Tunable particle-agglomeration and magnetic coupling in bi-magnetic nanocomposites †

A set of non-stoichiometric Zn–Co-ferrite nanoparticles (NPs) was prepared by thermal decomposition of metallic complexes, in the presence of oleic acid, and, after a ligand-exchange process, was coated by a hydrophilic surfactant: these NPs were used as seeds in a sol–gel self-combustion synthesis to prepare nanocomposites (NCs) with a fixed weight ratio. Our focus here is the development of an efficient synthetic approach to control the magnetic coupling between a hard-magnetic matrix (Sr-ferrite) and NPs. The physico-chemical synthetic conditions (temperature, pH, colloidal stability) were optimized in order to tune their effect on the final particles’ agglomeration in the matrix. We demonstrate that our synthetic approach is a novel way to produce strongly magnetically coupled NCs, where the final extrinsic properties could be tuned by controlling (i) the agglomeration of seeds in the matrix and (ii) their elemental doping.


Introduction
Bi-magnetic nanocomposites (NCs) have gained a lot of interest over the last years, due to the possibility to finely control and modify their features at the nanoscale by combining two prototypical phases. 1 The novel properties arise from interfacial physical effects, owing to the coupling interaction between the oxides' interfaces which can be tuned via structural and morphological modification.3][4][5][6] The main challenge is to design advanced synthetic approaches to obtain homogeneously-dispersed systems at the nanoscale.Among the several techniques to synthesize nanoparticles (NPs), sol-gel chemistry has emerged as an advantageous way to prepare transition-metal oxides from homogeneous solutions containing the precursors, 7 and found application in a wide range of nanomaterials (e.g., thin-films, 8 monodisperse nanoparticles, 9 nanocomposites 10 ).The main advantage of the sol-gel method is to enable the complete conversion of precursors which is normally prevented in solid-state methods, a solid being limited by mass transport. 11A possible way to tune the sol-gel process, relies upon the choice of molecular species for the starting solution, as the key to controlling the morpho-structural features of the products, and thus designing novel systems.6][17] Furthermore, combining two or several magnetic phases was demonstrated to be an efficient way to achieve superior magnetic performances. 5,18,19Within this framework, the need for sharp interfaces with large contact areas to maximise the magnetic interactions between particles has required the development of more advanced synthetic strategies to control particle agglomeration, and thus morpho-structural features.

Synthesis of Co x Zn y Fe 3À(x+y) O 4 seeds
The synthesis of Co x Zn y Fe 3À(x+y) O 4 seeds NPs of 6(1) nm was carried out using standard thermal decomposition of metallic complexes in a high-boiling solvent under an inert atmosphere in the presence of stabilizing surfactants (see Section 1.1 of ESI †). 33he Fe/Co/Zn precursors ratio has been chosen to obtain Zn 0.4(1) -Co 0.5(1) Fe 2.1(1) O 4 and Co 1.0(1) Fe 2.0(1) O 4 (referred to as CZFO and CFO, respectively).The amount of Zn was chosen since Co-Zn ferrites from thermal decomposition synthesis are expected to increase their saturation with Zn 2+ up to Zn 0.4À0.5 Co 0.6-0.5 Fe 2 O 4 (see also Fig. S2, ESI †). 25,26

Synthesis of SrFe 12 O 19 matrix
The strontium ferrite SrFe 12 O 19 matrix (SFO) was prepared by a sol-gel combustion process. 12,28The pH effect upon the final compound was evaluated by preparing SFO at pH = 4, 7 and 10. (see Section 1.2 of ESI †).

