Seeded Growth Synthesis of Au–Fe3O4 Heterostructured Nanocrystals: Rational Design and Mechanistic Insights

Multifunctional hybrid nanoparticles comprising two or more entities with different functional properties are gaining ample significance in industry and research. Due to its combination of properties, a particularly appealing example is Au–Fe3O4 composite nanoparticles. Here we present an in-depth study of the synthesis of Au–Fe3O4 heterostructured nanocrystals (HNCs) by thermal decomposition of iron precursors in the presence of preformed 10 nm Au seeds. The role of diverse reaction parameters on the HNCs formation was investigated using two different precursors: iron pentacarbonyl (Fe(CO)5) and iron acetylacetonate (Fe(acac)3). The reaction conditions promoting the heterogeneous nucleation of Fe3O4 onto Au seeds were found to significantly differ depending on the precursor chosen, where Fe(acac)3 is considerably more sensitive to the variation of the parameters than Fe(CO)5 and more subject to homogeneous nucleation processes with the consequent formation of isolated iron oxide nanocrystals (NCs). The r...


Introduction
In recent years, nanochemistry research has moved towards the realization of complex multifunctional heterostructured nanocrystals (HNCs), able to combine two or more different inorganic materials in a single nanosystem by a direct atomic bond at their interphase without the presence of any molecular bridge. These heterostructured systems provide new degrees of freedom to control nanocrystals (NCs) functionality. [1][2][3][4][5][6] In this context, the combination of gold (Au) and magnetite (Fe 3 O 4 ) into one nanostructure has attracted a great interest due to the enormous potential of this type of structure in a wide range of applications such as catalysis, [7][8][9][10][11] optics, 12,13 electronics, 14 sensing, 9,15 and biomedicine. 9 Indeed, the integration of both magnetic and plasmonic properties in a single nanosystem makes Au-Fe3O4 HNCs extremely promising for the improvement of the related applications and the development of new ones. 9 For example, the two functionalities can act in a synergic way, enhancing specific properties (e.g., heat release capability, catalytic activity) or serving as contrast agents allowing combination of the high spatial resolution of computed axial tomography (CAT) imaging with the high sensitivity of magnetic resonance imaging (MRI). Moreover, the study of the Au-Fe3O4 interphase and the interaction between the magnetic and plasmonic domains as a function of the morphology make these systems very interesting also from a fundamental point of view.

Au-Fe3O4
HNCs with a precise control over their size and morphology can be obtained by thermal decomposition, following a seeded-growth approach in which the iron oxide counterpart grows onto the gold NC surface. Although the Au seeds can be obtained in a previous step or in situ, the first approach usually offers a better control on the monodispersity of the obtained HNCs with respect to the one-pot reaction. The most common iron oxide precursors used for this synthesis have been iron pentacarbonyl (Fe(CO)5), 16,17 iron(III)acetylacetonate (Fe(acac)3) 18 and iron(III) oleate, [19][20][21] where the [Fe]:[Au] molar ratio, together with surfactants, solvent, heating rate and reaction time were employed to control the morphology.
Despite the great interest raised by Au-Fe3O4 HNCs, the mechanism of formation and role of the diverse synthetic parameters to achieve different morphologies are still not fully understood and are, in fact, not free of controversy. 16,17 Additionally, no comparison has been reported yet between the structural and physical properties of Au-Fe3O4 HNCs obtained from different iron precursors, since it is known the iron oxide moiety can be strongly influenced by the kind of precursor used, where, for example, some precursors and processes can give secondary uncontrolled iron phases (e.g., Fe or Fe1-xO). [22][23][24][25][26][27] The purpose of this work is to investigate the effect of the different reaction parameters on the morphology of Au-Fe3O4 HNCs obtained by thermal decomposition of iron precursors (Fe(CO) 5 and Fe(acac)3) in the presence of preformed Au NCs. In particular, the aim is to define the optimal conditions for the formation of dimers and flower-like HNCs, which are of particularly interest compared to the core-shell morphology, since these morphologies have both domains exposed. It must be noted that the heterogeneous nucleation of iron oxide onto gold surfaces is not a trivial process, because of the large lattice mismatch (~3%) between the two phases. Within this context, the seeded-growth strategy was systematically investigated by changing the iron oxide precursor (Fe(CO)5 or Fe(acac)3), Fe precursor/Au seed ratio, surfactant/Fe ratio and concentration, solvent and temperature. Fe(CO)5 has been largely reported as precursor in the formation of Au-Fe3O4 HNCs, however its use presents some issues related to toxicity, stability and high cost, 28 factors that can hamper a scale-up procedure. In contrast, Fe(acac)3 has several advantages related to its stability, low cost and low toxicity. 29,30 Besides, despite Fe(acac)3 has been widely used as iron precursor for the synthesis of pure iron oxide NCs, its use in the formation of Au-Fe3O4 HNCs is rather uncommon. 18,31 The characterization of the obtained products highlights that the conditions required for the controlled formation of Au-Fe3O4 HNCs and their morphology critically depend on the iron precursor chosen, as much as on the surfactants, temperature, time and the other reaction conditions. Finally, considering the great interest of these heterostructures for their use as multifunctional nanoplatforms in biomedical applications, silica coating was demonstrated to be a feasible and effective strategy to transfer Au-Fe3O4 HNCs from organic to aqueous media.

