**Novel Nanocomposites Based on Functionalized Magnetic Nanoparticles and Polyacrylamide: Preparation and Complex Characterization**

**Eugenia Tanasa 1,2, Catalin Zaharia 3,\*, Ionut-Cristian Radu 3, Vasile-Adrian Surdu 1,2, Bogdan Stefan Vasile 1,4, Celina-Maria Damian <sup>3</sup> and Ecaterina Andronescu 1,4**


Received: 1 September 2019; Accepted: 23 September 2019; Published: 27 September 2019

**Abstract:** This paper reports the synthesis and complex characterization of nanocomposite hydrogels based on polyacrylamide and functionalized magnetite nanoparticles. Magnetic nanoparticles were functionalized with double bonds by 3-trimethoxysilyl propyl methacrylate. Nanocomposite hydrogels were prepared by radical polymerization of acrylamide monomer and double bond modified magnetite nanoparticles. XPS spectra for magnetite and modified magnetite were recorded to evaluate the covalent bonding of silane modifying agent. Swelling measurements in saline solution were performed to evaluate the behavior of these hydrogels having various compositions. Mechanical properties were evaluated by dynamic rheological analysis for elastic modulus and vibrating sample magnetometry was used to investigate the magnetic properties. Morphology, geometrical evaluation (size and shape) of nanostructural characteristics and the crystalline structure of the samples were investigated by SEM, HR-TEM and selected area electron diffraction (SAED). The nanocomposite hydrogels will be further tested for the soft tissue engineering field as repairing scaffolds, due to their mechanical and magnetization behavior that can stimulate tissue regeneration.

**Keywords:** magnetic nanoparticles; polyacrylamide; functionalization; nanocomposite; hydrogel

#### **1. Introduction**

Polymeric hydrogel-like materials are a category of soft materials containing crosslinked hydrophilic networks with a high swelling ability. The hydrophilic nature of the macromolecular chains is based, in general, on side hydrophilic active groups [1–4]. The cross-linking reaction of hydrophilic chains is an absolute requirement for dissolution avoiding of polymeric material. The generation of a cross-linked network assumes formation of inter and intramolecular bridges, which do not allow the solvent molecules to solve and unfold the macromolecules. Thus, the solvent can only penetrate among polymeric molecules and swell the material [5,6]. In the swollen state, the polymeric hydrogel exhibits brittleness and obvious low mechanical properties. These disadvantages seriously limit their usage in special biomedical applications. The use of polymeric hydrogels is directly related to the intrinsic mechanical properties in the swollen state. A relatively new concept of polymeric nanocomposite hydrogels has started to overcome these problems by combining the advantage of polymeric hydrogels with the advantage of polymeric nanocomposites [7–15]. Nanocomposite

hydrogels have been developed by various methods, such as in situ polymerization or pre-modified inorganic nanoparticles [16–21]. Modified inorganic nanomaterials have gained special attention, as they can be used as inorganic crosslinkers. These types of modified crosslinkers exhibit a unique flexible intrinsic structure with a serious contribution to improving mechanical properties [22]. The major limitation of the swollen hydrogels is related to the network generation process based on traditionally low molecular weight organic crosslinkers. The limitation of classic organic crosslinkers, due to their relative low number of available groups for reactions with polymeric chains, can be overcome by inorganic nanoparticles modified with multiple groups. A suitable modification involves designing molecular architectures with long and short intermolecular and intramolecular bridges at the same time [7,23–27]. The mechanical stress generates the fracture first of short chains to partially dissipate the elastic energy, while the long chains take the remaining loading. Meanwhile, the hydrogel is still intact [4,28–31]. Inorganic nanoparticles as crosslinkers possess high stretchability, elasticity and superior toughness for polymeric nanocomposite hydrogels, with potential use in soft tissue applications. Inorganic nanoparticles such as magnetite exhibit a high potential for modification with functional groups, due to the presence of hydroxyl groups. They show outstanding physico-chemical properties due to the presence of both species of iron [32–34]. Furthermore, magnetite has been used with great success for various biomedical applications [34–37], including cellular imaging [38] or cancer diagnosis, monitoring and treatment [39].

This research study is focused on the development of nanocomposite networks crosslinked by highly-functionality modified magnetite with enhanced stretchability and elasticity for biological tissue applications.

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

#### *2.1. Materials*

The reagents used for the synthesis of the magnetic iron oxide nanoparticles were iron chloride iron (III) chloride (FeCl3, 97%), ferrous sulfate heptahydrate (FeSO4·7H2O) and ammonium hydroxide solution (NH4OH). The acrylamide monomer, 3-trimethoxysilyl propyl methacrylate modifier agent and potassium persulfate initiator were used for the preparation of hydrogels. All the reagents were supplied by Sigma-Aldrich, 3050 Spruce Street, St. Louis, MO, United States.

#### *2.2. Synthesis of Magnetite (Fe3O4) Nanoparticles*

The synthesis of the Fe3O4 nanoparticles (MNPs) was carried out at room temperature, by co-precipitation method, starting from iron (III) chloride, ferrous sulfate heptahydrate and ammonium hydroxide solution [40,41]. The iron chloride was dissolved in deionized water to give a clear solution. Under vigorous magnetic stirring, the FeSO4·7H2O was added to the solution (Fe2+/Fe3<sup>+</sup> <sup>=</sup> 1:2 molar ratio). Independently, an aqueous solution of ammonium hydroxide is prepared, and the mixture solution resulting from the iron chloride and ferrous sulfate heptahydrate was added to it. Magnetite nanoparticles formed and precipitated. The MNPs were separated from the reaction medium using a strong magnet. The powder was rinsed several times with distilled water until reaching a neutral pH (pH = 7) in the washing solution. After washing, the precipitate was dried for 12h in air oven, at 60 ◦C.

#### *2.3. Synthesis of Double Bond Modified Magnetite Nanoparticles*

The surface modification of the magnetic nanoparticles with double bonds was carried out in several steps, as follows (Figure 1). Briefly, 2 g of MNPs were reacted with 4 mL of 3-trimethoxysilyl propyl methacrylate (3-TPM) by dispersion in 40 mL of toluene for 24 hours at room temperature under magnetic stirring. The modified magnetic nanoparticles (denoted by MMNPs) were then washed several times with toluene to remove the unmodified MNPs and unreacted 3-TPM by centrifugation and then dried.