Synthesis of bi-magnetic NCs
The sol-gel self-combustion synthesis was slightly modified to include the seed NPs in the SFO matrix as follows (Fig. 1).Firstly, a ligand-exchange process has been performed to replace the hydrophobic coating (oleic acid) with a hydrophilic layer (di-hydro caffeic acid, DHCA) using tetrahydrofuran (THF) (see procedure described in Section 5 of ESI †). 34After dissolving the precursors of SFO in a Teflon beaker and adding the citric acid, the seeds NPs (previously stabilized in a basic NaOH aqueous dispersion) are added to the dispersion while adjusting the pH to the final desired value (pH = 7 and 10 were considered in this study).After the mass correction accounting for the coating of the seeds by thermogravimetry (TG), the final estimated weight fraction of seeds added to the synthesis with respect to SFO is around 7 wt%.Next, it is heated in a silicon oil bath to 80 1C for B4 h to form a gel, and then let it quickly dry for 1 h at 120 1C.Next, the dried gel is ground to guarantee the homogeneity of the phases and burnt in a pre-heated oven at 300 1C.The resulting powders are ground and annealed at 850 1C for 4 h under air with a ramp of 5 1C min À1 .We refer to the final nanocomposites as NC@CZFO@pH10 (with CZFO seeds, pH = 10), NC@CZFO@pH7 (with CZFO seeds, pH = 7) and NC@CFO@pH10 (with CFO seeds, pH = 10).

Physical, structural, and magnetic characterization
The powder samples were characterized by using a Bruker D8 Advance diffractometer (solid state rapid LynxEye detector, Cu Ka radiation, Bragg-Brentano geometry, DIFFRACT plus software) in the 101-1401 2y range with a step size of 0.0131 (counting time was 4 s per step).The powder samples were grounded in an agate mortar and suspended in ethanol.A Si substrate was covered with several drops of the resulting suspension, leaving randomly oriented crystallites after drying.Rietveld analysis was performed on the X-ray powder diffraction (XRPD) data by using the FULLPROF program. 35The diffraction peaks were described by a modified Thompson-Cox-Hastings pseudo-Voigt function.A peak asymmetry correction was made for angles below 401 (2y).Background intensities were estimated by interpolating a set of points.In the model, seeds were described as cubic Fd% 3m space group.For CFO, the tetrahedral (Td) and octahedral (Oh) sites were assumed to be fully occupied, with 2/3 Fe 3+ and 1/3 Co 2+ .Since Fe and Co are indistinguishable from XRPD (electron scattering density is the same), we assume a random distribution along tetrahedral and octahedral sites.For CZFO, the cationic distribution was assumed to be (Zn 0.32 Fe 0.68 ) Td [Zn 0.08 Co 0.46 Fe 1.46 ] Oh O 4 (according to a recent study by some of the authors 25 ).SFO was modelled as a hexagonal P6 3 /mmc space-group.The composites were refined accordingly.A NIST LaB 6 660b standard was measured under the same conditions as the samples to account for the instrumental contribution to the peak broadening.
Structural and morphological investigations were performed via high-resolution transmission electron microscopy (HRTEM) and scanning transmission electron microscopy (STEM) in high-angle annular dark-field mode (HAADF).These studies were achieved on a field emission gun FEI Tecnai F20 microscope.STEM analysis was combined with electron energy loss spectroscopy (EELS) in the Tecnai microscope by using a GATAN QUANTUM energy filter in order to obtain compositional maps.For electron microscopy measurements, samples, in powder form, were dispersed in ethylic alcohol and submitted to ultrasonic agitation for one minute.Then, a drop of the suspension was deposited on a commercial holey carboncoated TEM grid and kept in the air until complete alcohol evaporation. 57Fe Mo ¨ssbauer spectra were recorded using a 57 Co/Rh g-ray source mounted on an electromagnetic transducer with velocity modulated according to a triangular waveform.The spectra were obtained at 300 and 77 K without external applied field.
The hyperfine structure was modeled by means of a leastsquare fitting procedure involving Zeeman sextets composed of Lorentzian lines.The isomer shift (IS) values were referred to that of a-Fe at 300 K. Dynamic Light Scattering (DLS) measurements were performed with a Malvern Zetasizer Nano ZSP equipped with a 10 mW He-Ne red laser (632.8 nm), operating in backscattered geometry (1731).The suspensions of NPs were analysed in plastic disposable cuvettes after filtration and sonication.More information about the measurements is reported in Section 6 of ESI.† Elemental analysis was performed by X-ray fluorescence (XRF) on the selected composites, showing only slight deviations from the target composition.Stoichiometric SFO and CFO powder samples were adopted as standards for the analysis, by checking their atomic composition (Fe/Sr = 12 and Fe/Co = 2) by inductively coupled plasma optical emission spectroscopy (ICP-OES).The estimation of spinel ferrite NPs weight fractions was carried out as reported in Section 7 of ESI.† The elemental composition of seeds NPs was confirmed by ICP-OES.
Magnetic measurements were performed at room and low temperature (T = 300 and 5 K, respectively) by using a Quantum Design superconducting quantum interference device (SQUID) and Physical Property Measurement System (PPMS) with Vibrating Sample Magnetometer (VSM) option magnetometers, which can supply maximum fields of 5 T and 9 T, respectively.To avoid any displacement of the nano-powders during the measurements, the samples were immobilized in a chemical neutral ethyl cyanoacrylate glue (no significant magnetic contribution from the glue was observed).Isothermal field-dependent magnetization loops were recorded by sweeping the field in the À5T to +5T range, and the obtained magnetization values were normalized by the weight of powders present in the sample and expressed in A m 2 kg À1 .To get information about the irreversible processes, direct current demagnetization (DCD) remanence curves were measured by applying and removing a progressively higher DC reverse field to a sample previously saturated under a (negative) field of À5T and by recording, for each step, the value of the remanent magnetization, M DCD , which is then plotted as a function of the reverse field.The corresponding switching field distributions (SFDs) were obtained from the first order derivative of the M DCD curves.We refer to the average switching field as H SW .