Experimental section.
The detailed synthetic procedures for all the samples presented in this article can be found in the Supporting Information (SI).

Materials
All the samples were prepared under Ar atmosphere using commercially available reagents.
Synthesis of Au-Fe3O4 HNCs from Fe(CO)5 (HNCs_CO). A mixture of Au NCs (0.035 g in ~6 mL of hexane, from sample Au NCs_A; see SI for details on the synthesis of the Au NCs seeds), oleylamine (0.247 g, 0.92 mmol), oleic acid (0.261 g, 0.92 mmol) and 1-octadecene (20 mL) was heated up to 120 °C under vacuum in order to remove the hexane and the residual moisture. Later, Fe(CO)5 (0.05 mL, 0.37 mmol) was injected into the reaction and the temperature was increased to 315 ºC at 4 ºC/min under argon flow. Finally, the reaction was kept at 315 ºC for 50 min. The mixture was removed from the heating source and allowed to cool down to room temperature before exposing it to air. The HNCs suspension was purified by centrifuging twice the reaction mixture with a mixture of 2-propanol and methanol and further redispersion in hexane.

Synthesis of Au-Fe3O4 HNCs from Fe(acac)3 (HNCs_acac).
A mixture of Au NCs (0.011 g in ~3 mL of hexane, from sample Au NCs_B; see SI), Fe(acac)3 (0.177 g, 0.5 mmol), 1,2dodecanediol (0.508g, 2.5 mmol), oleylamine (0.53 g, 2 mmol), oleic acid (0.565 g, 2 mmol) and 1-octadecene (50 mL) was heated up to 205 °C at 5 °C/min under Ar flow (under vacuum until 115 ºC in order to remove the hexane and the residual moisture) and kept at this temperature for 2 h. Then, the reaction was heated up to 315 °C at 6 ºC/min and kept at this temperature for 2 h before removing it from the heating source and allowed to cool down to room temperature before exposit it to air. The HNCs suspensions were purified by first centrifuging twice the reaction moisture with a mixture of 2-propanol and methanol and further redispersion in hexane.
Silica coating of Au-Fe3O4 HNCs. Silica coating was performed using a reverse-micelles method. Briefly, 0.35 mL of IGEPAL® CO-520 were added to a 6 mL of suspension Au-Fe3O4 HNCs in cyclohexane (1.5·10 13 HNCs/mL) and gently stirred mechanically, resulting in a clear solution. Then, 200 µL of an ammonia solution (28-30%) and 20 µL of tetraethyl orthosilicate (TEOS) were added, shaking the suspension between both additions. The resulting mixture was kept at room temperature for 24 h and then 5 mL of methanol were added to stop the reaction.
The obtained Au-Fe3O4 HNCs@SiO2 were recovered by centrifugation of the polar phase and dispersed in a mixture of water and methanol with a 1:1 ratio.

Characterization.
Morphology, particle size and size distribution of the samples were determined by transmission electron microscopy (TEM), using a 120 kV JEOL JEM-1400 microscope. High particles, analyzing the recorded images with the ImageJ software. 32 For non-spherical shaped objects, the size measurements were performed as schematized in Figure S1.
Powder X-ray diffraction measurements were performed with a PANalytical X'PERT PRO MPD X-ray Diffractometer using CuKα radiation and operating in θ-2θ Bragg Brentano geometry at 40 kV and 40 mA. The lattice parameters of each single phase were evaluated using the software MAUD. 33 The determination of the mean crystallite diameter, dXRD, was performed on the (111) and (311) peaks for Au and magnetite phase, respectively, using the Debye-Scherrer's equation.
UV-VIS spectra were recorded using a Shimadzu UV-2100 spectrometer in the 300-800 nm range using 5 mL quartz cuvettes.
Magnetic measurements were performed a MPMSXL-7 Quantum Design SQUID magnetometer. Samples were tightly packed onto a Teflon tape in order to prevent preferential orientation. Zero Field Cooled-Field Cooled (ZFC/FC) curves were obtained by measuring the magnetization as a function of temperature, after cooling the sample in the presence (FC) or in the absence (ZFC) of an applied magnetic field (2.5 mT).

