**Achieving Secondary Dispersion of Modified Nanoparticles by Hot-Stretching to Enhance Dielectric and Mechanical Properties of Polyarylene Ether Nitrile Composites**

#### **Yong You, Ling Tu, Yajie Wang, Lifen Tong, Renbo Wei \* and Xiaobo Liu \***

Research Branch of Advanced Functional Materials, School of Materials and Energy, University of Electronic Science and Technology of China, Chengdu 610054, China

**\*** Correspondence: weirb10@uestc.edu.cn (R.W.); liuxb@uestc.edu.cn (X.L.); Tel.: +86-028-8320-7326 (X.L.)

Received: 5 June 2019; Accepted: 10 July 2019; Published: 12 July 2019

**Abstract:** Enhanced dielectric and mechanical properties of polyarylene ether nitrile (PEN) are obtained through secondary dispersion of polyaniline functionalized barium titanate (PANI-*f*-BT) by hot-stretching. PANI-*f*-BT nanoparticles with different PANI content are successfully prepared via in-situ aniline polymerization technology. The transmission electron microscopy (TEM), fourier transform infrared spectroscopy (FTIR), X-ray photoelectron spectroscopic instrument (XPS) and Thermogravimetric analysis (TGA) results confirm that the PANI layers uniformly enclose on the surface of BaTiO3 nanoparticles. These nanoparticles are used as functional fillers to compound with PEN (PEN/PANI-*f*-BT) for studying its effect on the mechanical and dielectric performance of the obtained composites. In addition, the nanocomposites are uniaxial hot-stretched by 50% and 100% at 280 ◦C to obtain the oriented nanocomposite films. The results exhibit that the PANI-*f*-BT nanoparticles present good compatibility and dispersion in the PEN matrix, and the hot-stretching endows the second dispersion of PANI-*f*-BT in PEN resulting in enhanced mechanical properties, crystallinity and permittivity-temperature stability of the nanocomposites. The excellent performances of the nanocomposites indicate that a new approach for preparing high-temperature-resistant dielectric films is provided.

**Keywords:** nanocomposites; surface-functionalization; secondary dispersion; hot-stretching

#### **1. Introduction**

With the increasing requirements of modern microelectronic components, the miniaturized and flexible dielectric materials are attracting more and more attention for various applications [1,2]. However, up to now, a single component material has been unable to meet these demands. Although the traditional inorganic ceramic dielectrics are widely used owing to their high dielectric constant, their inherent characteristics of heavy weight, difficult processing and brittleness fail to meet the current practical application [3,4]. In comparison, polymeric materials have exhibited the advantages of being lightweight and flexible, but their low dielectric constant also limits their application to a great extent [5–7]. Therefore, combining the two-component materials is an effective way to overcome these limitations [8–10].

In recent years, polymer-based nanocomposites have proved to be an important dielectric material by virtue of the high dielectric permittivity, flexibility and excellent thermal stability for widely using as dielectrics in the electronic system [11–15]. It is mainly because the nanocomposites can absorb the dominants of polymer matrix and inorganic nanofillers. Nevertheless, the prerequisite for the nanocomposites demonstrating excellent properties is to realize good compatibility between nanoparticles and polymer matrix [16,17]. In general, modifying the micro-interface of the nanoparticles

can greatly improve the compatibility of nanoparticles with matrix, which can also regulate the dielectric permittivity and thermal stability of composites [18–20]. Thus, it is important to design and fabricate surface functionalized nanoparticles on the basis of maintaining the properties of nanoparticles while improving the compatibility. Polyaniline (PANI), as a conducting polymer, has been widely used as the filler to compound with polymer matrix [21] or as the surface agent to modify the nanoparticles [22,23] by virtue of its excellent conductivity after doping.

Although surface-functionalized nanofillers can effectively improve their compatibility with the matrix enhancing the dispersion of nanofillers in the matrix, they will inevitably agglomerate at a high filler level, resulting in reduced mechanical properties and permittivity-temperature stability. Therefore, it is crucial to find a technology that can achieve secondary dispersion of fillers and improve the overall performance of composites [24]. Uniaxial hot-stretching, a method to achieve high orientation of polymer materials under the action of external force, realizes a secondary uniform dispersion of the fillers along the orientation direction. In addition, the hot-stretching technology can effectively promote the regularity of polymer molecular chains and improve their crystallinity [23–28].

In this paper, novel PANI-*functionalized*-nanoparticles of different polymer content via in-situ polymerization technology are fabricated and characterized in detail. Also, these surface functionalized nanoparticles are used as functional fillers to promote the performances of polyarylene ether nitrile (PEN). In addition, the PEN-based nanocomposites are uniaxial hot-stretched by 50% and 100% at 280 ◦C. The corresponding properties of the oriented nanocomposite films are investigated in detail.

#### **2. Experimental**

#### *2.1. Materials*

BaTiO3 (~60 nm, cubic) was bought from TPL Co., Dallas, Texas, USA. Potassium carbonate (K2CO3), 2, 6-dichlorobenzonitrile (DCBN), biphenol (BP), ammonium persulfate ((NH4)2S2O8), aniline (C6H7N), hydrochloric acid and alcohol were supplied by Chengdu KeLong chemicals, Chengdu, China. *N*-methyl-2-pyrrolodone (NMP) was bought from Chengdu Changzheng chemicals, Chengdu, China.

#### *2.2. Preparation of PEN*

PEN was prepared by 2, 6-dichlorobenzonitrile and biphenol in our laboratory through the previously reported method [28].

#### *2.3. Preparation of PANI-f-BT Nanoparticles*

The PANI-*f*-BT nanoparticles were fabricated by in-situ aniline polymerization method [29]. The specific steps are shown in Figure 1a. Firstly, BaTiO3 (~60 nm, 1.0 g) was added into 100 mL deionized water and ultrasonicated for 1 h. Then the dispersion was cooled to 0–5 ◦C with an ice bath (step A). At the same time, a certain amount of aniline dissolved in 50 mL HCl (0.1 M) was also cooled in another ice bath. Next, the aniline solution was quickly dripped into the BaTiO3 dispersion under nitrogen atmosphere (step B). After stirring for 1 h, pre-cooled ammonium persulfate dissolving in deionized water was added into the mixture of BaTiO3/aniline for oxidative polymerization for 18 h (step C and D). Finally, the PANI-*f*-BT nanoparticle was obtained through filtration and drying (step E). In addition, the preparation diagram of PANI-*f*-BT nanoparticles is shown in Figure 1b. In this system, the molar ratio of ammonium persulfate to aniline is controlled to be 1.2:1, and the amounts of aniline are 0.1, 0.2 and 0.3 mL for PANI-*f*-BT-a, PANI-*f*-BT-b and PANI-*f*-BT-c, respectively.

**Figure 1.** Experimental steps (**a**) and schematic diagram (**b**) of the polyaniline functionalized barium titanate (PANI-*f*-BT) nanoparticles.

#### *2.4. Preparation of Nanocomposites*

PEN-based nanocomposite films with 40 wt% pure BT and PANI-*f*-BT nanoparticles were prepared by solution casting method [30], which were named PEN/BT, PEN/PANI-*f*-BT-a, PEN/PANI-*f*-BT-b, PEN/PANI-*f*-BT-c, respectively.

#### *2.5. Preparation of Oriented Nanocomposites by Hot-Stretching*

The orientation process of nanocomposite film was carried out in a 280 ◦C oven by uniaxial hot-stretching method according to the previously reported paper [25]. All the nanocomposite films were stretched by 50% and 100%, respectively. The detailed steps were as follows: first of all, the two ends of the films (10 mm × 150 mm) were fixed by clamps, where the distance was controlled to be 100 mm; next, one clamp was suspended at the top of the oven, and the other one was hung with a 200 g balancing weight. The distance between the bottom of the oven and the balancing weight was adjusted to be 50 mm and 100 mm, corresponding to the stretching ratios of 50% and 100%. After the films were stretched to the required length, they were quickly removed and quenched in cold water. For comparison, the un-stretched nanocomposite films were also treated at 280 ◦C for the same time.