Tuning particle-agglomeration
The thermal decomposition method has allowed the preparation of monodisperse, highly uniform spherical CZFO NPs.A typical STEM-HAADF image of the CZFO seeds is shown in Fig. 2(a).The NPs have a spherical shape with a uniform diameter of about 5.4 nm AE 1.4 nm, in agreement with XRPD analysis (see Section 3 of ESI † for more details).The seeds are well crystallized as evidenced by HRTEM analysis showing the atomic planes extending on the entire particle size, Fig. 2(b).The inset of Fig. 2(b) is the Fast Fourier Transform (FFT) of the image and allows to identify the cubic spinel structure of the particle.The composition of the seeds was confirmed by STEM-EELS compositional maps, and a uniform distribution of Zn was observed, Fig. S4 (ESI †).From the ZFC/FC curves (see Section 4.1 of ESI †), the extracted maximum temperature, irreversible temperature, and average blocking temperature (T max , T irr and T b respectively) were found to decrease for CZFO, due to the lowering of anisotropy upon replacement of Co 2+ with Zn 2+ , as reported in Table 1.This effect is indeed accompanied by the corresponding decrease of the coercivity H C . 26,27o ¨ssbauer spectrometry confirmed the dynamic properties of CZFO, highlighting the presence of sextets in the spectrum at 77 K, which account for the larger fractions of blocked NPs (see Section 4.3 of ESI †), in comparison to the large singlet at 300 K typical of unblocked, i.e. superparamagnetic, NPs (or for particles whose relaxation time is near the measurement time scale).
This journal is © the Owner Societies 2023 The CZFO NPs underwent a ligand-exchange process which permitted their controlled partial agglomeration upon tuning the pH conditions of the starting aqueous suspension (given by the presence of citric acid as a chelating agent). 36This enabled us to exploit the chemical structure of DHCA in a specific range, pH = 7-10, for which citric acid is completely deprotonated. 37eplacing the hydrophobic coating (oleic acid) with the hydrophilic one (DHCA), did not only allow us to obtain stable suspensions of nanoparticles in water, but provided one more degree of freedom to tune the morphology of the seeds.DHCA has three dissociation constant pK a (4.5, 9.4 and 11.7), 38 allowing us to tune the degree of deprotonation by simply adjusting the acidic/basic water suspension, which in turn affects the stability of NPs in water. 38Moving from a neutral (i.e., pH = 7) environment to basic (i.e., pH = 10) environment, yields a larger deprotonation of DHCA, being well above the second acid pK a (with a corresponding total deprotonation of the carboxylic moiety, and more than 50% of the catechol), and a consequent larger electrostatic stability due to the higher repulsion between the ligand molecules, which are likely attached to the NPs surface through both the carboxylic and the catechol functional groups. 39While at pH = 7 the lower stability favours a higher degree of agglomeration, with a consequent growth of larger crystallites during the annealing step, compared to those at pH = 10.As a result, the partial agglomeration of CZFO NPs was evidenced by DLS measurements, which highlight a multiple distribution decay of the correlation function for pH = 7 (Fig. 1 and ESI, † Section 6). 40The subsequent addition to the sol-gel self-combustion route, during the pH adjustment-step, resulted in an effective way to control the initial stability of CZFO particles and thus their final size in the composite after the annealing step, which causes a different growth of this phase in the SFO matrix.
Rietveld analysis of XRPD patterns for NC@CZFO@pH10 and NC@CZFO@pH7, Fig. 3(a) and (b) respectively, reveals that the average crystallite size of the spinel fraction increases to 32(4) nm and to 57(5) nm, respectively (see Table 2); interestingly, the SFO fraction increases as well, although we have observed that the size of SFO (without seeds) seems rather insensitive to the change of pH from 10 to 7 (see Section 3 of ESI †). 42,43Therefore, it is likely to be the segregation of the two phases for the composite at pH = 7 (NC@CZFO@pH7), which is inducing the independent growth of SFO.In contrast, a more homogeneous distribution of the CZFO phase prevents the growth of both phases, demonstrated earlier. 5,44,45In addition, the presence of additional impurities, in the pattern at pH = 7, 41 suggests an incomplete growth of SFO.
The morphology of the composites has been investigated by STEM-HAADF revealing a perfect agreement with the XRPD results.In particular, Fig. 3(c) and (d) respectively show typical images of NC@CZFO@pH7 and NC@CZFO@pH10: the two nanocomposites are constituted by crystallites, strongly agglomerate in the sample obtained with pH = 7, while in that with pH = 10 they give rise to a more porous structure.Furthermore, comparing the two images, the crystallites in the pH = 7 composite are on average larger than in the pH = 10 one, as also evidenced by XRPD measurements.In such images, it is not easy to distinguish the spinel CZFO phase from the hexaferrite SFO one and for this reason, electron energy loss spectroscopy (EELS) measurements were performed in STEM mode to analyse the elemental distribution.
Two typical STEM-EELS compositional maps are reported in Fig. 4. In particular, in the top panel of Fig. 4 the composition of the NC@CZFO@pH7 sample is shown, and in the bottom panel of the same figure the composition of NC@CZFO@pH10 is reported.In both images it is possible to clearly distinguish CZFO and SFO structures, observing that in NC@CZFO@pH7 Fig. 3 Rietveld refinement analysis of the composites containing CZFO obtained at different pH for starting solutions: (a) pH = 10 (NC@CZFO@ pH10) and (b) pH = 7 (NC@CZFO@pH10) For NC@CZFO@pH7, * indicate a possible impurity. 41STEM-HAADF image of (c) NC@CZFO@pH7 and (d) NC@CZFO@pH10 showing the different sizes of the crystallites and their agglomeration.
Table 1 Composition from ICP-OES analysis, maximum temperature (T max ), irreversible temperature (T irr ), and average blocking temperature (T b ) from ZFC-FC curves, coercive field (m 0 H C ), saturation magnetization (M 5T S ) and reduced remanent magnetization (M R /M 5T S ) from magnetization (M) vs. temperature (T) at 5 K, and calculated anisotropy constant (K eff the CZFO is in partial contact with the SFO, while in the NC@CZFO@pH10 the CZFO is fully embedded in the SFO.The latter ensures an optimal connection between the two structures and consequently an excellent magnetic coupling.Furthermore, looking at the CZFO size, a reduced particle dimension of this phase is observed in pH = 10 (B35 nm) compared to pH = 7 (B100 nm) in agreement with XRPD measurements and with the effect of pH on the tendency of CZFO seeds to agglomerate.Indeed, as a further confirmation of the particle size's change, the Mo ¨ssbauer spectrum of NC@CZFO@pH10 recorded at 300 K (in Fig. 5(a)) displays overlapping mixed sextets due to SFO/CZFO particles with long relaxation time compared to the timescale of the technique (i.e., blocked particles), 46 but also a doublet due to a fraction of CZFO with shorter relaxation times, whose relative area decreases upon decreasing temperature (see Fig. 5

