One-pot Electrosynthesis of Multi-layered Magnetic Metallopolymer Nanocomposites

Researchers have been investigating various methodologies for fabricating well-defined, homogenous composites consisting of nanoparticles (NPs) dispersed in a matrix. The main challenges are to prevent particle agglomerations during fabrication and to obtain nanoparticles whose size distribution could be tuned on demand. One of the methods that can provide these features is electrodeposition. We report for the first time the fabrication of a thin magnetic multilayer nanocomposite film by electrodeposition from one bath containing both a monomer and metal salts. Cobalt and cobalt-nickel NPs were deposited on conductive polymer polypyrrole thin films using different electrodeposition potentials and times. Multilayer nanocomposite films were fabricated by subsequent electrodeposition of polymer and nanoparticle layers. Scanning electron microscopy analysis showed that a wide range of NPs (70-230 nm) could be synthesized by manipulating growth potentials and times. The cobalt-nickel NPs were found to contain hexagonal close-packed (hcp) and face-centered cubic (fcc) phases based on X-ray diffraction and selected area electron diffraction. Magnetic measurements proved that both the single and the multi-layered nanocomposites were magnetic at room temperature.


Introduction
Polymers have become essential materials in the fields of micro-and nanoelectromechanical systems (MEMS and NEMS, respectively). Due to their low density and flexibility, polymeric materials are not only employed as small components, but they can also serve as stamps for building blocks in MEMS and NEMS devices [1][2][3][4][5][6] . Functional parts such as micro pumps, valves or microfluidic channels are often made of polymers due to their mechanical properties, ease of fabrication, and low manufacturing costs 2 . Some of the most common MEMS and NEMS polymers are polydimethylsiloxane (PDMS), SU-8, polymethylmethacrylate (PMMA), parylene, and polyimide 2 . To extend their performance and functionality, many nanostructures such as nanotubes, nanosheets, and nanoparticles (NPs) have been incorporated in polymeric matrices to enhance their flexibility, mechanical, electrochemical or catalytic properties [7][8][9][10][11] . Polymer-matrix nanocomposite materials have been employed for a variety of applications such as energy harvesting, sensing and actuation, catalysis, and biocidal food packaging [8][9][10][11][12] . There is also interest in the fabrication of magnetic-field responsive polymer nanocomposites. Small components and parts made of magnetic materials are extensively used in transducers, microfluidic control systems, and drug delivery platforms. The interesting magnetic behaviour of nanostructures combined with the wide functionality of polymers has recently resulted in devices built of magnetic polymer nanocomposites (MPN). For example, Suter et al. demonstrated the fabrication of microcantilevers made of photocurable epoxy SU-8 containing superparamagnetic magnetite nanoparticles (SMNP) 13 . Kwon and co-workers have shown that it is possible to build microactuators with programmed anisotropy consisting of a poly(ethyleneglycol)diacrylate (PEGDA) matrix with dispersed SMNP 14 . Sotiriou et al. have fabricated multi-layered plasmonic and phosphorescent superparamagnetic actuators made of PMMA matrices 15 . Some investigations are also being conducted on hydrogel-based magnetic nanocomposites. For example, Olsson et al. manufactured aerogel magnets and magnetic nanopapers using bacterial cellulose matrices filled with ferrimagnetic cobalt ferrite NPs 16 .
Several approaches have been established to engineer MPN. The most commonly used method is in-situ polymerization in the presence of magnetic NPs 13,14,16 . However, these methods have several disadvantages. One of the most fundamental issues is the agglomeration of particles due to strong magnetic and physical interactions. In this case, complex chemical formulations containing dispersing agents are employed to avoid the formation of agglomerates. Other problems arise from the uneven distribution of filler NPs within the polymer matrix. To overcome these drawbacks, some nanocomposites are fabricated by alternating several times the deposition of the polymer followed by the deposition of the nanoparticles. Layerby-layer deposition or spin coating combined with flamesynthesis deposition are among these methods 9,15 . However such techniques require combination of sequential fabrication steps that increases the complexity of fabrication and costs 15,17 . Another technique for fabricating MPN is electrodeposition. By combining suitable anodic and cathodic deposition phases, MPN can be fabricated in a layer-by-layer manner from one bath containing both the monomer and metal salts. For this purpose, conductive polymers can be electrodeposited using positive potentials (anodic phase) with metallic NPs deposited at negative potentials (cathodic phase). This method offers several advantages like ease of operation, reduced costs, and precise control of dimensions and composition by manipulation of deposition potentials and time 18 .
Among the conducting polymers, polypyrrole (Ppy) has been widely studied based on its high stability and good conductivity. Various metal (Cu, Ni, Au, Ag) NPs have been grown electrochemically on thin polypyrrole films [17][18][19][20][21] . In most of these cases, deposition of polymer and metal nanoparticles are performed using separate electrolytes. It has also been shown that nanocomposite fabrication can be conducted from a single electrolyte containing both the monomer and the metal salts 18,19 . In addition, Ppy is an excellent substrate to control the size of the electrodeposited nanoparticles; varying sizes can be obtained by simply changing the thickness of the Ppy 17,20 . However in all cases, only single layer nanocomposites were fabricated, and there have not been any reports on the fabrication of multi-layered composites by electrodeposition.
Here, we report one-pot electrochemical synthesis of multilayered thin film magnetic nanocomposites. Cobalt and cobaltnickel nanoparticles were electrodeposited on an electrochemically grown polypyrrole film by reverse pulse plating from an electrolyte containing both the metal salt and the pyrrole (Py) monomer. Multi-layered nanocomposites were fabricated by electrodeposition of alternating subsequent polymer and nanoparticle layers.