Results and Discussion
Conventionally, the process of formation of NCs is composed by two different steps: the formation of small nuclei (homogenous nucleation) and their successive growth, leading to the final NCs. 34,35 When applied to the synthesis of Au-Fe3O4 HNCs through the seeded-growth approach, the nucleation step assumes a crucial role in obtaining HNCs and in determining their final morphology. In   Figure   S4), they tend to cluster in Au and Fe3O4 rich areas, and the formed nanoparticles are clearly separate from each other by several nm due to thickness of their respective surfactant coating layers. As expected, no significant change, due to reactive etching, was observed in the mean size of the Au domains if compared to those of the starting Au NCs used as seeds ( Figure S5).
Importantly, for almost each of the Au domains (~96%) corresponds a single iron oxide counterpart and no presence of isolated Fe3O4 or a secondary population with multiple iron oxide domains was detected. The minority population of unreacted isolated Au NCs can be easily purified through magnetic separation. Starting from these reaction conditions and taking sample HNCs_CO as a reference, the effect of the different synthetic conditions was investigated in order to individuate the key parameters involved in the control of the nucleation of iron oxide on top of Au NCs and its successive growth.

Figure 2:
Representative TEM pictures of Au-Fe3O4 HNCs obtained a) from Fe(CO)5 (HNCs_CO) and b) form Fe(acac)3 (HNCs_acac) precursors. The light and dark arrows identify the Au and iron oxide domains, respectively. Note that in our bright field TEM images, darker contrast corresponds to gold, while lighter contrast to iron oxide as a consequence of their difference in electron density.
Effect of the iron precursor. In the first instance, the effects of the nature of the iron oxide precursors were evaluated. Remarkably, when Fe(acac)3 was used instead of Fe(CO)5, keeping constant all the other reaction parameters, significant differences were observed in the morphology of the formed HNCs. As it can be observed (Figure 3a), the Au NCs are surrounded by more than one iron oxide domain, with less than 20% of binary dimer HNCs and they show a more rounded shape instead of the elongated one obtained when Fe(CO)5 was used. The total HNCs mean size was found to not vary significantly with respect to sample HNCs_CO, while the iron oxide domain was found to be reduced down to ~7 nm. Undesirably, a secondary consistent population of single iron oxide NCs (~6.5 nm) was also obtained, as a consequence of a simultaneous occurrence of homogenous and heterogeneous nucleation process.
The different results obtained from the two different iron precursors suggest a different nucleation/growth regime arising from their different chemistry and reactivity. It is well known that the formation of iron oxide NCs when starting from a Fe(CO)5 precursor occurs through the initial formation of metallic Fe NCs and their posterior oxidation, 36 while iron oxide NCs are directly obtained with the Fe(acac)3 precursor. 37 Similarly, in the seeded-growth synthesis of Au-Fe3O4 HNCs starting from Fe(CO)5 the mechanism proposed is the initial nucleation of the intermediate Fe metallic phase onto the Au NCs surface, 16,38 followed by oxidation. In addition, it has been reported that Fe(CO)5 experiences a delayed sudden nucleation when injected in solution in the presence of carboxylic acids surfactants, 39 which complex the Fe(0) and prevents the rapid nucleation of Fe nanoparticles. Subsequently, as the temperature is raised the formed complex decomposes, leading to a burst nucleation. This mechanism suggests that, using Fe(CO)5 in the presence of Au NCs, the heterogeneous nucleation onto the gold surface and the successive growth of the resulting iron domain are strongly favored with respect to the homogeneous nucleation, resulting in the strong preference to form dimer-like structures.
Similarly, Fe(acac)3 is also reported to form intermediate complex, related with the complexation of Fe(III) with oleic acid and oleylamine ligands. 40 However, in this case the strong prevalence of heterogeneous nucleation over the homogeneous one does not occur, resulting in a mixture of single Fe3O4 NCs and Au-Fe3O4 HNCs, the latter ones having isotropic flower-like and incomplete core-shell morphologies.
These results suggest also that the optimal conditions for the formation of Au-Fe3O4 HNCs have to be tuned depending on the iron precursor used. In the literature, the synthetic protocols   (Figure 4a). Notably, the absence of isolated iron oxide NCs and the high amount of dimers still present suggests that even in the presence of higher iron precursor concentration, heterogeneous nucleation and growth is still preferred. As a consequence, the higher amount of iron precursor available and its affinity for the Au NC surface will lead to multiple heterogeneous nucleations on the surface of Au NCs.   (Table S1).
In particular, high resolution TEM images (Figure S2 c-d and S3 c-d) show that both are composed of multiply twinned NCs, a particular feature typical for Au NCs obtained from reduction oleylamine in organic solvents. 17,46 In the case of Fe(CO)5, an increase of the HNCs size (from ~22 nm to ~28 nm) and variations in HNCs morphology were observed moving from Au NCs_A (Figure 5a, left) to Au NCs_B (Figure 5a, right). On the other hand, in the case of Fe(acac)3 no significant differences were observed in the obtained HNCs (Figure S7), even if the product obtained from Au NCs_A ( Figure S7a) shows the presence of a secondary population of single iron oxide NCs of ~ 9 nm.
