) Raman spectra collected from different points on the P–AgNPs@PMMA substrate for R6G molecules. The R6G concentration is measured at 10−<sup>7</sup> M; (**b**) the peaks at 614, 774, and 1362 cm−<sup>1</sup> demonstrate relative intensities collected from the Raman spectra; (**c**) the Raman spectra of the P–AgNPs@PMMA substrate was measured every five days at room temperature; (**d**) the Raman intensity of the 614 cm−<sup>1</sup> peaks for R6G from (c).

Aside from the excellent homogeneity and stability, the P–AgNPs@PMMA substrate also exhibits good mechanical stability (stretching and bending) in practical applications. For tensile testing, the flexible substrate was stretched to ~110%, 120%, 130% (*L*/*L*0; *L*: length after stretching; *L*0: original length), as shown in Figure 6a. The flexible films were then bent in half with various bending cycles (Figure 6b). The optical image of stretching and bending are presented in Figure 6a1, b1. The Raman signals were analyzed for 10−<sup>7</sup> M R6G molecules absorbed on the substrate after each mechanical stimulus (Figure.6a2, b2). In order to draw comparisons, the relative intensities of the characteristic peaks with error bars (same batch but different positions) at 614 and 774 cm−<sup>1</sup> were collected to plot histograms for the stretching and bending processes, as displayed in Figure 6a3, b3. The histograms suggest that the Raman fingerprint peaks at 614 and 774 cm−<sup>1</sup> remain mostly unchanged compared to the original state before mechanical stimulation. It can be seen that the SERS substrate can achieve mechanical deformation with little loss of SERS performance, and can better meet the requirements of SERS detection for uneven surfaces. This may be due to the good mechanical resistance of the PMMA film and the stability of the embedded AgNPs. It is well-known that the intensity of SERS is not only related to the distance of gaps, but also the number of molecules between gaps. During mechanical stimuli, the distance between silver nanoparticles becomes larger, which creates a large "hot spot" zone. The density of "hot spot" decreases within a certain incident laser area, while the number of adsorbed molecules increases in the "hot spot" zone. Therefore, mechanical changes in a certain range will not change the Raman intensity of the corresponding target molecule.

**Figure 6.** Durability tests with mechanical stimuli of the P–AgNPs@PMMA substrate. Schematic illustration: comparative Raman spectra and the SERS intensity of R6G at 614 and 774 cm−<sup>1</sup> peaks; (**a**) after stretching the P–AgNPs@PMMA substrate to ~10%, 20%, 30%; (**b**) after the bending the P–AgNPs@PMMA substrate in half; (**a1**) the optical image of stretching; (**b1**) the optical image of bending; (**a2**) raman signals of stretching; (**b2**) raman signals of bending; (**a3**) histograms for the stretching; (**b3**) histograms for the bending.

During mechanical strain, stretching the P–AgNPs@PMMA substrate yields smaller gaps between neighboring particles, resulting in control over the electromagnetic fields coupling on a nanoscale [32]. To better understand the SERS behaviors of the flexible substrate during tensile testing, we employed commercial COMSOL software to model the local electric field properties of the flexible substrate. Herein, we model the electric field distributions of the SERS film to investigate the enhancement behavior. The height (∼3 μm) and space (∼4 μm) of the P–Si geometry was chosen according to the actual sample. The AgNPs embedded in the PMMA film is modeled as a sphere with a diameter of 78 nm and an interparticle distance of 15 nm, which mirrors the actual size in terms of SEM (Figure S1). Additionally, the incident light is 532 nm, and the refractive index of PMMA is 1.5, according to the actual experiments. Figure 7a–d respectively show the local electric field distributions at the y–z cross-section of the SERS film after stretching to ∼0%, 10%, 20%, and 30%. The bottom left in Figure 7d exhibits the theoretical model of P–AgNPs@PMMA substrate. SEM image of the single pyramid-shaped AgNPs structure is also present in the inset of Figure 7d (bottom right). These results indicate that the electric field near the surface of the pyramid does not obviously change as the stretching length increases. Although some subtle differences have occurred, the plasmon coupling between high-density AgNPs embedded in the PMMA film remains unaffected, allowing for almost constant Raman signals, which is consistent with our experimental results.

**Figure 7.** The respective Y–Z views of the electric filed distributed on the P–AgNPs@PMMA substrate after stretching to ~0% (**a**), 10% (**b**), 20% (**c**), and 30% (**d**).