#### *2.6. Characterization*

The chemical structure of PANI-*f*-BT nanoparticles was characterized on a fourier transform infrared spectroscopy (FTIR, 8400S, Shimadzu, Japan) in the transmission mode between 4000 and 500 cm−<sup>1</sup> by incorporating PANI-f-BT in the KBr. The elemental analysis was tested on an X-ray photoelectron spectroscopic instrument (XPS, ESCA 2000, VG Microtech, UK) using a monochromic Al Kα (*hv* = 1486.6 eV) X-ray source. The micro-structures of PANI-f-BT were carried out on the transmission electron microscopy (TEM, JEM-2100F, JEOL, Japan) at 200 kV by dispersing PANI-f-BT on the copper network. The micro-structures of the PANI-f-BT nanocomposites were also obtained using a scanning electron microscopy (SEM, 6490LV, JSM, Japan) at 20 kV by sputtering gold on the fractured surface of PANI-f-BT. The crystalline structure of PANI-f-BT was characterized by X-ray diffractometer (XRD, RINT2400, Rigaku, Japan) with Cu Kα radiation. Thermal properties of samples were tested under N2 atmosphere by differential scanning calorimetry (DSC, Q100, TA Instruments, New Castle, USA) from 40 ◦C to 380 ◦C with a heating rate of 10 ◦C/min and Thermogravimetric analysis (TGA, Q50, TA Instruments, New Castle, USA) from 50 ◦C to 800 ◦C with a heating rate of 20 ◦C/min. The mechanical properties of the compounds were measured by Universal Testing Machine (SANS CMT6104, China) with a stretching speed of 5 mm/min. All the films were cut into standard strips (10 mm × 150 mm), and the reported data is the average value obtained by testing five samples. The Dielectric properties of the polymeric compounds were tested on a precision LCR meter (TH 2819A, Tonghui, China). The films were cut into regular pieces (10 mm × 10 mm) and both sides were coated with the conductive silver paste to form a plate capacitor.

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

#### *3.1. Microstructure and Morphology of PANI-f-BT*

In this work, enhanced dielectric and mechanical properties of PEN are obtained by hot-stretching. PANI-*f*-BT nanoparticles with different polymer content are fabricated via in-situ aniline polymerization technology by controlling the content of aniline, and then incorporated into PEN matrix by ultrasonication achieving the first dispersion of the fillers in PEN matrix. In addition, the PEN-based nanocomposites are uniaxial hot-stretched by 50% and 100% at 280 ◦C, obtaining the second dispersion of the fillers. Resulting from the excellent compatibility between PANI-*f*-BT and PEN and the second dispersion of PANI-*f*-BT in PEN by hot-stretching, the obtained composites demonstrate enhanced crystallinity, mechanical and dielectric properties.

In order to characterize the microstructure of the functionalized nanoparticles, BT and PANI-*f*-BT are characterized by TEM, as shown in Figure 2a,b. It can be seen from Figure 2a that BT shows a smooth surface without distinct interface at its periphery. In comparison, Figure 2b shows different interfaces at the edges of PANI-*f*-BT. A layer of polymer corona is uniformly coated around the BT, indicating that the surface of BT is wrapped with a compact polyaniline layer [31]. The chemical structure of PANI-*f*-BT is characterized by FTIR (Figure 2c). It is clear that the strong band at 567 cm−<sup>1</sup> spectra of pristine BT and PANI-*f*-BT is corresponding to the vibration of Ti-O [32]. Besides, obvious absorption bands at 1586 and 1497 cm−<sup>1</sup> can be found on the FTIR spectrum of PANI-*f*-BT, which belong to the skeleton vibration of benzene rings from polyaniline [33]. Compared to BT, the additional characteristic absorption peaks at 3428 and 1189 cm−<sup>1</sup> from the spectrum of PANI-*f*-BT are the absorption peaks of N-H and Ar-N vibration, which proves that polyaniline exists in the PANI-*f*-BT [33].

**Figure 2.** Transmission electron microscopy (TEM) images of (**a**) barium titanate (BT) and (**b**) polyaniline functionalized barium titanate (PANI-*f*-BT-b); (**c**) Fourier transform infrared spectroscopy (FTIR) spectrum of BT and PANI-*f*-BT-b; (**d**) Thermogravimetric analysis (TGA) curves of the nanofillers.

In addition, the chemical composition of the obtained PANI-*f*-BT is further characterized by XPS measurement (Figure 3). As shown in Figure 3a, it is obvious that the Ba3d, Ba4d, Ba4p, Ti2p and O1s peaks can be observed on the full scanned XPS spectrum of PANI-*f*-BT indicating the existence of BT. In addition, two peaks at 286 eV and 402 eV on the spectrum of PANI-*f*-BT are corresponding to C1s and N1s from polyaniline. Ba3d spectrum of PANI-*f*-BT presents two peaks at 779.6 and 794.9 eV which belong to Ba3d5/<sup>2</sup> and Ba3d3/2, respectively (Figure 3b) [34]. The Ti2p spectrum of PANI-*f*-BT also shows two peaks at 457.9 eV (Ti2p3/2) and 463.9 eV (Ti2p1/2) (Figure 3c) [32]. What is more, the N1s spectrum of PANI-*f*-BT can be differentiated into three peaks: 398.1 eV (–N=), 398.9 eV (–NH–) and 400.1 eV (N+), respectively [35] as shown in Figure 3d.

**Figure 3.** The X-ray photoelectron spectroscopic instrument (XPS) spectrum of PANI-*f*-BT-b: (**a**) full scanned spectrum; (**b**) Ba3d; (**c**) Ti2p; (**d**) N1s.

Beside the characterization of PANI-*f*-BT, the contents of polyaniline in PANI-*f*-BT are determined by TGA test. As shown in Figure 2d, it is clear that BT nanoparticle does not demonstrate any weight loss, even when heated to 800 ◦C. In comparison, the residue of PANI-*f*-BT-a, PANI-*f*-BT-b and PANI-*f*-BT-c is 83.4%, 79.4% and 65.5% at 800 ◦C, respectively. The decrement of the residue indicates the existence of PANI in PANI-*f*-BT. Simultaneously, the less residue of the PANI-*f*-BT nanoparticles, the more PANI in the PANI-*f*-BT nanoparticles. Therefore, all these results including TEM, FTIR, XPS and TGA suggest that PANI is successfully grown on the surface of BT after the in-situ polymerization procedure.

#### *3.2. Morphology of PEN*/*PANI-f-BT Composites*

After the fabrication and characterization of the PANI-*f*-BT nanoparticles, they are introduced into the PEN matrix to prepare the PEN/PANI-*f*-BT composites. The miscibility between PANI-*f*-BT and PEN matrix, which is one of the most important factors affecting the properties of the nanocomposites, is firstly investigated by SEM measurement. Figure 4a shows the microstructures of PEN/BT nanocomposites, which exhibit a poor interfacial adhesion. In addition, a large number of BT nanoparticles are observed at the cross-section of the PEN matrix with serious spherical agglomeration. This phenomenon is mainly due to the high content of BT as well as the poor miscibility between PEN and BT [36]. Contrarily, the PANI-*f*-BT nanoparticles reveal a homogeneous dispersion in PEN matrix without agglomeration (Figure 4b). This result is mainly caused by that the PANI layer on PANI-*f*-BT nanoparticles which improves compatibility with PEN. Therefore, the modification of BT with PANI can effectively improve the compatibility between BT and PEN [36].

**Figure 4.** Cross-sectional scanning electron microscopy (SEM) images of (**a**) PEN/BT; (**b**) PEN/PANI-*f*-BT-b; (**c**) PEN/PANI-*f*-BT-b hot-stretched by 50%; (**d**) PEN/PANI-*f*-BT-b hot-stretched by 100%.

To achieve the secondary dispersion of PANI-*f*-BT in PEN, the PEN/PANI-*f*-BT composites are hot-stretched in a home-made oven. Figure 4c,d are the cross-sectional SEM images of PEN/PANI-*f*-BT-b after hot-stretching at a stretching ratio of 50% and 100%, from which obvious orientation of the sample caused by the directional arrangement of polymer molecular chains under the action of external forces is observed. After hot-stretching, the PANI-*f*-BT nanoparticles are secondarily dispersed in the PEN matrix along the orientation direction. A schematic model of the evolution process of PANI-*f*-BT in the polymer matrix during uniaxial stretching is presented in Figure 5. Before the hot-stretching, the spherical PANI-*f*-BT is isotropically dispersed in the PEN matrix. With the commencement of the hot-stretching, the PANI-*f*-BT nanoparticles rearrange along the orientation direction of the stretching. Finally, the enhanced dispersion of PANI-*f*-BT in PEN matrix is obtained after the hot-stretching. Combining the improved compatibility between PANI-*f*-BT and PEN and the secondarily dispersion of PANI-*f*-BT in PEN matrix induced by hot-stretching, enhanced properties of the PEN/PANI-*f*-BT composites can be imaged.