(b) and Section 8.4 of ESI †).
In contrast, the spectra of NC@CZFO@pH7 at both 300 and 77 K do not show any such doublet (Fig. 5(c) and (d)), as a result of high degree of agglomeration.
The Rietveld analysis does not show any secondary impurity upon introduction of Zn 2+ in the spinel-structure of CZFO, while the (a) lattice parameter increases, as expected for similar Zn-doped particles, up to 8.4091 Å (see Section 8 of ESI †). 26,47owever, the annealing step in air for the composites might induce a cationic redistribution, as hinted by the change of lattice parameter (up to 8.4162 Å): [48][49][50] this is also in agreement with the observed segregation of hematite for a reference CZFO sample, whose resulting weight fraction (B11 wt%) after annealing matches with the iron excess determined by ICP (B13 wt%).Interestingly, the extracted lattice parameter for the CZFO phase in the composite increases even slightly further (up to B8.4251 and 8.4225 Å at pH = 10 and 7, respectively) (see ESI, † Section 8.3 for details).This increase in the composite might be due to Sr 2+ substituting cations (Co 2+ or Zn 2+ ) at the interface with the spinel phase.Previous studies have observed that the addition of Sr 2+ in CFO and CZFO nanoparticles can increase the lattice parameter, 51,52 up to B8.42 Å for Sr 2+ contents as low as 0.01 (1%) in CZFO samples with Co 2+ / Zn 2+ composition close to our case (e.g., Co 0.59 Zn 0.4 Sr 0.01 Fe 2 O 4 ).Additional strain effects induced by the contact between the two phases cannot be excluded.