These results indicate that the presence of 1,2-dodecanediol in the coating layer of the Au NCs has a significant effect on the reaction, with different results depending on the iron oxide precursor considered. Note that one could expect that amines will compete successfully for the Au NC surface. However, as lability of functional groups decreases with increasing molecule chain length, the short di-alcohol can thus compete efficiently for the longer amine terminated  observed for all the ratios tested, however, the product with the best characteristics in terms of morphology and size distribution was obtained with a ratio of 5. The obtained data shows how the degree of protection of the seed NCs surface, a dynamic coating strong enough to avoid aggregation but weak enough to allow the seeded-growth process, can be modulated by varying the composition and the amount of surfactants in the reaction mixture, as much as the temperature (vide infra). Indeed, both oleylamine and oleic acid are necessary for obtaining well- Effect of the reaction temperature. Another way to control the seeded-growth nucleation process is acting on the reaction temperature. With the decrease of temperature in the presence of Au NCs seeds, it has been proposed that the heterogeneous nucleation will be favored at the expense of the formation of new nuclei, a process that requires the overcoming of a greater energy barrier than growth. 47 In order to investigate this aspect, the effect of the decrease of the reaction temperature was investigated for both precursors. Unexpectedly, in the case of Fe(acac)3, taking as reference a total volume of 20 mL and 0.15 mmol of iron precursor (Figure   6a, left), the decrease of the temperature reaction from 315 to 290 ºC led to homogenous nucleation of Fe3O4 NCs, with the complete absence of Au-Fe3O4 HNCs (Figure 6a, right). The observed data suggests that in these conditions the temperature was not high enough for the growing iron oxide phase to overcome the energy barrier imposed by the oleylamine coating the Au NCs surface, leading to a complete lack of heterogeneous nucleation. To confirm this hypothesis, the reaction was repeated reducing the concentration of free oleylamine from 2 to 0.2 mmol (Figure 6b). In these conditions, the process of homogenous nucleation was almost completely inhibited and a population of flower-like HNCs of ~37 nm was obtained (Figure 6b).
Furthermore, the complete removal of free oleylamine from the reaction mixture has a detrimental effect, leading to the coalescence of Au NCs (Figure 6b), indicating that this molecule has a key role in the stabilization of Au NC and controlling access to its surface.
Moreover, as the amount of free oleylamine is reduced, a modest increase of the mean size of the gold domain of HNCs was observed with respect to the original Au NCs, that can be ascribed to Ostwald ripening, to a further reduction of Au precursor or to atomic rearrangement occurring during the reaction at high temperature.
In this regime of low oleylamine concentration (0.2 mmol), a further reduction of the reaction temperature to 270 ºC leads to a double population of single Au and small (~5 nm) iron oxide NCs (Figure 6c) evidencing again the lack of sufficient thermal energy to overcome the surfactant layer (as temperature decreases the NC-surfactant bond hardens and the equilibrium between surfactant molecules at the surface of the NCs and in solution is slowed down).
Interestingly, using a solvent with higher coordination capacity for the Fe-carboxylate complex such as benzyl ether (Bz2O), flower-like HNCs with a total mean size around 37 nm (Figure 6c) could be obtained. Besides, the use of a lower boiling point solvent such as phenyl ether (Ph2O, b.p. 260 ºC) makes the iron oxide domain smaller (Figure 6c)  Primarily, a parameter that is expected to control the mean size of the iron oxide phase is the amount of iron precursor. However, in the case of Fe(CO) 5 no significant differences in the mean size of the obtained HNCs with dimer morphology (~20 nm) were observed when the [Fe]: [Au] ratio was varied from 1.6 to 9.3: the mean size of the iron oxide domain was found to be constant around 10 nm. The principal effect of the use of an higher [Fe]:[Au] ratio was the rising of a secondary population (up to 36%) of HNCs with more than one iron oxide domain, with total mean size slightly larger (~26 nm) (Figure 4a). In the case of Fe(acac)3, the higher tendency to HNCs with dimer morphology synthesized using different precursors (HNCs_CO and HNCs_acac). In both cases, two phases were identified, matching the reference patterns of gold (PDF-00-004-0784) and of magnetite (PDF-00-019-0629) or maghemite (PDF-00-039-1346), and no evidence of any additional inorganic phase was observed. The lattice parameters, a, reported in Table 1, were found to be in excellent agreement with the reference values of magnetite (8.3919Å) and gold (4.0786Å). The mean crystalline size of Au domains (Table 1), estimated with the Scherrer analysis on the (111) diffraction peak was found to be smaller with respect to the mean size obtained from TEM: this behavior was already observed in the case of the Au NCs seeds (see SI) and can be ascribed to the polycrystalline multiple-twinned structure of nanosized gold. On the other hand, in the case of the iron oxide domains, due to their irregular shape, the direct comparison between the obtained values is more complex: the mean crystalline size found for the magnetite phase, estimated on the (311) diffraction peak, was found to be higher with respect to the TEM reported mean size for both samples.  dTEM: HNCs average diameter and standard deviation obtained from TEM analysis; a, dXRD: lattice parameter and crystallite mean size obtained from XRD data analysis (errors are reported in brackets); max: wavelength of the plasmonic resonance.