**Figure 5.** The theoretical model of the evolution process of inner network during uniaxial stretching: (**a**) the original composite film, the composite film hot-stretched by (**b**) 50% and (**c**) 100%.

#### *3.3. Thermal Properties and Crystallization of PEN*/*PANI-f-BT Composites*

As a crystalline polymer, the crystallization behavior of PEN is another important factor affecting the properties of the nanocomposites. The crystallization behavior of PEN/PANI-*f*-BT composites is studied by DSC and XRD. Figure 6 shows the DSC curves of the PEN/BT (Figure 6a) and PEN/PANI-*f*-BT (Figure 6b–d) nanocomposites before and after hot-stretching. It clearly shows that the melting peaks of all nanocomposites are not observed before hot-stretching, while they appear after hot-stretching. The melting enthalpy (Δ*H*m) of PEN/PANI-*f*-BT-b at the stretching ratios of 0%, 50% and 100% is 0, 6.4 and 9.1 J/g, respectively (Table 1). With the increase of stretching ratios from 50% to 100%, the Δ*H*<sup>m</sup> of these composites increases gradually, meaning that the crystallinity of the nanocomposites increases [25,37]. This would be due to the rearrangement of PEN molecular chains during the hot-stretching resulting the transitions of the samples from amorphous regions to crystalline regions and from irregular crystals to regular crystals [28]. Besides, the half peak width of the melting peaks also shows the crystalline information of the samples. Generally, the perfect crystals exhibit smaller half peak width than the imperfect crystals. The half peak width is 9.34, 6.55, 5.68 and 5.71 ◦C for PEN/BT, PEN/PANI-*f*-BT-a, PEN/PANI-*f*-BT-b and PEN/PANI-*f*-BT-c respectively, at the stretching ratio of 100%. The widest half peak width of PEN/BT at 100% stretching ratio is mainly owing to the worst compatibility and dispersion of BT in the PEN matrix as confirmed by the SEM observation (Figure 4a). In addition, DSC curves also demonstrate that the glass transition (*T*g) of the all nanocomposites increases slightly during the hot-stretching process. For instance, the *T*g of PEN/PANI-*f*-BT-b increases from 218.6 ◦C to 221.7 ◦C as stretching ratios increases from 0% to 100% (Figure 6c and Table 1). This result can also be explained by the arrangement of the PEN chains after hot-stretching, leading to the harder movement of them. Moreover, the increase of crystallinity further limits the movement of the chain segments [38]. Furthermore, the DSC curves for both cooling and heating are shown in Figure S1, as can be seen from the figure that during the first cooling scan and the second heating scan, the *T*gs obtained from the curves are lower than the one obtained from the first heating scan due to the supercooling effect [39]. What is more, the melting point disappears during the first cooling scan and the second heating scan due to the semi-crystalline property of the polymers and slow crystalline rate of the polymers [40]. In addition, we also characterize the crystal structure state of the samples after treatment at 200 ◦C, which is shown in Figure S2. It can be clearly seen from the figure that the PEN/PANI-*f*-BT-b nanocomposites with a 50% and 100% stretching ratio still show an obvious melting peak in the second heating curve after treating them at 200 ◦C for 10 min. It is indicated that the crystal structure of the PEN/PANI-*f*-BT-b nanocomposites is thermodynamic stable when used at 200 ◦C. More importantly, as shown in Figure S2d, it is clear that the melting peak is also maintained in the second heating curve of the PEN/PANI-*f*-BT-b nanocomposites with a 100% stretching ratio after treating at 300 ◦C for 10 min. Therefore, all these results confirm that the crystal structure of the hot-stretched PEN/PANI-*f*-BT nanocomposite films can still maintain good thermal stability during the practical application (<200 ◦C).


**Table 1.** Thermal and mechanical properties of nanocomposites at different stretching ratios.

**Figure 6.** The differential scanning calorimetry (DSC) curves of nanocomposites with different stretching ratios: (**a**) PEN/BT; (**b**) PEN/PANI-*f*-BT-a; (**c**) PEN/PANI-*f*-BT-b; (**d**) PEN/PANI-*f*-BT-c.

XRD is usually employed to study the crystals and crystallinity of samples. Figure 7 typically shows the XRD patterns of PEN/PANI-*f*-BT-b at different stretching ratios. As shown in the figure, the diffraction peaks at around 32◦, 38◦ and 45◦ are observed from all three samples which are coming from the diffractions of (110), (111) and (200) of BT. As for PEN, no crystalline peak is observed before hot-stretching. In comparison, two diffraction peaks at 17◦ and 23◦ which are coming from the diffractions of (111) and (112) of PEN are observed after the hot-stretching. The crystallinities of PEN/PANI-*f*-BT-b with the stretching ratios from 0% to 100% are 0%, 11.2% and 16.4%, which are calculated from wide-angle XRD spectrogram by using Jade 6 software [41].

**Figure 7.** The wide-angle XRD patterns of PEN/PANI-*f*-BT-b at different stretching ratios.

#### *3.4. Mechanical Properties of PEN*/*PANI-f-BT Composites*

Resulting from the improved compatibility and crystalline of the composite, their enhanced properties are further investigated. Tensile strength and tensile modulus, two most important mechanical properties of the high-performance engineering plastics, are typically studied. The tensile properties of the PEN-based nanocomposites at different stretching ratios are shown in Figure 8. The tensile strength of PEN/BT nanocomposites is 67.2, 90.3 and 119.7 MPa, at the stretching ratio 0%, 50% and 100%, respectively. As expected, the tensile strengths of all nanocomposites demonstrate a significant increase after modifying the surface of BT and hot-stretching, which are shown in Figure 8a. Moreover, tensile strength of PEN/PANI-*f*-BT-b nanocomposite is 83.8 MPa without stretching. What is more, it increases to 161.1 MPa when the stretching ratio is 100%, with an increment of 90%. The detailed mechanical data of the nanocomposites are listed in Table 1. It is obvious that the tensile strengths of PEN/PANI-*f*-BT nanocomposites are higher than that of PEN/BT nanocomposites, which are attributed to the better dispersion and compatibility between PEN and PANI-*f*-BT. In addition, it can be concluded that a substantial increase in mechanical property of the nanocomposites is contributed by the high orientation of the molecular chains and the newly formed oriented crystals after hot-stretching process [25,41]. The tensile modulus of the nanocomposites exhibits a similar tendency as that of tensile strength, which results from the same reasons as mentioned above [25].

**Figure 8.** The mechanical properties of nanocomposite films at different stretching ratios: (**a**) tensile strength and (**b**) tensile modulus.

#### *3.5. Dielectric Properties of PEN*/*PANI-f-BT Composites*

PEN has shown prospective application in film capacitors and other electronic devices. However, the permittivity of PEN is relatively low (~4.0 at 1 kHz) which failed to meet the high permittivity requirement for film capacitors. Herein, PANI-*f*-BT nanoparticles are used as a filler to mix with PEN for preparing PEN/PANI-*f*-BT nanocomposites. The dependence between dielectric properties of the obtained nanocomposites and varying frequency (100 Hz to 1 MHz) is shown in Figure 9a. Compared with PEN/BT, the dielectric constants of all PEN/PANI-*f*-BT nanocomposites are slightly lower than those of PEN/BT. The decreasing of dielectric constant of PEN/PANI-*f*-BT nanocomposites is caused by the PANI layer on the surface of BT nanoparticles which will hinder the charge movements from BT to PEN matrix. This usually leads to a decrease in interfacial polarization of the system [42]. In addition, it is clear that the permittivity of PEN/PANI-*f*-BT nanocomposites is more stable than that of PEN/BT nanocomposite with the change of frequency (100 Hz to 1 MHz). This is due to that the introduction of organic shell layer which enhances the compatibility between nanofillers and PEN matrix which depresses the Maxwell-Wagner polarization [30]. The dielectric loss of the PEN/PANI-*f*-BT nanocomposites is shown in Figure 9b. Although the content of nanofillers is up to 40 wt%, the dielectric loss of PEN/PANI-*f*-BT nanocomposites is still below 0.028 (1 kHz). This phenomenon also resulted from the improved compatibility between PANI-*f*-BT and PEN [30]. What is more, the dielectric loss of the PEN/PANI-*f*-BT nanocomposites demonstrates a similar trend with the changing of frequency and fillers as that of their permittivity. Furthermore, the electrical conductivity of the studied samples is shown in Figure S3. It can be seen from the figure that the electrical conductivity of PEN/PANI-*f*-BT nanocomposites is almost the same (10−<sup>10</sup> S cm<sup>−</sup>1) at 100 Hz (Figure S3a), which indicates that all the nanocomposite films are insulators.