**Figure 1.** Modification of magnetite nanoparticles with double bonds.

#### *2.4. Preparation of Polyacrylamide*/*MMNPs Nanocomposite Hydrogels (PAA*/*MMNPs)*

Hydrogels were obtained by free-radical polymerization of acrylamide and MMNPs in aqueous solution (Figure 2). Briefly, various ratios between acrylamide monomer and MMNPs (90/10; 80/20; 70/30; 60/40 and 50/50 *w*/*w*) were prepared. The MMNPs were dispersed in water by sonication and added in a mixture of 15 wt. % aqueous acrylamide solution and initiator (potassium persulfate). The ratio between organic phase (acrylamide) and MMNPs was varied in order to enhance the mechanical properties of the hydrogels. The nanocomposite hydrogel samples were added in circular glass matrix and put at 60 ◦C for 24 h. Finally, the nanocomposite hydrogel samples were removed from the glass matrix and immersed in distilled water for 5 days to remove residual monomer and final purification. Hydrogels were cut as disks for further mechanical investigations (rheological measurements).

**Figure 2.** Preparation of polyacrylamide (PAA)/modified magnetic nanoparticles (MMNPs) nanocomposites.

#### *2.5. Swelling Measurements*

Swelling behavior of the hydrogels was performed in saline solution at 37 ◦C. The weight changes of the hydrogels were recorded at regular time intervals during swelling. The swelling degree of the hydrogels was determined according to the following equation [42,43]:

$$SD = \frac{\mathcal{W}\_t - \mathcal{W}\_o}{\mathcal{W}\_0} \cdot 100,\tag{1}$$

where W and W0 denote the weight of the wet hydrogel at a predetermined time and the weight of the dry sample, respectively. The equilibrium swelling degrees (ESD) were measured until the weight of the swollen hydrogels was constant. At least three swelling measurements were performed for each hydrogel sample and the mean values were reported.

Swelling kinetics. The dynamics of the water sorption process was studied by monitoring the saline solution absorption by the hydrogels at different time intervals. For diffusion kinetic analysis, the swelling results were used only up to 60% of the swelling curves. Fick's equation was used [42–49]:

$$f = k \cdot t^{\prime},\tag{2}$$

where f is the fractional water uptake, k is a constant, t is swelling time and n is the swelling coefficient that indicates whether diffusion or relaxation controls the swelling process. The fractional water content f is Mt/Mn where Mt is the mass of water in the hydrogel at time t, and Mn is the mass of the water at equilibrium.

#### *2.6. Characterization Methods*

FTIR analysis. FTIR spectra of native magnetite and 3-TPM modified magnetite were recorded on a Bruker Vertex 70 FT-IR spectrophotometer with attenuated total reflectance (ATR) accessory with 32 scans and 4 cm−<sup>1</sup> resolution in mid-IR region.

XPS analysis. The X-ray photoelectron spectroscopy spectra for magnetite and modified magnetite were recorded to evaluate the covalent bonding of silane modifying agent. The spectra were recorded on a K-Alpha instrument from Thermo Scientific, using a monochromated Al Kα source (1486.6 eV), at a pressure of 2 <sup>×</sup> <sup>10</sup>−<sup>9</sup> mbar.

#### 2.6.1. Evaluation of the Rheological Properties for the Nanocomposite Hydrogels

Rheological tests were performed with a rotational rheometer Kinexus Pro, Malvern Instruments, and a temperature control unit. In oscillating mode, a parallel plate and a geometric measuring system were used, and the gap was set according to the force value. The tests were performed on samples of 20 mm diameter with parallel plate geometry in a frequency range 1 to 30 Hz.

#### 2.6.2. Magnetic Properties by Vibrating Sample Magnetometry (VSM)

Vibrating sample magnetometry (LakeShore 7404-s VSM) was used in order to investigate the magnetic behavior of the hydrogels. Hysteresis loops were recorded at room temperature with an applied field up to 15 kOe, increments of 200 Oe and ramp rate of 20 Oe/s.

#### 2.6.3. Morphological Characterization by Scanning Electron Microscopy (SEM) and Transmission Electron Microscopy (TEM)

The microstructure of the samples was analyzed by Scanning Electron Microscopy (SEM) using a Quanta Inspect F50, with a field emission gun (FEG) having 1.2 nm resolution and an energy dispersive X-ray spectrometer (EDXS) having 133 eV resolution at MnKα. Morphology, geometrical evaluation (size and shape) of nanostructural characteristics and the crystalline structure of the samples were investigated by high-resolution transmission electron microscopy (HR-TEM) and selected area electron diffraction (SAED) using a TECNAI F30 G2 S-TWIN microscope operated at 300 kV with energy dispersive X-ray analysis (EDAX) facility.

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

#### *3.1. Swelling Measurements*

The most important property of a hydrogel is its ability to absorb and hold an amount of solvent in its network structure. The equilibrium swelling of a hydrogel is a result of the balance of osmotic forces determined by the affinity to the solvent and network elasticity. Hydrogel properties depend strongly on the degree of cross-linking, the chemical composition of the polymer chains, and the interactions of the network and surrounding liquid. Figure 3 shows the water swelling behavior of the PAA/MMNPs hydrogels. The swelling curves show a decreasing trend of swelling degree with the increase of the modified magnetite nanoparticles content (Figure 3). These results are sustained by the fact that a higher amount of MMNPs lead to a higher crosslinking density. The crosslinking of the hydrogel comes from the reaction between the double bonds from NPs surface and the double bonds of the acrylamide monomer without the adding of any other crosslinker.

**Figure 3.** Swelling degree versus time in saline solution at 37 ◦C for PAA/MMNPs hydrogels.

Next, the swelling mechanism is evaluated by Equation (2). Here, by plotting ln f versus ln t, we may calculate the swelling coefficient n as the slope of the linear graph. It is known that the swelling process could be controlled by a Fickian-type mechanism, by relaxation of the chain or by both mechanisms depending on the composition. The values of n were below 0.5 for 2 samples (PAA/MMNPs 70/30, 60/40 ratio), which means a diffusion-controlled process (Fickian mechanism). The other three nanocomposite samples (PAA/MMNPs, 90/10, 80/20 and 50/50 ratio) are governed by a diffusion swelling coefficient with values above 0.5 and a water molecules transport model, done by chain relaxation [50,51]. These data are shown in Table 1.