Results and Discussion
The nanocomposites were obtained by anodically electrodepositing an initial Ppy film on a gold substrate, followed by the cathodic electrodeposition of the metallic NPs. A pulse plating approach was employed and the pulses for all layers were fixed to 1 ms with a duty cycle of 50 %. The growth times given represent total deposition time (tT) including the on (tON) and off (tOF) times. The multi-layered samples were obtained by subsequently repeating the process several times. The fabrication steps are given in Fig. 1. Using the designed electroplating technique, it was possible to tune the number of layers in the nanocomposites.

Cyclic Voltammetry
In order to assess the growth conditions, cyclic voltammetry scans for the Co and Co-Ni baths were performed (Fig. 2). The oxidation peak for cobalt can be clearly seen in the Py-Co bath. However, for Py-CoNi, this peak is significantly smaller, suggesting that the dissolution of Co is much higher than that of CoNi. In addition, reduction for both metals starts at -1 V, which was taken as the upper limit for electrodeposition. Ppy deposition potential was chosen to be +1 V.

Nanoparticle growth on polypyrrole thin films
The scanning electron microscopy (SEM) images showed that NPs were successfully grown on polypyrrole thin films for both baths. Fig. 3 shows selected SEM images for cobalt and cobaltnickel NPs-polypyrrole bilayer nanocomposites.
Different electrodeposition conditions were investigated for the nanoparticle growth on Ppy, within the range of -2 V to -5 V. In order to keep the charge density similar in all cases, the deposition time was reduced for higher electrodeposition potentials. By controlling the electrodeposition conditions it was possible to successfully tune the cobalt (Co) and cobaltnickel (CoNi) average nanoparticle size from 60 to 250 nm ( Table 1). As expected, for a given potential particle size typically decreases with decreasing deposition time ( Fig. 3 (a) and (b)). However, such differences in size are not so evident when comparing very similar deposition times (c.f. average NP size for Co and CoNi at -3V for 1.1 s and 0.88 s) due to the inherent variability of the deposition process and fluctuations of current efficiency with time growth. At -2 V a combination of rounded and hexagonal star-like structures was found for both Co and CoNi NPs (Fig. 3 (c)). The occurrence of hexagonal star-like NPs was more pronounced for CoNi. Current efficiency also affected the particle size as noted in Fig. 3 (a) and (d), where deposition charge was kept almost constant but the deposition potential was more negative for Co (-5 V) than CoNi (-3 V). This is probably due to favoured H2 evolution at higher potentials. Also, potentials higher than -3 V were not suitable for CoNi particle production due to increased hydrogen  An initial layer of Ppy is achieved by pulse plating on gold coated silicon substrates. This is followed by the electrodeposition of metallic NPs. This procedure is performed multiple times to produce the multilayer nanocomposite. The pulses are not to scale and the rest phase introduced between Ppy and metallic NPs is not shown for simplicity. evolution. At -5 V H2 evolution was greatly enhanced, and almost no deposition took place for CoNi particles since the current efficiency was very low. The particle size distributions for the conditions of Figure 3 are shown in Fig. S1.
EDX studies showed that the electrodeposition conditions can also be used to tailor the composition of CoNi NPs. For example, CoNi composition was found to be 46% cobalt and 54% nickel for particles produced at -2 V, whereas cobalt composition was decreased to 23% (77% nickel) at -3 V. A representative EDX spectrum is given in Fig. 4. Table 1 Deposition conditions (applied potential and total deposition time) with the resulting approximate nanoparticle size are given. For all cases, pulse time was fixed to be 1 ms with a duty cycle of 50%. CoNi NPs could not be observed for all conditions and are not applicable (n.a.)