**Figure 9.** (**a**) Dielectric constant and (**b**) dielectric loss of the nanocomposites with the changing of frequency.

Compared with the commonly used biaxially oriented polypropylene (BOPP) and Poly(vinylidene fluoride) (PVDF), PEN, which is a kind of special thermoplastic engineering material, has demonstrated its potential application as high temperature dielectrics. Therefore, the permittivity of all samples at different temperature is further researched. As shown in Figure 10a, the dielectric constant of nanocomposites is measured at 1 kHz in the range of 25 to 250 ◦C. It is clear that the dielectric constant of all nanocomposites is stable before their *T*g, and it increases hastily and obviously when the temperature exceeds their *T*g. This is due to the molecular chains of the PEN that are frozen at a temperature lower than its *T*g. However, as the temperature increases, the macromolecular chains are thawed which strengthens the mobility of electrons and enhances polarization in the system [43]. As a result, the *T*g of the polymeric dielectrics can be obtained from their permittivity-temperature curves. As shown in Figure 10a, the *T*g of PEN/PANI-*f*-BT-c obtained from the permittivity-temperature curve is 218 ◦C which is the same as that obtained from its DSC curve. The *T*<sup>g</sup> of the other composites is also around 218 ◦C. The high *T*<sup>g</sup> of the composites ensures the potential application of them at high temperature. Beside the qualitative description of the stable permittivity of the composites before

their *T*g, the quantitative value of dielectric constant with the change of temperature (temperature coefficients of dielectric constant) is calculated according to Equation (1) [44]:

$$
\pi\_{\varepsilon} = \frac{\varepsilon\_{\text{T2}} - \varepsilon\_{\text{T1}}}{\varepsilon\_{\text{T0}}(T\_2 - T\_1)} \tag{1}
$$

where τ<sup>ε</sup> is the temperature coefficient of dielectric constant, εT0 is the dielectric constant of room temperature, εT1 is the dielectric constant of initial temperature and εT2 is the dielectric constant of the final temperature. *T*<sup>1</sup> and *T*<sup>2</sup> are the initial and final temperatures, respectively. According to Equation (1), the calculated results of the temperature coefficients of dielectric constant of all nanocomposites are shown in Figure 10b. The temperature coefficients of dielectric constant of the composites are lower than 5 <sup>×</sup> <sup>10</sup>−<sup>4</sup> ◦C−<sup>1</sup> within the temperature range from 25 to 100 ◦C. They are still lower than <sup>3</sup> <sup>×</sup> <sup>10</sup>−<sup>3</sup> ◦C−<sup>1</sup> even though in the temperature range from 25 to 200 ◦C, indicating that they are extraordinarily stable, even at temperature up to 200 ◦C. In addition, it is notable that the temperature coefficients of dielectric constant of the composites decrease with the increase of the PANI content in the PANI-*f*-BT. This result is also largely due to the introduction of organic shell layer can improve the compatibility between nanoparticles and PEN matrix, which can effectively reduce the charge accumulation and interfacial polarization between inorganic nanoparticles and matrix [30].

**Figure 10.** (**a**) Permittivity-temperature dependences and (**b**) temperature coefficients of dielectric constant of the nanocomposites; (**c**) permittivity-temperature dependences and (**d**) temperature coefficients of dielectric constant of PEN/PANI-*f*-BT-b at different stretching ratios.

Moreover, to make a clear view of the properties of the PEN/PANI-*f*-BT nanocomposites, a comparison of dielectric constant, dielectric loss and working temperature at room temperature and 1 kHz of reported polymer-based composites are summarized in Table 2. As can be seen from the table, the dielectric constant and loss of PEN/PANI-*f*-BT are 14 and 0.025 when the content of PANI-*f*-BT nanoparticles is 40 wt%. Although the dielectric constant is lower than that of the most widely used PVDF based composites, it is comparable to the P(VDF-HFP)/BT-OPA, PES/BT-CuPc and PAEN/BT@CPAEN system. In addition, the dielectric loss of PEN/PANI-*f*-BT nanocomposites maintains at a relatively low level and is lower than that of most PVDF based composites. More importantly, the inherent low temperature resistance of PVDF (<120 ◦C) will limit its application in high temperature environment. In comparison, in this work, the PEN/PANI-*f*-BT nanocomposites can be used as flexible dielectric films at around 200 ◦C, which provides a new approach for preparing high-temperature-resistant dielectric films.

**Table 2.** Dielectric constant at 1 kHz and 25 ◦C, dielectric loss at 1 kHz and 25 ◦C, working temperature of typical polymer-based composites.


The dielectric properties of the PEN/PANI-*f*-BT nanocomposites are further improved by hot-stretching. As shown in Figure 10c, the dielectric constant of PEN/PANI-*f*-BT-b is 14.0 at 1 kHz without stretching (at 25 ◦C). After hot-stretching, its dielectric constant increases to 15.9 and 18.7 when the stretching ratio is 50% and 100% respectively (at 25 ◦C). It is well-known that the micro-capacitor networks are often formed in the nanocomposites [28]. During hot-stretching, the disordered nanoparticles are realigned to form the oriented micro-capacitors in the polymer matrix along the orientation direction (Figure 5), contributing to the enhancement of the dielectric constant of nanocomposites. In addition, the electrical conductivity of PEN/PANI-*f*-BT-b nanocomposites presents a slight increase with the increase of stretching ratio (Figure S3b). This is a result of the second dispersion of PANI-*f*-BT-b in the system. What is more, the permittivity of the PEN/PANI-*f*-BT-b nanocomposites after hot-stretching at different temperatures is also investigated in this work, as shown in Figure 10c. It can be seen that the *T*<sup>g</sup> of PEN/PANI-*f*-BT-b obtained from the permittivity-temperature curve increases to 219 ◦C and 221◦C at the 50% and 100% respectively, indicating that the composites can be used at higher temperature after hot-stretching. It is well-known that the service temperature of BOPP is higher than PP due to the stretching of the sample. Furthermore, the temperature coefficients of dielectric constant of PEN/PANI-*f*-BT-b nanocomposites at different stretching ratios are shown in Figure 10d. As can be seen, the temperature coefficients of dielectric constant of PEN/PANI-*f*-BT-b nanocomposites exhibit a gradual downward trend. More importantly, the temperature coefficients of dielectric constant of PEN/PANI-*f*-BT-b are lower than 5 <sup>×</sup> 10−<sup>4</sup> ◦C, which are very important for its application at high temperature. This is because the high orientation of the molecular chain and the increase of crystallinity limit the movement of the chain segments, which is consistent with the conclusion of the thermal properties [52]. The results reveal that the modification of the filler and hot-stretching method can effectively increase the dielectric constant and the stability of the nanocomposites at different temperatures, which presents great potential for it to be used as a high performance dielectric film in harsh environments.

#### **4. Conclusions**

In conclusion, enhanced dielectric and mechanical properties of polyarylene ether nitrile are obtained through secondary dispersion of polyaniline functionalized barium titanate (PANI-*f*-BT) by hot-stretching. PANI-*f*-BT nanoparticles with different PANI contents are prepared via in-situ aniline polymerization technology, and then characterized by TEM, XPS, FTIR and TGA. The results confirm that the polymer layers have uniformly enclosed on the surface of BaTiO3 nanoparticles. The obtained PANI-*f*-BT nanoparticles are used as functional fillers to compound with PEN for preparing the PEN/PANI-*f*-BT nanocomposites. In addition, these PEN-based nanocomposites are uniaxial hot-stretched by 50% and 100% at 280 ◦C to obtain the oriented nanocomposite films. The results show

that the PANI-*f*-BT nanoparticles present well compatibility and dispersion in the PEN matrix, and the hot-stretching can achieve a second dispersion of the PANI-*f*-BT nanoparticles in PEN, which can effectively enhance the comprehensive properties of nanocomposites like mechanical properties, crystallinity, dielectric constant and so on. When the stretching ratios increase from 0% to 100%, the tensile strengths of PEN/PANI-*f*-BT-b nanocomposite film increase from 83.8 to 161.1 MPa, and the crystallinities increase from 0% to 16.4%. Most importantly, the permittivity temperature dependences of nanocomposites after hot-stretching are more stable than that of original nanocomposites. The excellent performances of the stretched composites indicate that these samples present a great potential to be used as a high-performance dielectric film in high-temperature environments.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2079-4991/9/7/1006/s1, Figure S1: The DSC curves of PEN/PANI-*f*-BT nanocomposite films with different stretching ratios. Figure S2: The DSC curves of PEN/PANI-*f*-BT-b nanocomposite films with different stretching ratios. Figure S3: The electrical conductivity of PEN/PANI-*f*-BT nanocomposites.