In order to have a better insight of HNCs structure at the nanoscale, a detailed high resolution TEM (HR-TEM) characterization was performed (Figure 8). In the two types of samples, a high crystallinity of both the Au and iron oxide domains was observed (Figure 8c,f). The gold domains showed the usual metallic gold cubic Fd ̅ m crystal lattice (although with planar defects; Figure S11), whereas the iron oxide moiety presents a cubic spinel structure (space group Fd3m) consistent with Fe3O4 (Figure 8b,e). The most commonly observed relative orientations between Au and Fe3O4 are compatible with (100)[010]Au//(100)[010]Fe3O4 epitaxial relations ( Figure   S11). This indicates that a unit cell of magnetite (lattice parameter 8.4Å) is matched by 2x2 gold unit cells (lattice parameter 4.1Å), which translates into a lattice mismatch of 2.8%. Note that relative orientations with [100]Au zone axis parallel to [112]Fe3O4 are also occasionally seen (see Figure S11). Importantly, no significant differences have been detected between the HNCs_CO and HNCs_acac samples. An Electron Energy Loss Spectroscopy (EELS) study was also carried out, with the aim to investigate the elemental composition and the homogeneity of the two different phases in the HNCs. The spectrum images (Figure 9) clearly show that for both synthesis approaches the two phases are separated by a very sharp interface without interdiffusion. Interestingly, the relative composition Fe signal/(O signal+Fe signal) is constant along the iron oxide domain (Figure 9c,i and Figure S12), indicating an homogeneous composition of the iron oxide counterpart. The analysis of the fine structure of the EELS spectra 49 (Figure 9f,n and Figure S12e) shows that they are equivalent to that of magnetite, Fe3O4, reference spectrum. 50 Additionally, no changes in the oxidation state were detected in the sample, thus, the presence of a secondary phase of metallic Fe or FeO can be excluded Figure   S12d,e).   (Table S2) consistent with the data reported in the literature for HNCs with similar size. 27 The temperature dependence of the magnetization, M(T), of both samples cooled in the absence (Zero Field Cooled, ZFC) or the presence (Field Cooled, FC) of a magnetic field of 2.5 mT (Figure S13) shows the thermal irreversibility typical of nanosized magnetic materials. The shape of M(T), with a rather flat FC M(T), is typical of strongly interacting magnetic nanoparticles. Interestingly, the blocking temperature, TB, is clearly above room temperature. Thus, TB is somewhat high compared to Fe3O4 particles of similar sizes, 51,52 even taking into account dipolar interactions. The possible origin of such enhanced TB are manifold, e.g., shape effects, strain effects (due to the lattice mismatch), or even charge transfer at the Au-Fe3O4 interface. 27,[53][54][55][56] The optical properties of the obtained Au-Fe3O4 HNCs were investigated by UV-VIS absorption spectroscopy. In the case of HNCs_CO, the spectrum shows a strong damping of the plasmon resonance peak with respect to the starting Au NCs, with no significant shift of the maximum wavelength (Figure 7c). On the other hand, HNCs_acac shows the presence of a well-defined plasmonic peak (although broader than the corresponding Au seeds) shifted towards higher wavelengths (557 nm) with respect to the starting Au NCs (519 nm). Both red shift and damping of the plasmonic peak are phenomena commonly observed in Au-Fe3O4 HNCs 16,17,21,31,57 and are related to the presence of the iron oxide in direct contact with the Au domain. Au NCs are very sensitive to the refractive index of their surroundings, 58 therefore the high refractive index and absorption of Fe3O4 can induce optical changes in the HNCs, which increase as the Au NCs surface is covered by the iron oxide. Consequently a sample with inhomogeneous iron oxide coverage will lead to both red shift and broadening of the plasmonic band. On the other hand, the high absorption of the iron oxide generates additional broadening of the plasmonic band due to the damping of the localized plasmonic resonance in the Au NCs. 56 The different optical behavior observed in HNCs_CO and HNCs_acac samples can be explained considering the differences in surface coverage of Au NCs by iron oxide. By comparing the TEM images (Figure 2) it can be noticed that in the case of HNCs_acac the Au domain is surrounded by a higher amount of iron oxide with respect to HNCs_CO. In the latter, a large portion of the Au is exposed to the external environment and the Au-iron oxide interface is less extended with respect to HNCs_acac, resulting thus in a lower red shift. Moreover, the peculiar morphology of HNCs_CO sample can favor the inter-particle interactions between different Au domains: this phenomenon can favor the near-field interaction between the plasmonic parts in the HNCs, which can induce additional broadening of the plasmonic peaks.