To further investigate the practicability of the P–AgNPs@PMMA substrate, the in-situ detection of a solution containing biochemical molecules was performed. Adenosine is the metabolite of adenine nucleotides, which are regarded as major neuromodulators. Furthermore, as the core molecule of ATP and nucleic acids, adenosine forms a unique link among cell energy, gene regulation, and neuronal excitability [18,33]. Here, adenosine molecules were dissolved in deionized water, and the molecular concentration was measured at 10−<sup>8</sup> M. The substrate was placed on the surface of the prepared adenosine aqueous solution. Next, in-situ detection was conducted, as exhibited in Figure 8a. The optical picture of the in-situ detection of adenosine molecules can be observed in Figure 8b. The detected results are presented in Figure 8c. There were no obvious Raman signals (orange line) obtained without the P–AgNPs@PMMA substrate. However, when the activated surface of the P–AgNPs@PMMA substrate encountered the solution, the characteristic peaks at 729, 1257, and 1329 cm−<sup>1</sup> of adenosine molecules were easily detected (green line), which was attributed to the active surface of the substrate interacted with probe molecules, enhancing the Raman signal. The experimental results suggest the substrate's promising application to practical in-situ detection of biochemical molecules. Recently, MB is often used as the fish medicine or the disinfector for the fishpond. If MB molecules have not been completely removed from the skin of fish, it would be harmful to humans' health. We perform MB molecular detection by swabbing the surface of the fish skin using the P–AgNPs@PMMA substrate shown in Figure 8d. Figure 8e exhibits the SERS spectra of MB molecules. Obviously, the Raman intensity decrease with the decrease of the MB concentration. Based on these particle detection, the P–AgNPs@PMMA substrate has shown great potential in noninvasive and ultrasensitive molecular detention, such as the detecting on the surfaces with any arbitrary morphology and aqueous solution.

**Figure 8.** (**a**) Schematic illustration showing the in-situ detection process of the adenosine solution; (**b**) the optical picture of the in-situ detection of adenosine molecules; (**c**) Raman spectra of adenosine molecules before (orange line) and after (green line) placing the P–AgNPs@PMMA substrate on the adenosine solution to confirm the SERS effect. Insert: in-situ detection of adenosine molecule; (**d**) the optical picture of detecting of methylene-blue (MB) by swabbing the marine fish surface; (**e**) SERS spectra of MB obtained by swabbing fingerprint on the marine fish surface.

#### **4. Conclusions**

In summary, we proposed a rapid and convenient method for fabricating 3D P-AgNPs@PMMA flexible platforms, which enable highly sensitive single-molecule detection and provide a reproducible and stable Raman signal response. Benefiting from the effective oscillation of light inside the pyramid valley could provide wide distributions of 3D "hot spots" in a large space. The inclined surface design of the pyramid structure could facilitate the aggregation of probe molecules, achieving highly sensitive detection of R6G and CV. The AgNPs and PMMA composite structures provide uniform space distribution for analyte detection in designated hot spot zone, and the incident light can penetrate the external PMMA film to trigger the localized plasmon resonance of the encapsulated AgNPs, achieving enormous enhancement factor (~6.24 <sup>×</sup> 108). Additionally, the substrate maintains a stable SERS signal under various mechanical stimuli such as stretching and bending. As a practical application of the SERS substrate, we achieved the in-situ Raman detection of adenosine aqueous solution and the MB molecule detection of the skin of a fish. Our experimental results suggest promising application prospects for detection on the surfaces with any arbitrary morphology and aqueous solution.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4360/12/2/392/s1, Figure S1: SEM image and the EDS spectrum of the P–AgNPs@PMMA substrate; Figure S2: SEM images of different volume ratio AgNPs@PMMA/P–Si substrate.

**Author Contributions:** Conceptualization, Y.J., Y.T. and H.W.; data curation, T.Z. and M.W.; writing—original draft preparation, M.W. and M.L.; writing—review and editing, C.L., Z.L. and M.L.; funding acquisition, C.Z., Q.S., B.M. and M.L.; All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Natural Science Foundation of China (Grant No. 11774208 and 11974222) and Shandong Provincial Natural Science Foundation, China (Grant No. ZR2018MA040).

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

#### **References**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).