**Author Contributions:** Y.Y. and X.L. conceived and designed the experiments; Y.Y. performed the experiments; L.T. (Ling Tu), Y.W. and L.T. (Lifen Tong) analyzed the data; Y.Y. wrote the paper. R.W. and X.L. edited and revised manuscript. All the authors approved the final version of the manuscript.

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

**Acknowledgments:** The authors wish to give thanks for the financial supports from the National Natural Science Foundation of China (No. 51773028 and 51603029), China Postdoctoral Science Foundation (2017M623001) and National Postdoctoral Program for Innovative Talents (BX201700044).

**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/).

### *Article* **Rapid Self-Assembly of Metal/Polymer Nanocomposite Particles as Nanoreactors and Their Kinetic Characterization**

#### **Andrew Harrison 1, Tien T. Vuong 1, Michael P. Zeevi 1, Benjamin J. Hittel 1, Sungsool Wi <sup>2</sup> and Christina Tang 1,\***


Received: 24 January 2019; Accepted: 18 February 2019; Published: 28 February 2019

**Abstract:** Self-assembled metal nanoparticle-polymer nanocomposite particles as nanoreactors are a promising approach for performing liquid phase reactions using water as a bulk solvent. In this work, we demonstrate rapid, scalable self-assembly of metal nanoparticle catalyst-polymer nanocomposite particles via Flash NanoPrecipitation. The catalyst loading and size of the nanocomposite particles can be tuned independently. Using nanocomposite particles as nanoreactors and the reduction of 4-nitrophenol as a model reaction, we study the fundamental interplay of reaction and diffusion. The induction time is affected by the sequence of reagent addition, time between additions, and reagent concentration. Combined, our experiments indicate the induction time is most influenced by diffusion of sodium borohydride. Following the induction time, scaling analysis and effective diffusivity measured using NMR indicate that the observed reaction rate are reaction- rather than diffusion-limited. Furthermore, the intrinsic kinetics are comparable to ligand-free gold nanoparticles. This result indicates that the polymer microenvironment does not de-activate or block the catalyst active sites.

**Keywords:** nanoreactor; catalyst confinement; Flash Nanoprecipitation; diffusion

#### **1. Introduction**

Self-assembled amphiphilic molecules, both small molecules and macromolecules, that confine catalysts to micelle, vesicle, and Janus particle nanoreactor systems have proven to offer an efficient approach to perform organic reactions using water as a bulk solvent [1–3]. Using Janus particles, catalysts can be incorporated into a portion of the nanocomposite particle and the other portion imparts stability to the system. Asymmetric catalyst loading can facilitate particle motion driven by a chemical reaction [3]. In other nanoreactor systems, surfactant micelles that incorporate organic, metal (homogeneous), and metal nanoparticle catalysts have been used for a wide range of coupling reactions in water [4–6]. Confining catalyzed organic reactions to the nanoreactor environment can be leveraged to speed up various chemical reactions [7,8]. Improved yield and selectivity when compared to traditional organic solvents has been reported [6,9].

Core-shell polymer systems have also been considered. Polymeric micelles have been used for several reactions such as asymmetric aldol reactions catalyzed by L-proline [10], acylation [11], hydroaminomethylation of octane catalyzed by Ru-based nanoparticles [12], etc., with extensive reviews available elsewhere [13]. Another approach has been to immobilize metal nanoparticles within

polyelectrolyte-brushes synthesized on a polystyrene core [14]. The polymer microenvironment of these systems can lead to increased local concentrations of reactants, which can accelerate reactions, facilitate reactions of otherwise non-reactive species [9,10,15–17], confer temperature or pH dependent catalytic activity [18], and/or provide specificity based on hydrophobicity [15].

Generally, these promising approaches have involved design and synthesis of amphiphiles, block copolymers, or polyelectrolytes that contain catalyst or ligand for covalent attachment of the catalyst. Additionally, nanoreactor properties, such as catalyst loading and nanoreactor size, are related to the molecular properties of the synthesized material. Thus, varying the nanoreactor properties would require additional syntheses. Approaches to metal nanoparticle catalyst-polymer nanocomposite particle fabrication that would facilitate (1) modular material (off-the-shelf polymer, catalyst) selection, (2) tunable properties (size and catalyst loading), and (3) rapid, scalable productionwould be beneficial to expanding their potential application.

Flash NanoPrecipitation (FNP) is a rapid, scalable method of polymer self-assembly that may be useful for producing nanoreactors. In Flash NanoPrecipitation, an amphiphilic block copolymer and hydrophobic core material are dissolved in a water miscible organic solvent and rapidly mixed against water using a confined impinging jet mixer. Upon mixing, the rapid decrease in solvent quality causes the hydrophobic core material to precipitate and the block copolymer to micellize directing formation of the overall nanocomposite particle. This particle assembly ends when the hydrophobic block of the block copolymer adsorbs on the precipitating core material preventing further growth, while the hydrophilic block sterically stabilizes the nanoparticle. Given the molecular weight of the block copolymer, dynamic exchange of the block copolymer does not occur [9,19,20], so the resulting structure is kinetically-trapped.

Hydrophobic, inorganic nanoparticles have been incorporated into nanocomposite particles by dispersing the nanoparticles with the dissolved block copolymer and then mixing with confined impinging jets. Upon mixing, colloidal aggregation and block copolymer self-assembly occur due to the decrease in solvent quality. Nanocomposite particle assembly is complete when sufficient hydrophobic blocks of the block copolymer adsorb to the nanoparticle clusters to prevent further colloidal aggregation. For example, Gindy et al. demonstrated fabrication of polymer nanostructures containing colloidal gold using Flash NanoPrecipitation [21]. More recently, Pinkerton et al. encapsulated iron oxide nanoparticles for medical imaging applications [22]. For medical imaging, ~100 nm composite nanostructures with tunable inorganic nanoparticle loading were achieved. These studies suggest that Flash NanoPrecipitation is a suitable method for nanoreactor fabrication. However, the ability to independently tune inorganic nanoparticle loading and nanocomposite particle size has yet to be demonstrated.

Other important considerations when using the nanocomposite particles as nanoreactors are the reaction and diffusion within the system. In small molecule micelle systems that are thermodynamically stable, there is constant molecular exchange between the bulk solvent, and the confined hydrophobic mesophase facilitates reaction [6,9]. In the kinetically-trapped systems produced by Flash Nanoprecipitation, reactants and products reach the catalyst by partitioning from the bulk and diffusing through the nanoreactor structure [17]. The potential mass transfer limitations and the effect of incorporation into the nanocomposite particle on reactivity of the catalyst need to be established.

In this work, we use Flash NanoPrecipitation for rapid and scalable self-assembly of hybrid metal nanoparticle catalyst-polymer nanocomposite nanoreactors. Independently tuning the nanoreactor properties, namely size and gold loading, is investigated. We focus on fundamental understanding of reaction and diffusion using the reduction of 4-nitrophenol as a model reaction. Kinetic and scaling analysis following the induction time are also discussed.

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

#### *2.1. Materials*

Citrate stabilized 5 nm gold nanoparticles were purchased from Ted Pella. Polystyrene (PS, MW 800–5000 g/mol) was purchased from Polysciences, Inc. Sodium borohydride and 4-nitrophenol were purchased from Sigma Aldrich (St. Louis, MO, USA). Dodecanethiol (DDT) stabilized 5 nm nanoparticles, tetrahydrofuran (THF), HPLC grade), ethanol (ACS reagent grade), and diethyl ether (ACS reagent grade) were purchased from Fisher Scientific (Fairmont, NJ, USA). Environmental Grade Hydrochloric Acid 30-38% and Environmental Grade Nitric Acid 70% were purchased from GFS Chemicals (Columbus, OH, USA). The 1H-NMR solvent D2O with 4,4-dimethyl-4-silapentane-1-sulfonic acid DSS as an internal standard was purchased from Cambridge Isotope Lab, Inc (Andover, MA, USA). These chemicals and materials were used as received. Polystyrene-b-polyethylene glycol (PS-b-PEG, PSm-b-PEGn where m = 1600 g/mol and n = 5000 g/mol) was obtained from Polymer Source (Product No. P13141-SEO). Prior to use, PS-b-PEG was dissolved in THF (500 mg/mL) and precipitated in ether (~1:20 *v*/*v* THF:ether). The PS-b-PEG was recovered by centrifuging, decanting, and drying under vacuum at room temperature for 2 days.