#### *3.2. FTIR Analysis*

The modification of magnetite nanoparticles with 3-TPM was proved by FTIR investigation (Figure 4). FTIR spectrum of modified magnetite shows several new peaks specific to organic modifier 3-TPM. Therefore, the peak at 1170 cm−<sup>1</sup> can be assigned to stretching vibration of ester bonds; peaks at 1299 cm−<sup>1</sup> and 1325 cm−<sup>1</sup> can be assigned to the stretching vibration of -Si-methylene- from the internal structure of modifier agent; peaks at 1454 cm−<sup>1</sup> and 1412 cm−<sup>1</sup> can be assigned to the bending vibration of methyl and methylene groups from the internal structure of the modifier agent; the peak at 1638 cm−<sup>1</sup> is specific to the stretching vibration of –C=C– from the internal structure of the modifier agent; the peak at 1719 cm−<sup>1</sup> is specific to the stretching vibration of carbonyl –C=O from the internal structure of the modifier agent [52]. Considering all of the attributed peaks, FTIR analysis was a very useful tool to evidence the modification of the magnetite nanoparticles with double bonds.

**Figure 4.** FTIR spectra for magnetite and double bond functionalized magnetite nanoparticles.

#### *3.3. XPS Analysis*

XPS analysis for both magnetite and double bond modified magnetite was carried out in order to reveal the interstitial organic/inorganic character of new generated magnetite lattice. The results for surface modification are well correlated with the reaction mechanism and morphological results. There is an increasing of C1s in the elemental composition up to the main elemental percent, due to the modification on the surface of magnetite nanoparticles. Figure 5 highlights the high resolution spectra of the O1s species from crude magnetite with two deconvoluted peaks, the first centered at 530.35 eV, which can be attributed to O-Fe in magnetite phase [53], and the second centered at 531.01 eV, probably corresponding to the hydroxyl bonding within magnetite lattice. Furthermore, Figure 5 reveals the high magnification spectra of O1s species for functionalized magnetite nanoparticles with three secondary deconvoluted peaks. The two O1s peaks at 529.67 eV and 531.13 eV can be attributed to the crude magnetite structure and the new peak centered at 533.01 eV can be attributed to a Si-O new formed species by covalent bonding of silane with magnetite hydroxyl groups [22].

**Figure 5.** XPS spectra of magnetite and double bond modified magnetite.

#### *3.4. Evaluation of the Rheological Properties for the Nanocomposite Hydrogels*

Rheological behavior of novel nanocomposites was performed on swollen samples in aqueous NaCl 0.9 wt% solution at swelling equilibrium. The investigation involves the stress optimization in order to maintain a linear viscoelastic domain and samples to be dependent only on frequency and not on the applied stress. The elastic modulus for nanocomposite with 10% modified magnetite nanoparticles showed a unique behavior with significant differences, as compared to other samples. Figure 6 reveals a slow decreasing elastic of the modulus G' up to 20 Hz, followed by a fast increasing until 30 Hz for the sample with 90% PAA and 10% modified magnetite nanoparticles. This behavior can be explained by a low amount of modified magnetite nanoparticles, which act as a crosslinking agent. The low amount of inorganic modified agent does not allow the specific elastic network to adapt to environmental mechanical changes [22]. The nanocomposite samples with a higher amount of modified magnetite nanoparticles (30%, 50%) showed a different specific elastic behavior with frequency variation, presenting a constant elastic modulus increasing from 1Hz up to 30 Hz. The specific elastic behavior allows for the environmental changes, due to the formation of elastically active chains by bridging multiple surrounding chains with various lengths. In the case of 30% modified magnetite nanoparticles, the elastic modulus exhibited higher values over the frequency range. This is probably due to the nanoparticles concentration that is optimal for a good dispersion into polymer matrix. In the case of the 50% modified magnetite nanoparticles, the elastic modulus showed lower values, probably due to a lower dispersion in the matrix, with significant influences on the segmental mobility of the 3D network.

**Figure 6.** Elastic modulus G' versus frequency.

#### *3.5. Magnetic Properties by Vibrating Sample Magnetometry (VSM)*

The magnetic properties of the magnetic iron oxide nanoparticles (Fe3O4 NPs) and of the hydrogels were investigated by vibrating sample magnetometry (VSM) at room temperature. In Figure 7, the magnetic hysteresis loops that are characteristic of superparamagnetic behavior can be observed for all of the samples, due to the presence of the magnetite nanoparticles. Superparamagnetism is the responsiveness to an applied magnetic field without retaining any magnetism after removal of the applied magnetic field. The measured saturation magnetization (Ms) of the Fe3O4 NPs is 63.128 emu/g. For PAA-MMNPs 90:10, the saturation magnetization was found at 9.74 emu/g, the lowest measured saturation of the hydrogels. The saturation magnetization for the PAA-MMNPs 70:30 was found at 26.73 emu/g and the highest saturation magnetization was at 31.88 emu/g, corresponding to the PAA-MMNPs 50:50, the hydrogel with the highest concentration (50%) of MMNPs. These results show that the magnetization of the hydrogels increases with the increase of the concentration of MMNPs present in the hydrogels.

**Figure 7.** Vibrating sample magnetometry (VSM) magnetization curves of the Fe3O4 nanoparticles and the nanocomposites hydrogels.

#### *3.6. Morphological Characterization by SEM and TEM*

#### 3.6.1. SEM Analysis

The microstructure of the PAA-MMNPs hydrogels was studied by SEM in cross-section and the results are shown in Figure 8 (PAA-MMNPs 90:10) and Figure 9 (PAA-MMNPs 50:50). The image in Figure 8A, (magnification ×2.000) shows submicronic areas of bright contrast (functionalized magnetite aggregates) evenly distributed in a dark contrast PAA matrix. At higher magnifications (×200.000, Figure 8B) it can be observed that the areas of bright contrast are aggregates of MMNPs. Also, the image shows that the modified Fe3O4 nanoparticles showed a good distribution in the polymer matrix by the presence of areas with high dispersed MMNPs and areas with local agglomeration of MMNPs. However, even the local agglomerations revealed that the modified magnetite nanoparticles (MMNPs) seem to be addressed by the polymer polyacrylamide matrix due to the effect of the crosslinking agent of the MMNPs (Figure 8A,B). Thus, the polymer matrix covering the MMNPs is chemically linked by the MMNPs and the whole ensemble displays a crosslinked network-like architecture.