Fabrication of multilayer nanocomposite
The first attempts to grow multilayer nanocomposites were carried out using the Py -Co bath, from which successive layers of Ppy (deposition at 1V, tT of 15 seconds) and Co NPs (deposition at -5 V, tT of 1 second) were electrodeposited on goldcoated silicon substrates. Although bilayer Ppy -Co nanocomposites were fabricated successfully ( Fig. 3 (a) and (b)) it was observed that the multilayer nanocomposite did not contain any NPs when analysed by FIB-SEM suggesting that Co NPs are mainly re-dissolved into the solution during growth. This is consistent with results obtained from the cyclic voltammetry scans (Fig. 2) in which during the anodic phase Co oxidation is very high leading to the re-dissolution of the Co NPs back into the bath during the anodic Ppy growth phase.
In order to avoid substantial re-dissolution of the metal counterpart during multilayer nanocomposite growth, fabrication was performed using the Py -CoNi bath, which shows a considerably smaller oxidation peak for CoNi than pure Co (Fig. 2). Fabrication using the Py -CoNi bath resulted in the successful growth of Ppy -CoNi NPs multi-layered nanocomposites from a single bath by electrodeposition (Fig.  5). In the nanocomposite cross-section shown in Fig. 5, subsequent CoNi NP layers separated by thin Ppy films can be clearly observed, and the presence of CoNi was confirmed by EDX (the particle was composed of 46% Co and 54% Ni). The multilayer thin films were approximately 700 nm thick on average. In all cases, it was observed that the average thickness of the different CoNi layers was similar, implying that CoNi NPs did not re-dissolve even for high Ppy deposition potentials.   Importantly, for a successful nanocomposite it was necessary to include a rest phase between CoNi NP and Ppy pulses to fabricate uniform If the rest phase was potentiostatically controlled, film detachment was observed probably due to slight bubble formation between the substrate and the growing thin film (Fig.  S3). To avoid this, the electrochemical cell was programmed to automatically switch off during each rest phase.
In order to check the degree of coverage of CoNi NPs by the Ppy layer, Ppy deposition was carried out at two different potentials, 1 V and 2 V, with similar charge densities. It was seen that while the Ppy deposition at 1 V with relatively long deposition times was not sufficient to completely cover the NPs (Fig. S4), deposition at 2V resulted in a homogenous coverage. In order to assure the homogeneity of the Ppy layer, different total deposition times (0.75 -24 seconds) at 2 V were studied. It was observed that the particles are completely covered with Ppy for all deposition times and did not get damaged during the deposition of Ppy due to re-dissolution issues. An example of such a sandwich type Ppy -CoNi nanocomposite with a thick secondary Ppy layer is shown in Fig. 6. It was also seen that thick layers of the secondary Ppy inhibited NP growth. Therefore, the secondary Ppy deposition potential was chosen to be 2 V with total deposition time as 0.75 s. The CoNi NPs were grown at -2 V for 5.8 seconds, which was chosen due to the reduced H2 evolution during NP growth. This was found to be an important point for fabricating homogeneous and smooth nanocomposite films without defects. Additionally, as can be seen in Fig. 5, it was observed that the intermediate Ppy layers showed slight differences in thickness from sample to sample prepared in the same conditions. The Ppy layer separating the particles is seen more clearly in c, d as opposed to a, b and interconnections between particle layers are more frequent in the latter case. This is probably due to reduced efficiency of the Ppy electrodeposition which can be attributed to the slow decomposition of pyrrole in the bath due to air (which is commonly observed in pyrrole) 23 . The reduction in Ppy growth efficiency may cause thinner intermediate Ppy layers. The resulting thinner Ppy layer most probably creates "hot spots" of conductivity where CoNi electrodeposition efficiency is slightly higher, resulting in the observed interconnections. However, in both cases the Ppy and CoNi NPs layers can clearly be identified.  XRD patterns showed a combination of an amorphous background arising from the Ppy layer and some reflections attributed to the metallic particles (Fig. 7). For the Co -Ppy bilayer nanocomposite the reflections belong to the hexagonal close-packed (hcp) phase, whereas for the bi-and multilayer CoNi -Ppy nanocomposites a mixture of face-centered cubic (fcc) and hcp phases is observed. The presence of both hcp and fcc phases is expected taking into account the chemical composition of the CoNi NPs which was determined by EDX to be 46 wt % Co and 54 wt % Ni 22 . The ratio between the intensities of the different hexagonal peaks implies that the particles exhibit some degree of texture. A slight shift in fcc peaks towards higher angles was observed for CoNi compared to the position of tabulated fcc Co, which proves that a solid solution was actually formed. The fcc (200) peak (which is not overlapped with any hcp peak) has been considered to calculate the cell parameter, a, of CoNi NPs. A value of a = 3.527 Å is obtained for both the bilayer and the multilayer. According to the Vegard's law [Journal of The Electrochemical Society, 157, 2010, E92-E97], this value corresponds to a Ni percentage of around 50 at%, which matches the EDX data. The TEM image shown in Fig. 7 (b) presents a single CoNi NP featuring a rounded morphology for which crystallographic defects such as staking faults are visible. The nanoparticle is made up of hcp and fcc domains as can be seen in the corresponding SAED pattern (Fig. 7 (c)). Other particles showed the hexagonal starlike morphology as mentioned before (Fig. 3 (c)).
The literature suggests that the particle shape depends on the bath composition, deposition potential, and crystalline structure of resulting particles 24 . At low deposition potentials (e.g. -1.4 V) cobalt particles can grow in a hexagonal shape due to the preferential growth of hcp phases along the edges of the particles after nucleation 24 . The fact that the hexagonal particles are most clearly observed at -2 V for CoNi (Fig. 3 (c)), is probably due to the low deposition potential and their large size. In addition, the growth of perfectly spherical particles is most likely related to the presence of acetate in the bath which is reported to cause such particle geometry 24 .