#### *2.2. Nanoreactor Assembly*

For self-assembly, the gold nanoparticles need to be dispersed in a water miscible solvent such as THF with molecularly dissolved block copolymer. To disperse the gold nanoparticles in THF, the as-received dodecanethiol stabilized gold nanoparticles in toluene (1 mL) were precipitated into ethanol (45 mL) and filtered using a Buchner funnel. The filtered nanoparticles from the filter cake were resuspended in THF and concentrated via evaporation at room temperature overnight to achieve a nominal concentration of around 20 mg/mL. The final concentration was confirmed by inductively coupled plasma optical emission spectroscopy using an Agilent 5110 (ICP-OES, Santa Clara, CA, USA). UV spectra collected on an Ocean Optics FLAME-S-UV-VIS with a HL-2000-FHSA light source (Largo, FL, USA) were compared before and after the solvent switch to confirm processing did not significantly affect gold nanoparticle size.

For nanoreactor self-assembly, typically, PS-b-PEG (6 mg), dodecanethiol stabilized 5 nm gold nanoparticles (0.5 mg), and PS homopolymer (co-precipitate, 5.5 mg) were added to 0.5 mL of tetrahydrofuran (THF) and sonicated at 55 ◦C for 30 min. Using a manually operated confined impinging jet mixer with dilution (CIJ-D) [23,24] with achievable Reynolds' numbers >1300, the resulting THF mixture was rapidly mixed against 0.5 mL of water into a stirring vial of water (4 mL). The resulting dispersion (5 mL total) was stored at room temperature for further characterization and analysis without purification. The nanocomposite particle properties were tuned by adjusting the total solids concentration or the relative amounts of gold nanoparticles and the co-precipitate at a constant total mass or a constant total core volume based on the bulk density of gold and co-precipitate.

#### *2.3. Nanoreactor Characterization*

Nanoreactor size was measured after mixing using a Malvern Zetasizer Nano ZS (Westborough, MA, USA) with a backscatter detection angle of 173◦. Size distributions are reported using the average of four measurements of the intensity weight distributed with normal resolution. The reported size is the peak 1 mean intensity. The polydispersity index (PDI) is defined from the moment of the cumulant fit of the autocorrelation function calculated by the instrument software (appropriate for samples with PDI < 0.3) and is reported as a measure of particle size distribution. UV absorbance spectra (300 to 1200 nm) of the nanoparticle dispersions were measured at room temperature with an Ocean Optics FLAME-S-UV-VIS with a HL-2000-FHSA light source (Largo, FL, USA). For visualization by TEM, samples were submerged in a dilute dispersion of nanoreactors (10-fold dilution with water) for one hour and dried at ambient conditions overnight. Samples were imaged using a Zeiss Libra 120 TEM (Oberkochen, Germany) using an accelerating voltage of 120 kV. To determine the gold nanoparticle

concentration, nanoreactor dispersions were dissolved in THF and digested in aqua regia (1:3 nitric acid:hydrochloric acid by volume) and diluted to 5% *v*/*v* aqua regia. Gold concentration of the digested sample was measured using inductively coupled plasma optical emission spectroscopy measurements with an Agilent 5110 (Santa Clara, CA, USA).

#### *2.4. Kinetic Analysis*

The catalytic performance of the nanoreactors was evaluated using the reduction of 4-nitrophenol with sodium borohydride as a model reaction using well established procedures [25,26]. Briefly, the nanoreactors were diluted with 4-nitrophenol (aq.) and aqueous sodium borohydride (within 5 min of preparation) and the reduction of 4-nitrophenol was monitored using UV spectroscopy (Ocean Optics FLAME-S-VIS-NIR-ES, Largo, FL, USA, with a HL-2000-FHSA light source (300–1200 nm) with a CUV-UV cuvette holder placed on a stir plate). The final reaction mixture contained less than 0.01 vol% THF. The induction time and apparent reaction rate (*kapp*) were determined from tracking the absorbance at 425 nm as a function of time. The values of *kapp* and induction time are the averages (± standard deviations) of at least 3 trials of each experiment. Detailed procedures are provided in the Supporting Information.

#### *2.5. Langmuir-Hinshelwood Kinetics*

For more detailed kinetic analysis, we performed full kinetic analysis considering the two-step reaction mechanism previously established [26]. Full kinetic analysis is described by the reaction rate of each step and the Langmuir adsorption constants of 4-nitrophenol, borohydride, and the stable intermediate. We determined the rate constants for both steps by solving the coupled rate equations using the numerical method previously described and fitting the experimental data (average of three experimental trials) [26].

#### *2.6. NMR Measurements*

To evaluate effective transport of the 4-nitrophenol, 1H-NMR spectroscopy and pulsed field gradient (PFG) NMR, combined with saturated transfer difference (STD) spectroscopy, using a Bruker 800 MHz cryo-probe (Billerica, MA, USA) was performed in accordance with the methods described in the Supplemental Information. Briefly, 4-nitrophenol molecules in close proximity to the nanoreactor core were analyzed based on spin diffusion of selectively saturated polystyrene, in conjunction with an applied magnetic field gradient. Relevant intensities were analyzed as a function of gradient strength to determine the diffusion coefficient of the molecules. Since nanoreactors diffuse in free solution at least 3 orders of magnitude slower than molecules, the measured diffusion coefficient was considered the effective diffusion coefficient of the solute within the nanoreactor [27–29].

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

#### *3.1. Nanoreactor Self-Assembly*

To perform Flash NanoPrecipitation, dodecanethiol stabilized 5 nm gold nanoparticles were dispersed in THF with the molecular dissolved, PS and PS-b-PEG, and rapidly mixed with water using a hand-operated confined impinging jet mixer. The entire formation process was accomplished in less than a second; further, the process can be performed continuously at large scales [24,30,31]. Due to their hydrophobic nature and particle aggregation during assembly, the gold nanoparticles are expected to be in the hydrophobic core of the nanoreactor [3,21], forming a nanoparticle-macromolecular system [32]. Due to the high molecular weight of the polystyrene block, no dynamic exchange of the block copolymer is expected [20]. The resulting nanoreactors were ~130 nm indicated by a single Gaussian peak with PDI <0.2 measured by dynamic light scattering (DLS). The dispersions were stable when stored at room temperature for at least 2 months as there was no significant change in size or size distribution by DLS (Supporting Info, Figure S5), and no macroscopic precipitation of unencapsulated gold was observed.

We further characterized the nanoreactors using UV-Vis spectroscopy. Prior to Flash NanoPrecipitation, the dodecanethiol-stabilized nanoparticles dispersed in toluene showed a peak absorbance at 495 nm (as received and after switching solvents). The nanoreactors showed a peak absorbance of 520 nm (Figure 1b). The peak shift could occur due to differences in hydrophobicity of the surrounding environment [33]. Since the polystyrene microenvironment should have similar hydrophobicity as toluene, we attribute the red-shift to plasmonic coupling due to close proximity of the encapsulated gold nanoparticles, which has been previously observed with polymer-gold nanocomposite particles [34].

**Figure 1.** Polymer nanoreactors were fabricated via self-directed assembly. (**a**) DLS confirms the uniform size distribution of the ~130 nm self-assembled polymer nanoreactors (black circles) and confirms that the size is the same after the reduction of 4-nitrophenol (red squares). (**b**) UV-vis analysis shows that the absorbance of the gold nanoparticle remains unchanged through the solvent switch from toluene (black filled circles) to tetrahydrofuran (THF) (red open circles). A red-shift is seen upon encapsulation within polymer nanoreactors (blue open diamonds) due to close proximity of the encapsulated gold nanoparticles. (**c**) TEM imaging demonstrates that multiple gold nanoparticles were encapsulated within the core of the nanoreactors.

The structure of the nanocomposite particles was visualized using TEM. Based on TEM imaging, clustering of the gold nanoparticles during assembly resulted in multiple catalytic gold nanoparticles per nanoreactor. The majority of the gold nanoparticles appear to be in the nanoreactor core, although multiple polymer layers are not visible on TEM due to low electron density. This result is consistent with previous reports of encapsulated gold nanoparticles via Flash NanoPrecipitation (FNP) [21,35]. Based on TEM, some of the gold may also be associated with the PEG-layer of the nanoreactors whereas unassociated gold would be expected to precipitate out of the dispersion as well as affect the size distribution measured by DLS. Since we do not observe gold precipitate from the dispersion, and the size of the TEM size is consistent with DLS with PDI <0.2, we assume all the gold in the dispersion is associated with the nanoreactors. Finally, we confirmed the amount of gold by ICP-OES. We found the polystyrene nanoreactors retained 74% of the gold from the THF-gold nanoparticle solution and the loss can be attributed to the hold-up volume during mixing.