**Figure 8.** SEM micrographs of PAA-MMNPs 90:10 block hydrogel (**A**,**B**) and lyophilized PAA-MMNPs 90:10 hydrogel (**C**,**D**).

Figure 8C is a SEM backscattered electron image at a smaller magnification (×500), showing small MMNPs agglomerates (white spots) uniformly dispersed on a mesh of micro-pores. Figure 8D is a detail (×100.000 magnification) of a nano-size area from the central zone in Figure 8C, showing a nanostructure of the lyophilized hydrogel as fibrils having evenly incorporated MMNPs. The crosslinked network-like ensemble generated by the MMNPs is better highlighted by the lyophilized samples (Figure 8C,D). The fibrils revealed branches-like structures which are extending on the sample surface and evenly through the sample internal structure. The branched-like structures exhibited MMNPs linked to each other by the polymer matrix and serve as the basis of the crosslinked network-like ensemble.

The SEM image in Figure 9A (magnification ×2.000) shows a higher density in MMNPs clusters for the PAA-MMNPs 50:50 hydrogel due to the higher amount of modified magnetite, in comparison to the PAA-MMNPs 90:10 hydrogel from Figure 8A. Detail from Figure 9A is shown in Figure 9B (magnification ×200.000), proving that the clusters are made of nanoparticles. The polymeric matrix is not homogenous, due to the fact that it has smaller nanoparticle aggregates embedded. The cross-section of the lyophilized hydrogel shows microsize pores, with chains of MMNPs clusters, which seem to be located especially on the pore walls. At higher magnifications (Figure 8D), it can be observed that there are also nano-size areas having the same fibrils with branches-like structures with incorporated MMNPs. A very interesting result of the lyophilized sample of both PAA-MMNPs 50:50 and PAA-MMNPs 90:10 (Figure 8C,D and Figure 9C,D) showed less local MMNPs agglomeration with respect to un-lyophilized samples. This behavior can be explained by the lyophilization procedure. During the process, the polymer matrix between MMNPs swells and the space grows between them. Furthermore, the sublimation phenomenon leads to a rearrangement of the structure with the display of the MMNPs in the pore walls and fragmentation of the local agglomerates.

**Figure 9.** SEM micrographs of PAA-MMNPs 50:50 hydrogel (**A**,**B**).and lyophilized PAA-MMNPs 50:50 hydrogel (**C**,**D**); Energy dispersive X-ray (EDX) spectrum (**E**).

The EDXS spectrum (Figure 9E), acquired on a large area of the PAA-MMNPs 50:50 hydrogel surface and shows the presence in the sample of the elements Fe and O (from Fe3O4 NPs), C, N and Si (from 3-TPM and PAA).

#### 3.6.2. TEM Analysis

The morphology and nanostructural characteristics of magnetic nanoparticles (MNPs), modified magnetic nanoparticles (MMNPs) and of polyacrylamide modified magnetic nanoparticles (PAA-MMNPs) hydrogels were analyzed by TEM, selected area electron diffraction (SAED) and high resolution electron microscopy (HR-TEM).

3.6.3. TEM Analysis for Magnetite Nanoparticles (MNPS) and Modified Magnetite Nanoparticles (MMNPs)

Figure 10A–C are TEM micrographs of the MNPs. The bright field TEM image (Figure 10A) shows that the magnetic Fe3O4 nanoparticles are nearly spherical with diameters between 5 and 12 nm. The SAED pattern (inset of Figure 10A) of MNPs exhibits a typical face centered cubic (fcc) crystalline structure. The lattice spacing measured based on the diffractions rings is in accordance with the standard lattice spacing of Fe3O4 from the Powder Diffraction File (PDF) database (ICCD file no. 04-002-5683). The HRTEM images of MNPs (Figure 10B,C) clearly show the single crystallinity of Fe3O4 nanoparticles. The interplanar distances measured from the adjacent lattice fringes with Fast Fourier Transform (FFT) (inset of Figure 10B) are 2.53 Å, 2.10 Å and 1.62 Å, corresponding to (311), (400) and (511) crystalline family planes of Fe3O4 with crystalline structure, according to the PDF database. Nanocrystalline particles with diameter size between 5.7 and 8.6 nm are highlighted in Figure 10B. In the HRTEM image from Figure 10C it clearly shows the crystalline planes with 2.97 Å and 2.53 Å measured interplanar distances corresponding to crystalline family planes with (220) and (311) Miller indices.

**Figure 10.** TEM images on Fe3O4 nanoparticles (**A**–**C**) and modified F3O4 nanoparticles (**D**–**F**).

The TEM results of MMNPs are presented in Figure 10D,E,F. According to Figure 10D, modified Fe3O4 nanoparticles still keep the morphological properties of Fe3O4 nanoparticles. According to HRTEM images (Figure 10E,F), morphological and nanocrystalline properties of Fe3O4 nanoparticles are maintained, but it is clearly shown that the nanoscale Fe3O4 nanoparticles are modified in the MMNPs sample, because of the organic layer surrounding the Fe3O4 nanoparticles (highlighted in Figure 10F). The tailoring of magnetite nanoparticles by chemically functionalization with silane 3-TPM revealed by physico-chemical X-photoelectron spectroscopy is also sustained by the morphological characterization by TEM. The high magnification Figure 10E,F exhibits a less ordered organic layer consisted by silane 3-TPM, which addresses the magnetite nanoparticles. However, the surrounding organic layer displayed a specific order and arrangement structure, which will be further discussed. The Figure 10D revealed an overview result with a considering functionalization of the whole nanoparticles and not as an isolated modification. Thus, the magnetic nanoparticles were tailored with double bonds by the presence of the silane structure (Figure 1).