Silica coating
For the application of Au-Fe3O4 HNCs in the biomedical field (e.g., as contrast agents for imaging or as multifunctional heat mediators for hyperthermia) a successive step of transfer and stabilization in water media is needed. For this purpose, a simple, solid and robust strategy is the encapsulation of the HNCs in amorphous silica. The dimer Au-Fe3O4 HNCs were thus transferred to water using a reverse-emulsion method 57,58 (see Experimental Section). The Si precursor (TEOS) is hydrolyzed around each HNCs inside reverse micelles, which have the role in controlling the size of the final coating. As it can be observed in Figure 10, the Au-Fe3O4 HNCs were successfully encapsulated by a spherical silica shell, reaching an overall diameter of around 60 nm, without any alteration of HNCs morphology and size.

Conclusions
In conclusion, well defined Au-Fe3O4 HNCs (dimers or flower-like morphologies) were obtained by seeded-growth on preformed oleylamine-capped Au NCs using two different iron precursors, Fe(CO)5 and Fe(acac)3. Depending on the iron precursor used, the reaction conditions have to be varied in order to promote the heterogeneous nucleation and, thus, to obtain a single type of HNCs. The results suggest that a different nucleation/growth regime occurs in relation to the different chemistry and reactivity of the precursors. The synthetic strategy based on the thermal decomposition of Fe(CO) 5 precursor was found to be reproducible, robust and effective for obtaining monodisperse Au-Fe3O4 HNCs with dimer morphology, with a single nucleation of iron oxide domain for each Au seed. The study on the reaction parameters suggests that both oleic acid and oleylamine play an important role in the formation of well-defined HNCs and their amount with respect to iron helps in controlling the size distribution, with an optimal [surfactant]:[iron] ratio of 5. The increase of the [Fe]:[Au] ratio was found to have effect on the morphology of the obtained HNCs, leading to a mixture of dimer and flower-like HNCs.
Moreover, it was observed that a minimum reaction temperature, which depends on the kind of solvent used, is required in order to observe the formation of HNCs.
On the other hand, the reaction performed using Fe(acac)3 as precursor was found to be much

ACKNOWLEDGMENT
The authors acknowledge funding from Generalitat de Catalunya through the 2014-SGR-1015,