Magnetic coupling
Fig. 6 shows the loops field dependence of magnetization of the nanocomposites, obtained at different pH, at T = 300 (a) and 5 K (b).The main parameters extracted from the loops (saturation magnetization, M S , reduced magnetization, M R /M S , coercive field, H C , and switching/coercive fields ratio, H SW /H C ) are shown in Table 2.The loops show rather similar hysteretic behaviour, with a similar evolution of the coercivity as a function of temperature, which preserves large coercivities at 300 K, thus reflecting the dominant character of SFO over CZFO (SFO has an average size corresponding to a high coercivity range), with slight variations presumably due to the morphological differences together with the different size distributions of CZFO phase.
Besides, some significant differences may be noticed: the remanent magnetization from direct current demagnetization  Fig. 4 Left: STEM-HAADF images of NC@CZFO@pH7 (top panel) and NC@CZFO@pH10 (bottom panel) nanocomposites showing the size and distribution of the CZFO phase with respect to the SFO one.Right panels: SEM-EELS maps showing all elements, all cations but Fe, and the individual O K-edge at 532 eV (pink), Fe L-edge at 708 eV (green), Co L-edge at (DCD) experiments, M DCD , versus reverse magnetic field, 53 at pH = 7 (NC@CZFO@pH7) is crossing the field axis at values H SW (i.e.switching field) larger than H C (at both 300 and 5 K) in comparison to pH = 10, hinting at the presence of irreversible phenomena due to the less uniform particle distribution.This can be confirmed by comparing the respective differentiated curve of M DCD with respect to H, i.e. switching field distributions (SFDs) which represent the irreversible component of the susceptibility (w irr ) (inset of Fig. 5): the relative difference ((H SW À H C )/H C ) corresponds to 22% for NC@CZFO@pH7, compared to 10% for NC@CZFO@pH10, as may be generally observed for NPs systems. 54On the other hand, the harder magnetic SFO phase dominates the reversal of NC@CZFO@pH10, pointing out its role in preventing the demagnetization of the softer Zn-doped ferrite.We also point out that the irreversible component accounts for the thermally blocked NPs, so some discrepancies may arise from the different timescales in the relaxation process. 55The narrow and symmetric average SFD of NC@CZFO@pH10 suggests that the magnetic interaction here (whether or not direct exchange, i.e. dipolar-like) seems to be maximized in that system.The asymmetric SFDs of NC@CZFO@pH7 suggest the presence of more magnetic components at play. 56In addition, the average size of CZFO in NC@CZFO@pH7 is close to the single domain limit, above which the magnetization does not reverse through a uniform coherent rotation (with the corresponding nonmonotonic dependence of H C with size, observed at low temperature for the annealed CZFO, see Fig. S19, ESI †). 57,58Thus, the enhancement of remanent magnetization for NC@CZFO@pH10 (B13%) in comparison to NC@CZFO@pH7, can be ascribed to the more homogenous distribution of CZFO in the host SFO, due to the higher colloidal stability at pH = 10, which maximize the magnetic coupling between the two phases, as a result of the controlled increase of surface area at the interface. 54owever, understanding the coupling mechanisms in this kind of sintered magnets is a nontrivial task: such an effect might be hidden at room temperature, where the predominant harder magnetic nature of the Sr ferrite could mask the reversal of the small fraction of seeds.It was shown that a coupled system at room temperature might actually display features of a decoupled one at lower ones. 59To better highlight the presence of an enhanced magnetic coupling between the two phases in NC@CZFO@pH10, owing to the optimized interfaces, we compare in Fig. 6(c) and (d) the hysteresis loops at 300 and 5 K for the weighted sum (i.e., superposition)of SFO@pH10 and annealed CZFO seeds (with the corresponding weight ratio), and the experimental loops for NC@CZFO@pH10 as a reference.We should keep in mind that we expect the anisotropy of CZFO to increase as temperature is lowered, while the anisotropy of SFO decreases, 60 and thus the critical size for effective coupling changes significantly. 61Even if CZFO phases grow differently whether in the composite or individually (see Section 8 of ESI †), at 5 K it is evident (despite the low concentration of CZFO, 7 wt%) that the hysteresis for the superposition shows a double-loop response, typical of mixtures/phase-segregated Fig. 5 Mo ¨ssbauer spectra of NC@CZFO@pH10 at 300 K and 77 K (a), (b) and of NC@CZFO@pH7 at 300 K and 77 K (c), (d), respectively.systems. 5,17On the contrary, the composite shows an ordinary single-loop response at low temperatures, as a result of strong magnetic interactions between the NPs.Interestingly, we point out that at 300 K even hysteresis curve for the superposition (where no magnetic interactions are considered) displays a single loop, demonstrating the need of low-temperature measurement to fully characterize the reversal in those binary systems. 61