#### 3.6.4. TEM Analysis of Polyacrylamide-MMNPs Nanocomposite Hydrogels

The bright field TEM (BF-TEM) images from Figure 11A, 10D and 10G are results from PAA-MMNPs 90:10, PAA-MMNPs 70:30 and PAA-MMNPs 50:50 samples. These images show that all of the hydrogels have a similar morphology and nanostructure. The overview images (Figure 11A,D,G) revealed an expected decreasing of polymer matrix area with the increasing of MMNPs amount. All of the samples have isolated and local agglomerated magnetic Fe3O4 nanoparticles embedded within a polymer matrix. By comparing the BF-TEM images from PAA-MMNPs (Figure 11A,D,G) with the BF-TEM images from MNPs (Figure 10A) and MMNPs (Figure 10D), it can be concluded that the shape and the dimensions of the embedded nanoparticles are kept in the same range. The SAED image (inset of Figure 11G) shows that the PAA-MMNPs hydrogels contain similar Fe3O4 nanoparticles, well crystallized, with the same lattice spacing measured on the SAED image from MNPs (inset of Figure 10A). The diffraction of the matrix was not observed in the SAED image (inset of Figure 11G), which is probably because the organic layer and the PAA matrix are not highly ordered and are displaying short ordering range. In order to observe the detailed structure of PAA-MMNPs hydrogels, HRTEM was employed. Figure 11B,C (from PAA-MMNPs: 90-10), Figure 11E,F (From PAA-MMNPs: 70-30) and Figure 11H,I (from PAA-MMNPs: 50-50) show that the nanoparticles are embedded within a polymer matrix with amorphous structure. The nanoparticles have a round shape with diameters between 5 and 14 nm. The MMNPs are well integrated into polymer matrix revealing a clear interaction between the two phases. The nature of the interaction was revealed by the HR-TEM images, which rarely highlighted the specific order and arrangement structure of the organic layer from MMNPs. This result can be explained by the surrounding organic layer being in a chemical reaction with the acrylamide monomer by the consumption of the silane double bonds. Thus, the MMNPs act as an inorganic cross-linker by becoming generators of bridges between polymeric chains and development of a hybrid network (Figure 2). Also, the HRTEM results show that the nanoparticles are nanocrystals, disclosing the crystalline planes (220) and (311) of magnetite with 2.97 Å and 2.53 Å, respectively, which are characteristic interplanar distances. Furthermore, the HRTEM images also reveal a short ordering range in the matrix besides the amorphous phase, highlighted by squares (Figure 11F for PAA-MMNPs: 70-30 and Figure 11I for PAA-MMNPs: 50-50), which shows the structural arrangement of the polymer macromolecular chains compared with inorganic ordered magnetite nanoparticles.

**Figure 11.** Bright field (BF)-TEM and HR-TEM images on PAA-MMNPs 90:10 (**A**–**C**); PAA-MMNPs 70:30 (**D**–**F**) and on PAA-MMNPs 50:50 (**G**–**I**).

#### **4. Conclusions**

This study provides a comprehensive approach in the wide field of polymer nanocomposite materials. A new hybrid polymer network was successfully developed by double bond modified magnetic nanoparticles, using polyacrylamide as the crosslinked network structure, thereby overcoming the limitation of traditional organically crosslinkers. Functionalization of magnetic nanoparticles with the double bond was monitored by physico-chemical investigations. The details of the microarchitecture were shown by modern morphological characterization techniques, highlighting the nature of the interaction between the organic and inorganic phases. Furthermore, the obtained nanocomposite hydrogels may have an efficient applicability in the soft tissue engineering field, in the form of repairing scaffolds, due to their mechanical and magnetization behavior that can stimulate tissue regeneration.

**Author Contributions:** Formal analysis, E.T., V.-A.S., B.S.V. and C.-M.D.; Investigation, E.T. and I.-C.R.; Methodology, C.Z.; Project administration, C.Z.; Validation, C.Z.; Writing—original draft, E.T. and I.-C.R.; Writing—review and editing, C.Z. and E.A.

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

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

#### **References**


© 2019 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/).

#### **Zoran Cenev 1, Malte Bartenwerfer 2,\*, Waldemar Klauser 2, Ville Jokinen 3, Sergej Fatikow <sup>2</sup> and Quan Zhou 1,\***


Received: 11 September 2019; Accepted: 15 October 2019; Published: 19 October 2019

**Abstract:**The focused ion beam (FIB) has proven to be an extremely powerful tool for the nanometer-scale machining and patterning of nanostructures. In this work, we experimentally study the behavior of AISI 420 martensitic stainless steel when bombarded by Ga<sup>+</sup> ions in a FIB system. The results show the formation of nanometer sized spiky structures. Utilizing the nanospiking effect, we fabricated a single-tip needle with a measured 15.15 nanometer curvature radius and a microneedle with a nanometer sized spiky surface. The nanospikes can be made straight or angled, depending on the incident angle between the sample and the beam. We also show that the nanospiking effect is present in ferritic AISI 430 stainless steel. The weak occurrence of the nanospiking effect in between nano-rough regions (nano-cliffs) was also witnessed for austenitic AISI 316 and martensitic AISI 431 stainless steel samples.

**Keywords:** focused ion beam; nanospikes; martensite; stainless steel; gallium; bombardment; irradiation effects; sharp needle; incident angle

#### **1. Introduction**

The focused ion beam (FIB) technique has been established as a powerful tool for micro and nanoscale imaging [1], sputtering, deposition [2], 3D machining [3], and surface modifications [4]. When an incident ion comes into contact with a targeted material, the ion enters into a set of collisions (higher than normal thermal energies) with the target atoms, a process known as a collision cascade. Sputtering occurs when an incident ion comes into contact with a targeted surface and transfers its momentum to the host atoms. A host atom on the surface will absorb a part of the ion's kinetic energy. If the new energy state of the host surface atom is higher than the surface binding energy (SBE) of the targeted material, then the surface atom will be ejected as a sputtered particle [5]. A quantitative measure of sputtering is defined through sputtering yield, i.e., the number of atoms removed by an incident ion. The sputtering yield is affected by the material composition, angle of incidence, the crystal structure of the substrate, redeposition, scanning speed, temperature of the target, and surface contaminations [6].

The process which constrains the path of the ion in a crystalline solid is known as ion channeling [7]. Along low index directions in crystalline materials, ions may penetrate greater distances as compared to cascade collisions in amorphous materials. Since ion channeling has a direct impact on the ion penetration range, meaning the trajectory within the collision cascade, it also impacts the sputtering

yield. Variations on sputtering yield within a sample target cause roughening of the surface, which has been observed for aluminum [8], tungsten [9], and polycrystalline gold [10]. A nanometer sized spiky structure, an extreme form of nano-roughening, with distinct and visibly pronounced spikes, occurs during the anisotropic etching of single crystal (100) copper [8], tungsten [11], and 18 Cr-ODS (Oxide Dispersive Strengthened) steel [12]. Pyramidal and conical (faceted pyramid) micro/nanometer-sized structures have been observed much earlier on tin crystals [13], and monocrystalline [14] and polycrystalline copper [15] when irradiated by argon, as well as krypton ions [16]. The origins and stability of ion-bombarded copper surfaces have been heavily analyzed and discussed by Auciello and Kelly [17,18].