Magnetic properties
Measurements performed with vibrating sample magnetometer and (VSM) and magneto-optic Kerr effect (MOKE) prove that both bi-and multi-layered nanocomposites are ferromagnetic at room temperature. The hysteresis loops for the Co -Ppy and CoNi -Ppy bilayers are given in Fig. 8. In plane measurements for bilayer nanocomposites show that for Co (Fig. 8 (a)) coercivity (HC) and saturation magnetization (MS) are higher than for CoNi ( Fig. 8 (b)). The decrease in saturation magnetization is expected since Ni (489.5 emu/cm 3 ) has a much lower Ms than Co (1426.8 emu/cm 3 ). Nevertheless, the existence of cobalt and/or nickel oxides cannot be ruled out. Notably, the measured values of MS for Co (850 emu/cm 3 ) and CoNi (380 emu/cm 3 ), are approximately 40% of the tabulated values for hcp-Co and CoNi (taking into account the alloy composition Co0.46Ni0.54), which is in good agreement with the particle coverage observed by SEM (Fig. 3).
The difference in coercivity might be due to the presence of fcc phase in CoNi which has smaller magnetocrystalline anisotropy than the hcp phase 25 . However, other effects such as the presence of stacking faults may also play a role 26 . Additionally the coercivity values measured by VSM are lower than those measured by MOKE (Fig. 8 (a) and (b), inset). This could be due to dynamic effects since HC may depend on the sweep rate of the magnetic field 27 . However, since MOKE is a local measurement (i.e., a few tens of microns laser spot), there could also be some local inhomogeneities between different particles (e.g., size, shape, orientation or phase) which could also explain the slight differences in Hc between VSM and MOKE. Nevertheless, in both techniques Hc is observed to be larger for Co than for CoNi. Out-of-plane (i.e., polar) hysteresis loops could not be recorded for Co or CoNi -Ppy bilayer layer nanocomposites. This is thought to be related to shape anisotropy (i.e., platelet-like shapes for some particles) which would favor in-plane easy axis for the magnetization.
For the CoNi -Ppy multilayer nanocomposite, both inplane and out-of-plane hysteresis loops could be measured (Fig.  8 (c)). The shapes of the loops reveal that the multilayer nanocomposite does not have a clear easy axis (in-plane or outof-plane preferred orientation for the magnetization). The coercivity in the loop measured perpendicular to plane (Hc = 200 Oe) is slightly higher than in the one in-plane (Hc = 110 Oe). In turn, the squareness ratio is MR/MS = 0.15 for out-ofplane orientation and MR/MS = 0.25 for in-plane orientation. This lack of a clear easy axis is the consequence of the competition between the shape anisotropy arising from the flat (i.e., star shaped hexagonal NPs) geometry of the particles (which would tend to orient the magnetization in the film plane) and the magnetocrystalline anisotropy. XRD results reveal crystallographic texture along the (001) direction (c-axis) of the hcp phase and the (111) direction of the fcc phase (Fig. 7). In fact, these are the magnetic easy axis directions for hcp-Co and fcc-Ni. Hence, shape anisotropy likely favors in-plane orientation of the magnetization while magnetocrystalline anisotropy probably favors an out-of-plane easy axis. As a result, none of the two loops is particularly square.
It is also likely that magnetic dipolar (or exchange) interparticle interactions can affect the coercivity and squareness ratio of the measured samples, particularly in regions where the particles are very close to each other (dipolar) or in direct contact (exchange) [DIP]. In this sense, the reduction of coercivity and squareness ratio in the CoNi-Ppy multilayer (Fig. 8c) measured along the film plane, as compared to the CoNi-Ppy bilyaer (Fig. 8b), could also be ascribed to the stronger dipolar interactions between the particles in the former.