Effect of elemental doping on the anisotropy of NCs
In the previous section we have shown that the magnetic coupling between CZFO and the matrix can be tuned by the effect of the initial partial agglomeration.To evaluate the role of anisotropy of the seed phase on the overall composite magnetic behaviour, undoped Co-ferrite (CFO) NPs with the same size (6(1) nm) were synthesized using the same route and used to prepare a composite (NC@CFO@pH10) at pH = 10 starting conditions.XRPD analysis revealed a CFO phase of 36(2) nm, thus confirming the robustness of the pH-controlled agglomeration, as observed for NC@CZFO@pH10.Lattice parameters for SFO@pH10 are in good agreement with the other composites, while for CFO do not change significantly (see ESI †).The only difference is the small hematite impurity accompanied by unreacted SrCO 3 which resulted in a slightly lower SFO fraction (and in turn a larger CFO weight fraction (B8.5 wt%) as shown by XRF in ESI †).Nonetheless, our purpose here is to show that the strength of magnetic coupling, qualitatively evinced by demagnetization experiments, can be evaluated from the correlation between the change of magnetic anisotropy of composite and that of the seeds.
As shown in Fig. 7, the anisotropy of the composite changes according to the anisotropy of the corresponding spinel phase.At both 300 K and 5 K, the coercive field of NC@CFO@pH10 increases as expected for regular undoped CFO particles, while the saturation magnetization decreases.These results are confirming earlier observations reported in ref. 54 and 62.In the insets of Fig. 7, we note that the average switching field of NC@CZFO@pH10 is located at lower values than that of NC@CFO@pH10, as the corresponding magnetic anisotropy of CZFO is lower than that of CFO, 25 thus highlighting the effective tuning of the magnetic anisotropy of the composite. 27he effective magnetic anisotropy of the composites is calculated in the Section 8.5 of the ESI.† Interestingly, a much higher value is obtained for NC@CFO@pH10 at 5 K, suggesting a strong effect produced by the CFO seeds, as also indicated by the small secondary peak of the irreversible susceptibility at large fields (m 0 H SW B 1.2T). 54Indeed, the corresponding anisotropy field for non-agglomerated (smaller) particles is much larger.As also previously evidenced by Mo ¨ssbauer spectrometry for the NC@CZFO@pH10 case, we may expect to have smaller particles for NC@CFO@pH10 as well (owing to the colloidal stability at pH = 10) that are not coupled with the matrix and thus magnetically reverse independently (inset of Fig. 7(b)), since we observed a doublet assigned to the CZFO component in the spectrum at 300 K, which drastically decreases at 77 K (i.e.moving from thermally unblocked to blocked).In contrast, such contribution is hidden for CZFO, as the magnetic anisotropy is expected to be lower.We should point out that K eff considered here is used to compare the anisotropy for the composites.
4][65][66] Among other synthesis methods to produce composites with similar magnetic properties (see Section 9 of ESI †), [67][68][69] this design is essentially suitable for any type of particle, thus opening up the way for potential application of this pH-controlled agglomeration in other systems.