Here, we experimentally demonstrate the formation of nanospikes occurring on a martensitic AISI 420 stainless steel surface when bombarded with gallium ions. We also show that nanospikes can be made straight or angled depending on the incident angle of the FIB. To demonstrate potential applications, we FIB-treated an electrochemically etched stainless steel (AISI 420) tip to induce nanospiking and thus obtain a single tip nano-needle with a measured diameter of 15.15 nanometers. Additionally, we FIB-treated an electrochemically etched stainless steel (AISI 420) tip with micrometer sharpness to induce nanospikes. Finally, we also show that the nanospiking effect is present in ferritic AISI 430 stainless steel. The weak occurrence of the nanospiking effect in between nano-rough regions (nano-cliffs) was also witnessed for austenitic AISI 316 and martensitic AISI 431 stainless steel samples.

Future research should focus on using the single sharp nano-needle for creating localized magnetic fields, as in [19], or laser-induced electron emission, as in [20,21]. Due to the soft magnetic properties of the martensitic ASI 420 stainless steel, nanometer sized spiky magnetic tips could be applied in producing magnetic nano-devices, for example, magneto-gravitational traps [22,23]. Another line of research can focus on producing superhydrophobic/hydrophobic microneedles by subsequent fluoropolymer deposition (low adhesive polymer) to the surface of the microneedle with the nanometer sized spiky surface.

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

#### *2.1. Procedure of FIB Treatment of Martensitic, Austenitic and Ferritic Stainless Steel Plates*

The treatment of the martensitic AISI 420 (Fe-86,7/Cr13,0/C0,3) stainless steel sample (Goodfellow, Cambridge, UK) was carried out with a dual-beam high-resolution scanning electron/focused ion beam microscope, namely, the Lyra FEG (TESCAN, Brno, Czech Republic). A 0.5 cm2 piece was cut from the foil sheets. The piece was cleaned in an ultrasonic isopropanol bath, with a 10 min O2 plasma treatment under 40 kHz at 100 W in the plasma system, using the Femto instrument (Diener electronic GmbH and Co KG, Ebhausen, Germany). An area of 10 μm<sup>2</sup> was exposed to an ion dose of 19.4 nC/μm<sup>2</sup> at a 30 keV beam energy and emission current of 2 μA. The same treatment was applied for the other stainless steel samples, i.e., AISI316 (Fe/Cr18/Ni10/Mo3), AISI430 (Fe81/Cr17/Mn/Si/C/S/P) and AISI431 (Fe82/Cr16/Ni2), as were received from the supplier (Goodfellow, UK).

#### *2.2. Procedure of Fabrication of Martensitic Stainless Steel Needle with Nanometer Sharpness*

A one millimeter thick stainless steel AISI 420 wire (Goodfellow, Cambridge, UK) was thinned with up to micrometer sharpness, as previously reported in [19]. The etched needle was installed into the FIB-SEM dual beam system (Lyra FEG) and is shown in Figure S1a–c. Prior to FIB exposure, the needle was cleaned in an ultrasonic acetone bath and rinsed with isopropanol. A series of FIB exposures with a total ion dose of roughly equal to 2000 nC/μm2 at a 30 keV beam energy and emission current of 2 μA was applied in order to induce more spikes (Figure S1d). Once a prominent spike was obtained, it was isolated from further exposure, but the exposure was targeted towards removing the surrounding spikes and eventually providing the final result (Figure S1f).

#### *2.3. Procedure of Fabrication of Stainless Steel Microneedle with Nanospikes*

A one millimeter thick stainless steel AISI 420 wire (Goodfellow, USA) was installed into the collet of a milling 3-axis bridge router, as illustrated in Figure S2a. A face mill insert with four cutting edges was used for machining, where the wire would be thinned within a range of 0.4 and 0.7 mm thickness, with a length of about 3 mm. A thinned wire as such was mounted onto a holder of an in-house built electrochemical etching station, containing a 10% HCl bath, a computer-controlled voltage supply, and a motorized stage (further details can be found in [19]). The first step was electrochemical thinning, consisting of dipping the wire 3 mm into the HCl bath. The etching started when a voltage of 1V was supplied. Immediately after voltage application, the wire was pulled with a constant speed of 10 μm/s. The second etching step consisted of re-dipping the wire by 1 mm, with supply voltage of 1V and pulling the wire with a constant speed of 10 μm/s until the needle was completely out of the bath, as illustrated in Figure S2b. A sample micrograph of an etched needle can be seen in Figure S3a–c.

The electrochemically etched needle was installed into the FIB-SEM dual beam system, as illustrated in Figure S2c. Prior to FIB treatment, the machined/etched needles were cleaned in an ultrasonic acetone bath and rinsed with isopropanol. The needles were exposed to an ion dose of 10.6 nC/μm2 at a 30 keV beam energy and emission current of 2 μA (Figure S3d,e).

#### **3. Results**

#### *3.1. Nanospikes Formation on AISI 420 Martensitic Stainless Steels by FIB Treatment*

Figure 1a shows the surface morphology of an FIB-irradiated AISI 420 sample with gallium ions. Details of the sample preparation and FIB treatment settings are provided in Section 2.1. From the figure, it can be seen that the sharpness of the nanospikes is in the sub-micron range. One can also see that the nanospikes on the edge feature higher aspect ratios than the nanospikes in the middle of the trench.

To demonstrate the potential usability of the nanospiking effect, we have fabricated two different types of needles, i.e., an extremely sharp needle with radius of 15.15 nanometers (Figure 1b) and a micrometer-sized needle, featuring a nanometer sized spiky topology (Figure 1c). The fabrication procedure of both needles is similar, and they are explained in detail in Sections 2.2 and 2.3, respectively. One should note that the fabrication procedure for both needles includes a certain level of randomness, however, the sharpness of the nanospikes is very often in the low nanometer range (from several up to tens of nanometers).