Next, we aimed to independently tune the nanoreactor properties, size and gold loading, using formulation parameters. Nanoreactor assembly depends on the relative time scales of block copolymer micellization, gold nanoparticle clustering, and co-precipitate nucleation and growth. Therefore, the overall nanoreactor size can be affected by the ratio of core material to block copolymer, as well as the total concentration of components in the organic stream [22].

Varying the ratio of block copolymer to core materials has been an effective method for tuning nanostructure size via Flash NanoPrecipitation [23,36]. To vary nanoreactor size, the amount of block copolymer concentration can be increased (Supporting Information, Figure S6), but the gold loading is also affected. In order to vary the nanoreactor size while holding the gold loading constant, we varied the total solids concentration holding the mass ratio of gold to polystyrene co-precipitate constant. As expected, the nanoreactor size increased with increasing total solids concentration. This effect has been attributed to an increase in the rate of particle core relative to nucleation [36,37]. Using this approach, the nanoreactor size could be tuned between 100 and 200 nm with nominal gold loading of 4 wt % (Figure 2a). This level of gold loading is comparable other polymer nanocomposite systems with low volume additions of inorganic nanoparticles that demonstrate enhanced functional performance [32].

Next, we aimed to vary the gold loading independently of nanoparticle size. Holding the total core material mass constant and varying the ratio of gold to polymer resulted in a decrease in nanoparticle size with increasing gold concentration. In contrast, with gold nanoparticles and block copolymer without a co-precipitate, Gindy et al. observed that increasing the gold loading results in an increase in nanocomposite particle size that is attributed to the increase in the amount of gold core relative to the block copolymer [21]. The difference is our use of a co-precipitate. We attribute the trend observed in this case to the increase in the number density of gold nanoparticles that act as nucleating agents that seed particle growth via heterogeneous nucleation [37,38].

To guide nanoreactor formulation, the Smoluchowski diffusion limited aggregation model has previously been used to formulate inorganic nanoparticle-polymer nanocomposite particles via Flash NanoPrecipitation [36]. Based on the model, nanoreactor size can be predicted using:

$$R = \left( K \frac{k\_B T \varepsilon\_{core}^{5/3}}{\pi \mu \rho \varepsilon\_{BCP}} \right)^{1/3} \tag{1}$$

where *R* is the aggregate radius, *K* is a constant of proportionality for formation time, *kB* is Boltzmann's constant, *T* is the absolute temperature, *ccore* is the concentration of core material, *cBCP* is the concentration of block copolymer, *μ* is the solvent viscosity, and *ρ* is the core material density. This model suggests that the nanoreactor size is affected by the volume more than the mass of the core. Thus, as an alternative to holding the mass of the core constant, we held the volume of the core constant, according to:

$$V\_{CM} = \frac{m\_{AuNP}}{\rho\_{AuNP}} + \frac{m\_{PS}}{\rho\_{PS}} \tag{2}$$

where *Vcm* is the total volume of the core materials, *mAuNP* and *mPS* are the masses of the gold nanoparticles and polystyrene core materials, respectively, finally *ρAuNP* and *ρPS* are the densities of the gold nanoparticles and polystyrene core materials, respectively. The core volume was selected from the standard formulation, a nominal gold loading of 4% and nanoreactor concentration of 2.4 mg/mL. Using the density of bulk gold and polystyrene, which are 19.32 g/mL and 1.04 g/mL, respectively, the core material volume was found to be 5.33 μL. Using the approach of constant volume, the gold loading was tuned between 4 and 50 nominal wt % at a nanoreactor size of ~130 nm (Figure 2b).

Overall, nanoreactors were assembled in a rapid, scalable, single-step method using Flash NanoPrecipitation. Nanoreactor size could be tuned independently of gold nanoparticle loading by varying the total solids concentration at a constant ratio of gold to polystyrene. Interestingly, the gold nanoparticle loading was tuned independently of nanoreactor size by varying the ratio of gold to polystyrene at constant total core volume. The constant core volume approach may be useful for formulations of multiple components with disparate densities e.g., inorganic particle-polymer nanocomposite particles.

**Figure 2.** Hydrodynamic diameter of polystyrene nanoreactors measured by DLS with varying total nanoreactor material concentration in the formulation. (**a**) By varying the total material concentration with constant ratio of components tunable nanoreactor size between 100–200 nm. (**b**) By varying the gold to polystyrene co-precipitate ratio at a constant nanoreactor core volume (red squares), as opposed to constant mass ratio (black circles), the nominal gold loading of polystyrene nanoreactors can be tuned at constant nanoreactor size (~130 nm). The standard formulation (4 wt % nominal gold loading, 2.4 mg/mL) is shown by the red triangle.

#### *3.2. Initial Characterization of Nanoreactor Performance*

To evaluate the catalytic performance of the nanoreactor, the reduction of 4-nitrophenol by sodium borohydride was used as a model reaction [39]. First, we confirmed the nanoreactors remained intact following the reaction; no significant change in size or polydispersity was observed by DLS (Figure 1a). Further, no macroscopic precipitation of gold nanoparticles was observed following the reaction.

In these initial studies, we assume all of the gold nanoparticles included in the formulation are associated with the nanoreactor and contribute to the observed catalytic activity. From TEM (Figure 1c), the gold nanoparticles may be associated with the hydrophobic core or hydrophilic shell or may be unencapsulated. Unencapsulated gold was not observed precipitating from the nanoreactors and would not contribute to the observed activity (Table 1, dodecanethiol-stabilized gold nanoparticles (DDT)). This is likely due to the lack of solubility as other hydrophobic inorganic nanoparticles have shown activity in water:solvent reaction mixtures [40]. If the dispersions contained trace amounts of unencapsulated gold, the reported values for k1 would be slightly underestimated. The conversion of 4-nitrophenol confirmed the gold nanoparticles associated with the nanoreactors were catalytically active (Supporting Information, Figure S4a). The apparent reaction rate constant per surface area of gold, k1, for the nanoreactors was 0.414 ± 0.095 L m<sup>−</sup>2s−1, which is comparable to the citrate-stabilized, 5 nm gold particles.

**Table 1.** Rate constants and induction times for various gold nanoparticles. PS, polystyrene; DDT dodecanethiol-stabilized gold nanoparticles.


Comparing the performance of the nanoreactors with other metal nanoparticle-polymer systems using the reaction rate considering the amount of gold catalyst (e.g., k1 in Table 2), the nanoreactors demonstrate over 110-fold better catalytic activity than gold within polymer (PNIPAM-b-P4VP) micelles, despite a larger overall nanoreactor size. This difference may be attributed to P4VP-gold interactions that affect availability of active sites. Thus, the use of non-interacting co-precipitates and Flash NanoPrecipitation may provide an advantage to other polymer micelle systems that rely on gold-polymer interactions for self-assembly.

Further, the induction time and kinetics are similar to immobilized gold nanoparticles within polyelectrolyte brush shell on polystyrene core particle systems [26]. Specifically, the kinetics of the nanoreactors we report with 5 nm gold are comparable to polyelectrolyte brushes with 2.2 nm gold nanoparticles at the surface of the core-shell nanostructures, which are expected to have similar activities [42]. This result suggests that association of the catalyst with the nanoreactor does not sacrifice reactivity.


**Table 2.** Rate constants for various metal/polymer nanocomposite nanoreactors.

#### *3.3. Probing Potential Mass Transfer Limitations*

#### 3.3.1. Induction Time

Notably, the induction time of the encapsulated gold nanoparticles is ~50-fold longer than citrate-stabilized nanoparticles (Table 1). This relatively long induction time has been previously observed with gold-nanoparticle-polymer nanoreactor systems. It may be attributed, in part, to slow surface restructuring upon encapsulation within the hydrophobic polystyrene microenvironment [14]. Additional factors that may increase induction time include: poisoning of the active sites when encapsulated within the nanoreactor core, reduction of the dissolved oxygen present in the reaction dispersion, and/or diffusion limitations [44,45].