Electrochemical Synthesis of Nanocomposites
The nanocomposites were grown on gold substrates by using a standard three-electrode set up. A gold coated silicon substrate with an area of 0.25 cm 2 was selected as the working electrode. Silicon wafers (4-inch) were e-beam evaporated with 25 nm of titanium and 125 nm of gold, which were later diced into 1x0.5 cm chips. The counter electrode was a pure platinum wire, and the reference electrode was Ag.AgCl (3M KCl). All electrochemical experiments were done using an Autolab PGSTAT302N potentiostat. A nitrogen blanket covered the solution during electrodeposition to minimize oxidation of pyrrole due to air. Experiments were conducted with two different electrolytes; compositions are given in Table 2. Chemicals were received from Sigma Aldrich with analytical grade purity. Pyrrole was distilled prior to use as well as every 6 months and kept at 4°C, under N2. Adjustment of pH was done using para-toluene sulfonic acid. Electrodeposition for both baths was performed under N2 circulation and with constant stirring.
Initial polypyrrole layer was grown at 1 V using a total deposition time of 15 seconds. For the multi-layer nanocomposites, intermediate polypyrrole layers were electrodeposited at 2 V with the total deposition time as 0.75 seconds. Cobalt (Co) and cobalt-nickel (CoNi) NPs were grown at various potentials (-2 V, -3 V, -5 V) and deposition times (0.5 to 5.8 seconds). For both anodic and cathodic cycles pulse times were fixed at 1 ms and the duty cycle was 50%.

Characterization of Nanocomposites
SEM imaging was done with Inlens and SE2 detectors at 2 kV. FIB-SEM imaging was also done at the same conditions. Nanoparticle size was determined for each condition using the image analysis software ImageJ. Three images per condition was analysed on average and an average value for nanoparticle size was calculated.
For cutting the cross sections a gallium ion source was used with currents of 300 pA-1.5 nA. Samples were sputtered with a thin layer of gold prior to SEM and FIB-SEM. A thin carbon coating was deposited during imaging prior to FIB crosssectioning to provide smooth cuts. EDX analysis was done using 20 kV and was coupled to the FIB-SEM (Zeiss NVision 40).
X-ray diffraction patterns (XRD) were acquired on a Philips X'Pert Pro in the 40°-100° 2θ range using CuKα radiation with 0.025° of step size and 10 seconds of holding time.
Hysteresis loops were recorded at room temperature using a vibrating sample magnetometer (VSM), from Oxford Instruments, both along the in-plane and perpendicular-to-plane directions, applying a maximum field of 4 kOe. Local magnetic measurements were performed using a magneto-optic Kerr effect (MOKE) setup, from Durham Magneto-Optics, with a maximum in-plane applied field of 1500 Oe.

Conclusions
Multilayer metallopolymer magnetic nanocomposites were fabricated by electrodeposition from a single bath. A bath containing both pyrrole monomer and metal salts (Co and CoNi) was used. The results demonstrate the first successful growth of multilayer nanocomposites using Ppy and CoNi NPs. Magnetic measurements using VSM and MOKE show that both bilayer and multilayer nanocomposites are strongly magnetic at room temperature.