**Figure 1.** Nanospiking effects on martensitic AISI 420 stainless steel. (**a**) Nanometer sized spiky surface of the martensitic AISI 420 stainless steel sample plate after FIB treatment with gallium ions with a dose of 19.4 nC/μm2. Fabrication results of a (**b**) sharp needle with nanometer resolution, the circle denotes fitting to the curvature of the tip. The original raw image without fitting is given in Figure S1f. (**c**) A micrometer-scale needle with nanospikes.

#### *3.2. Morphological Evolution of AISI 420 during FIB Treatment*

We also examined the morphological evolution of the martensitic AISI 420 in a step-by-step manner. Figure 2 shows the surface morphology evolution, and finally, the formation of the nanospikes. The samples were exposed to 30,000 scans overall (1000 scans correspond to an ion dose of 1435 nC/μm2). The trench dimension was 10 <sup>×</sup> <sup>10</sup> <sup>μ</sup>m2. The AISI 420 surface was untreated at the beginning (0 scans), and after the first 500 scans, the appearance of a few pits on the surface was noted.

**Figure 2.** Evolution of nanospikes at normal incidence as a function of FIB dose from 0 to 30,000 scans (1000 scans correspond to a gallium ion dose of 1435 nC/μm2). The spiky structure shift downwards along with increase of the FIB exposure. Scale bar in each image is 5 μm. Orange arrows denote the formation and size variation of the firstly formed nanospike.

The indentation of the initial pits increased with the increase of the number of scans (1000 to 5000 scans). At 7500 scans, the formation of the first nanospike (denoted with an orange arrow) was noticed. Further exposure of the earlier formed nanospikes causes their increase in sharpness, but also causes a decrease in height, as indicated by the orange arrows (10,000 to 30,000 scans). With further increase of the irradiation, new spikes started to form, and they could be found more within the central region of the trench, rather than on the edges. The nanospikes on the edges feature much higher aspect ratios than the ones in the central region. This difference in aspect ratios can be observed from scans 15,000 to 30,000.

#### *3.3. Energy-Dispersive X-ray Spectroscopy (EDX) and X-ray Photoelectron Spectroscopy (XPS) Analysis of FIB Irradiated AISI 420 Stainless Steel Alloy with Gallium Ions*

We have performed energy-dispersive X-ray spectroscopy (EDX) analysis on the whole gallium irradiated trench, a spot on a single nanospike, and a non-irradiated area (Figure S4). The only difference that the EDX results show is the presence of gallium in irradiated regions in comparison to non-irradiated regions. No significant change in the presence of the iron or chromium content within the AISI 420 sample before and after gallium irradiation was determined.

We also have performed X-ray photoelectron spectroscopy (XPS) measurements with the Kratos Axis Ultra ESCA system (Kratos Analytical Ltd., Manchester, UK), analyzing the gallium irradiated circular trench (diameter of ~35 μm) and the non-irradiated area (Figure S5). The XPS results show a reduction of iron (9.42% to 2.2% for XPS aperture of 27 μm) and chromium (0.97% to 0.44% for XPS aperture of 27 μm) between the non-irradiated and irradiated regions. Here, it could be that the gallium in the non-irradiated region was deposited during the gallium irradiation of the sample in the FIB system. The existence of high oxygen and carbon concentrations is due to exposure of the sample to ambient conditions.

#### *3.4. E*ff*ect from Variation of Incident Angle*

The effect from the variation of the incident angle has been studied by the different orientation of an AISI 420 stainless steel probes during gallium irradiation (Figure 3). At first, a probe was installed in a vertical position (Figure 3a) and it was subjected to gallium irradiation in FIB system. After a pre-defined FIB dose was delivered to the probe, the nanospikes formed in the direction of the beam. The same nanospikes formation occurred for a horizontally positioned probe (Figure 3b) and 40◦ inclined probe (Figure 3c). From these results, one can infer that nanospikes form regardless of the incident angle in this specific martensitic steel alloy.

**Figure 3.** Spiking phenomena of AISI 420 stainless steel sample probes during gallium irradiation with different incident angles: (**a**) Vertical position of the probe; (**b**) Horizontal position of the probe; and (**c**) at an incident angle of 40◦ to the probe. Here, (**i**) and (**ii**) illustratively depict the orientation of the sample, the gallium irradiated regions and the spiking result, respectively. Here, (**iii**) and (**iv**) show the results obtained before and after the FIB irradiation. Here, (**v**) are close-ins of (**iv**).

#### *3.5. Nanospiking Effect on Austenitic AISI 316, Ferritic AISI 430 and Martensitic AISI 431 during FIB Treatment*

Figure S6 shows a comparison of FIB irradiated stainless steel plates with gallium ions of other three different stainless steel types, i.e., austenitic (AISI 316), ferritic (AISI 430) and martensitic (AISI 431) stainless steel plates. Details of the sample preparation and FIB treatment are provided in Section 2.1. All samples have an anisotropic etching behavior. The austenitic AISI 316 stainless steel sample (a and d) shows a mix of inhomogeneous nano-rough regions and regions with nanospikes (Figure S6d). Nano-rough regions look like mountain range or cliffs, therefore the notation "nano-cliffs", for instance, see the orange arrows in Figure S6. The ferritic AISI 431 (c and e) features only a region with nanospikes (Figure S6e). The AISI 431 displays a presence of nanospikes, however, these are seldom scattered in between the nano-roughed bottom (Figure S6f).

#### **4. Discussion**

As can be seen from Figures 2 and 3, the nanospikes on the edges have much higher aspect ratios than the ones in the central region of the trench. When sputtering occurs at the edges, it nucleates the edges (formation of nanospikes), due to the presence of the gallium ions outside of the beam spot (the beam power features Gauss distribution). The inner part of the trench continues to be sputtered, but the part outside of the trench is slightly affected by the satellite gallium ions. The satellite gallium ions also cause sputtering, but at significantly reduced rates than the inner part of the trench. The sputtering continuity of the inner part shifts the nano-spiky structure downwards into the bulk, but this shift barely occurs at the edges. This discrepancy in the structural shift can explain why the nanospikes in the edges feature higher aspect ratios than the nanospikes in the central region of the trench.