Conclusions
We have designed a seed-mediated approach to synthesize bimagnetic composites by controlling the aggregation of the preformed NPs via control over the colloidal stability inside a magnetically hard matrix.The nanocrystallites' sizes extracted from XRPD analysis match the results from STEM-EELS compositional maps, which evidence the inclusion of the NPs in the SFO.The observed enhancement of remanent magnetization for NC@CZFO@pH10 is related to the homogenous distribution of NPs in the composite, which favours a higher degree of magnetic interactions between the two phases in contact (i.e., an increase of surface area at the interface).Furthermore, we demonstrate that the change of magnetic anisotropy in the seed (i.e., CZFO) is extended to the composites, since their anisotropy changes accordingly.To conclude, we propose a simple synthetic route to obtain nanomaterials with controlled structural and well-defined morphological properties.contributed to the results discussion and revision of the article, which was written mainly by P. M. and R. M.

Table 1
reports the samples' details.
). See ESI for magnetic curves

Table 2
List of parameters: starting precursors' solution pH; lattice parameters a, b and c and crystallite sizes d for hexagonal and spinel ferrites extracted by XRPD analysis; compositional analysis by XRF; corresponding saturation at 5T (M 5T S ), and reduced remanent (M R /M 5T S ) magnetization, coercive field (m 0 H C ) and switching/coercive fields ratio (H SW /H C ) measured at 300 K and 5 K (in square brackets).*indicates the presence of an impurity (see main text for details)