To further understand the nature of the induction time in the nanoreactor system, we investigated both the sequence of addition and the time between adding the reactants (Figure 3). Under standard model reaction conditions, 4-nitrophenol was added first and allowed to equilibrate for 1 min, followed by the addition of the sodium borohydride. To probe potential diffusion limitations, we increased the time between adding the 4-nitrophenol and sodium borohydride 10-fold, and no significant change in induction time was observed. This result suggests that the induction time is not related to diffusion of 4-nitrophenol.

Moreover, switching the sequence to adding sodium borohydride first, followed by 4-nitrophenol after 1 min of equilibration did not significantly affect the induction time. Interestingly, when the equilibration time was increased in this case, the induction time was reduced by two orders of magnitude. This ~5 s induction time is comparable to the value measured for citrate-capped gold nanoparticles. This result indicates the long induction times relative to citrate stabilized gold nanoparticles may be attributed to diffusion of sodium borohydride. Further examining the effect of equilibration time, the induction time decreased from ~100 to 5 s when increasing the equilibration time from 1 to 3 min (Figure 4). Further increasing the equilibration time beyond 3 min did not significantly impact the induction time. Thus, it appears that it takes ~3 min for sufficient sodium borohydride to partition into the nanoreactor for the reaction to progress. This required equilibration time can be reduced by increasing the concentration of the borohydride (constant ratio of borohydride to 4-nitrophenol) (Supporting Information, Figure S7) which further indicates the relatively long induction time of the nanoreactors relative to citrate stabilized gold nanoparticles can be attributed to diffusion of the borohydride.

**Figure 3.** The effect of the sequence of reagent addition on the induction time of the 4-nitrophenol reaction. In all experiments, the 4-nitrophenol and sodium borohydride concentration followed standard conditions of 0.01 mM and 0.01 M, respectively. The indicated reagent was the first to be added, after which the reagent was allowed to equilibrate in the solution for either 1 min (black striped bars) or 10 min (red solid bars). The end of the equilibration period was the addition of the second reagent, at which point the reaction could progress.

**Figure 4.** The effect of sodium borohydride equilibration on the induction time of the 4-nitrophenol reaction. Standard reagent concentrations of 0.01 mM and 0.01 M were used for 4-nitrophenol and sodium borohydride, respectively. Data points marked with an asterisk (\*) are significantly different than each other (*p* < 0.1).

#### 3.3.2. Reaction Rate

Next, we further investigated potential mass transfer limitations on the observed reaction rate following the induction time. A useful tool for determination of diffusion limitations is the second Damköhler number (DaII), which is a ratio of the reaction rate to the diffusion rate given by:

$$
\beta DaII = \frac{k\_{app}C^{u-1}}{\beta a} \tag{3}
$$

where *n* is the reaction order, *β* is the mass transport coefficient (which is a quotient of the diffusion coefficient and the characteristic length of the system), and a is the interfacial area. To calculate DaII for a 130 nm diameter particle, the interfacial area (nanoreactor area per unit volume of nanoreactor dispersion) was estimated to be 2 × 104 <sup>m</sup>−<sup>1</sup> based on the number of nanoreactors estimated using the aggregation number of the block copolymer previously reported [46,47]. The diffusion coefficient for 4-nitrophenol in the nanoreactor system was experimentally determined by NMR. Using PFG-NMR in conjunction with the STD spectroscopy, the effective diffusion coefficient of the 4-nitrophenol within the nanoreactors was determined to be 1.91 ± 0.01 × <sup>10</sup>−<sup>8</sup> m2/s (Figure S2). Using this experimentally determined effective diffusion coefficient, the DaII is on the order of 10−<sup>6</sup> indicating the reaction is significantly slower than diffusion; therefore, the apparent kinetics are reaction-limited.

A complementary approach was to consider the bimolecular reaction between 4-nitrophenol and nanoparticle catalyst using the Smoluchowski diffusion limited reaction model [48,49]. We varied the gold concentration by (1) varying the nanoreactor concentration to probe potential external diffusion limitations, and (2) varying the gold loading at constant nanoreactor concentration to examine potential internal diffusion limitations (Figure S8). When the nanoreactor concentration or the gold loading was increased, *kapp* increased; the 2nd order rate constant was on the order of 10<sup>6</sup> M<sup>−</sup>1s−1. These values are much lower than the kbm ~ 108 M−1s−1, indicating that neither internal nor external diffusion from the bulk solution to the nanoreactor limited the apparent reaction kinetics.

Since there were no indications of diffusion limitations associated with the reaction following the induction time, we further characterized the reaction kinetics using Langmuir-Hinshelwood kinetics. Based on the previously established two-step reaction model [26], and fitting the measured concentration of 4-nitrophenol as a function of time (normalized after the induction time for conversions up to 30%) [26], the kinetics were comparable to other gold nanoparticle-polymer nanoreactor systems (Table 3 with plot, Figure S3, and full fit parameters, Table S1, in the Supporting Information). Interestingly, ka and kb observed for the gold encapsulated within the nanoreactors are comparable to ligand-free gold nanoparticles. This result suggests that the reactivity of the gold nanoparticle surface is not significantly affected by self-assembly and their incorporation into the nanoreactors.


**Table 3.** Langmuir-Hinshelwood rate constants obtained from fits to experimental data.

Overall, diffusion and partitioning of sodium borohydride into the polymer nanoreactor affect the induction time for the reaction. Sufficient equilibration time between adding the sodium borohydride and the 4-nitrophenol (~3 min) for the borohydride to partition and diffuse minimizes induction time. Notably, mass transfer effects are not observed after the induction time and the intrinsic kinetics are comparable to ligand-free gold nanoparticles.

#### **4. Conclusions**

Overall, we have presented rapid, scalable self-assembly of metal nanoparticle catalyst-polymer nanocomposite particles as nanoreactors. The size and gold loading of the nanoreactors can be tuned independently, with sizes and nominal loadings ranging from 100–200 nm and 4–50 wt% respectively. Using the 4-nitrophenol reduction as a model reaction, the induction time is affected by sequence or reagent addition, time between addition, and reagent concentration. Combined, our experiments indicate that the induction time is most influenced by diffusion of sodium borohydride. Scaling analysis and effective diffusivity measured using NMR, the observed reaction rate after the induction time are reaction- rather than diffusion-limited. Finally, the intrinsic reaction kinetics of gold associated with the polymer were comparable to ligand-free particles indicating the self-assembly process and resulting polymer microenvironment did not de-activate or block the catalyst active sites. Building on this foundational study, practical considerations such as nanoreactor recycling will be considered in future work.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2079-4991/9/3/ 318/s1, Figure S1: Representative Kinetic Data Analysis, Figure S2: PFG-NMR with STD, Figure S3: Langmuir-Hinshelwood Model Fitting, Figure S4: Confirmation of Nanoreactor Activity, Figure S5: Nanoreactor Stability, Figure S6: Nanoreactor Size Tuning with Block Copolymer Concentration, Figure S7: Effect of Sodium Borohydride Equilibration on Induction Time, Figure S8: External and Internal Mass Transfer, Table S1: Langmuir-Hinshelwood Fitting Parameters, Table S2: Rate Constant After Recycling.

**Author Contributions:** Conceptualization, A.H., S.W. and C.T.; Formal analysis, A.H., T.T.V., M.P.Z., B.J.H., S.W. and C.T.; Funding acquisition, S.W. and C.T.; Investigation, A.H., T.T.V., M.P.Z., B.J.H. and S.W.; Methodology, A.H., S.W. and C.T.; Project administration, C.T.; Supervision, C.T.; Validation, A.H. and C.T.; Visualization, A.H., S.W. and C.T.; Writing—original draft, A.H., S.W. and C.T.; Writing—review & editing, A.H., T.T.V., M.P.Z., B.J.H., S.W. and C.T.

**Funding:** This research was partially supported by startup funding at Virginia Commonwealth University, and NSF (Award number CMMI-1651957). A portion of this work was performed at the National High Magnetic Field Laboratory, which is supported by the National Science Foundation Cooperative Agreement No. DMR-1157490 and DMR-1644779 as well as the State of Florida.

**Acknowledgments:** The authors gratefully acknowledge funding for this research which was partially supported by startup funding at Virginia Commonwealth University, and NSF (Award number CMMI-1651957). This work was also supported by the National High Magnetic Field Laboratory through the NSF (DMR-1157490 and DMR-1644779) and by the State of Florida. We would also like to thank Joseph Turner of VCU and Christine Lacy of University of Richmond for their technical support.

**Conflicts of Interest:** The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results.

#### **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/).

*Article*