Polycrystalline alloys such as the martensitic AISI 420 stainless steel (and the other FIB treated stainless steels) investigated in this work, besides the difference in material content, feature domains with different crystallographic orientations. The sputtering rates of neighboring domains may vary greatly, depending on the structural configuration of the grains and the orientation of the lattices in the particular domain with respect to the incident ion beam. However, we have shown that nanospikes occurred in the gallium bombarded AISI 420 sample, but not as much in the AISI 431 sample, where nano-cliffs were more dominant, although both samples are martensitic stainless steels with very similar crystalline structure [24,25].

We have performed EDX and XPS analysis to investigate the material content within the non-irradiated and the irradiated regions of the martensitic AISI 420 stainless steel alloy. The EDX results (penetration depth up to 10 μm) show the presence of iron, chromium, and gallium in the gallium-irradiated regions. The XPS results show a decrease of the iron and the chromium in the irradiated trench with respect to non-irradiated surface. The XPS results do not indicate any saturation of a single element on the very surface in the irradiated regions.

Other studies have demonstrated that ion channeling affects the sputtering yield in polycrystalline materials such as [4,8,9], therefore inducing nano-roughening on the treated surface. However, we are not sure whether the same explanation can be attributed to the formation of the nanospikes. The nanometer sized spiky formations are special forms of the nano-roughed surface, and the exact mechanism has been recently discussed by Prenitzer et al. [8] and Ran et al. [12], but also heavily researched much earlier [13–18]. Auciello [18] claims that the micro/nanometer scaled pyramidal structures form due to sputtering differences in (1) the presence of intrinsic and/or bombardment-induced sub-surface defects, (2) the evolution of pre-existing and/or bombardment-induced asperities of convex-up curvatures, and (3) the erosion of nuclei formed by migration of sputter-deposited foreign atoms on the substrate of the surface. The observed nanospike formations by Prenitzer et al. [8] were attributed to a wide range of sputtering conditions, whereas the most likely one may be the quality of initial target surface. Ran et al. [12] imaged 18 Cr-ODS steel nanospikes with transmission electron microscopy (TEM), showing that two different crystal orientations do exist in one nanospike with distinct two grains and a clear grain boundary. The report claims that nanospike formation is not induced by grain recrystallization and

regrowth during Ga<sup>+</sup> ion bombardment, but rather due to an interplay between a curvature-dependent sputtering and defect accumulation near the surface. Both reports address the importance of the initial surface topology. This interplay between a curvature-dependent sputtering and defect accumulation near the surface seems to be a valid argument and might be used to interpret our experimental observations, since the morphological variation of the targeted surfaces greatly impacts the dynamic competition of available atoms on the substrate, the atom evacuation due to sputtering, and the gathering of vacancies.

#### **5. Summary and Conclusions**

The nanospiking phenomenon has been previously reported for copper [8], tungsten [11], and 18Cr-ODS steel [12]. In this communication, we have shown that nanospikes are formed on martensitic AISI 420 stainless steel when treated with FIB. The nanospikes can be made straight or angled depending on the incident angle between the sample and the beam. We also showed fabrication of a <16 nanometer sharp single tip needle and a micrometer-sized sharp needle with nanospikes. The nanospiking effect occurs in ferritic AISI 430 stainless steel sample too. A weak occurrence of the nanospiking effect in between nano-rough regions (nano-cliffs) was also witnessed for the austenitic AISI 316 and martensitic AISI 431 stainless steel samples. Unlike the intermediate existence of the nano-pyramidal structures reported in [16], the nano-spiky structures reported here are stable and occur at different irradiation doses.

The nanospiking phenomenon in martensitic AISI 420 stainless steels has promising capacity for future research. The single sharp nano-needle has potential of being used for creating localizing magnetic fields, as in [19], or laser-induced electron emission as in [20,21]. A micrometer-scaled needle with nano-spiky topology could be utilized for making superhydrophobic needles, performing droplet manipulation on open hydrophobic and superhydrophobic surfaces, where needle-droplet adhesion is less than droplet-substrate adhesion, similar to as in [26,27]. Since the martensitic stainless steel has soft ferromagnetic properties, the Ga<sup>+</sup> ion bombardment process can be used for fabricating magnetic nanospikes, which might find application in the development of novel quantum devices, e.g., magneto-gravitational traps [22,23].

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2079-4991/9/10/1492/s1, Figure S1: Intermediate steps of the fabrication process of martensitic stainless steel AISI420 needle with nanometer sharpness. Figure S2: Illustration of the fabrication procedure of microneedle with nanospikes. Figure S3: Intermediate steps of the fabrication process of martensitic stainless steel AISI420 microneedle with nanospikes. Figure S4: Energy-dispersive X-ray spectroscopy (EDX) results of (a) the completely irradiated trench; (b) a spot on a single nanospike; (c) non-irradiated area. Figure S5: X-ray Photoelectron Spectroscopy (XPS) results of (a) a ~35 μm in diameter gallium irradiated trench; (b) non-irradiated area. i and ii denote measurement with XPS aperture of 27 and 55 μm, respectively. Figure S6: Gallium irradiation of austenitic AISI 316 (a,d), ferritic AISI 430 (b,e), and martensitic AISI 431 (c,f) stainless steel plates with a dose of 19.4 C/μm2. (a–c) Before and (d–f) after gallium irradiation.

**Author Contributions:** Z.C., W.K., and M.B. have jointly observed the nanospiking phenomena of AISI 420 martensitic stainless steel and fabricated the needles. Z.C. has electrochemically etched the martensitic AISI 420 stainless steel wires, performed the EDX and the XPS measurements and analysis. W.K. and M.B. have performed the FIB treatments and investigated the spiking effects on the other stainless steel types. Z.C. and V.J. have conceived the idea of fabrication of microscopic needles with nanospikes. Q.Z., S.F. and M.B. supervised the whole research throughout the whole duration and ensured credible conduction of experimental work. All authors wrote the paper.

**Funding:** This research work was supported by the Academy of Finland (projects: #304843, #295006, #297360) and German Academic Exchange Service (DAAD) (project: #57247327). The authors express their gratitude to Micronova Nanofabrication Center for providing laboratory facilities for microfabrication.

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

#### **References**

1. Gupta, J.; Harper, J.M.E.; IV, J.M.; Blauner, P.G.; Smith, D.A. Focused ion beam imaging of grain growth in copper thin films. *Appl. Phys. Lett.* **1992**, *61*, 663–665. [CrossRef]


© 2019 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/).
