Magnetic Interaction of Multifunctional Core–Shell Nanoparticles for Highly Effective Theranostics

The controlled size and surface treatment of magnetic nanoparticles (NPs) make one-stage combination feasible for enhanced magnetic resonance imaging (MRI) contrast and effective hyperthermia. However, superparamagnetic behavior, essential for avoiding the aggregation of magnetic NPs, substantially limits their performance. Here, a superparamagnetic core–shell structure is developed, which promotes the formation of vortex-like intraparticle magnetization structures in the remanent state, leading to reduced dipolar interactions between two neighboring NPs, while during an MRI scan, the presence of a DC magnetic field induces the formation of NP chains, introducing increased local inhomogeneous dipole fields that enhance relaxivity. The core–shell NPs also reveal an augmented anisotropy, due to exchange coupling to the high anisotropy core, which enhances the specific absorption rate. This in vivo tumor study reveals that the tumor cells can be clearly diagnosed during an MRI scan and the tumor size is substantially reduced through hyperthermia therapy by using the same FePt@iron oxide nanoparticles, realizing the concept of theranostics.

The experimental setup for magnetically induced hyperthermia included a copper coil cooled with circulating water, and a resonant RLC circuit producing an AC magnetic field up to 18.8   kA/m at 630 kHz, which could increase the temperature of magnetic nanoparticles in water.
By using the optical fiber thermometer to probe temperature in the center of the sample, the specific absorption rate (SAR) of superparamagnetic material is deduced from the initial linear rise of the plot of temperature versus time, T/t (Figure 1c in the manuscript) and the heat capacity of the sample, normalized to the mass of magnetic material.The nanoparticle concentration in water was 1 mg/ml.SAR can be expressed as Equation S1 [1,2,3] : where C is the volumetric specific heat capacity of water, 4185 J/(kg•K).From the experimental data, commercially available Fe 3 O 4 nanoparticles (Resovist) and our cubic IONPs show SAR values of 0.39 kW/g and 0.92 kW/g, respectively, while FePt@IONPs exhibits a value of 1.21 kW/g.S2: T2*-weighted MR images of FePt@IONPs.  Figure S3 (a) The r 2 relaxivity of mPEG-FePt@IONP is 343 mM -1 s -1 .(b) The rates of temperature increase is 0.268℃/s for mPEG-FePt@IONP.
To improve bio-compatability of NPs, we change the surfactant CTAB to mPEG.The reuslts shown in Figure S3 reveal that no significant difference between CTAB and mPEG coated FePt@IONPs on the enhancement of MRI (r 2 ) and hyperthermia (SAR).The r 2 relaxivities of CTAB-FePt@IONP and mPEG-FePt@IONP are 360 and 343 mM -1 s -1 , respectively.The rates of temperature increase are 0.290℃/s and 0.268℃/s for CTAB-FePt@IONP and mPEG-FePt@IONP, respectively.The corresponding SAR values are 1.21 KW/g and 1.12 KW/g, respectively.
S5: Cytotoxicity evaluation of FePt@IONP NPs NPs.The DLS results reveal that the mean size of FePt@IONP is 15.5 nm and the size of NPs ranges from 11.7 to 20.5 nm (Figure S6a).The mean size and range of IONPs are slightly larger than those of FePt@IONP.No significant changes of mean size of both FePt@IONP and IONPs for 24 hrs, observed by DLS, indicating good stability of FePt@IONP and IONPs without aggregation (Figure S6b).S7: Exchange-coupled properties of FePt@IONPs.hard FePt core [4,5] Figure S8.Imaginary susceptibility as function of frequency for IONP (red circle) and FePt@IONP (blue triangle).Using the Cole-Cole model with two symmetric peaks (gray and green dash line), we can fit the AC susceptibility curve of IONPs.Only one symmetric peak is needed to fit the AC susceptibility curve of FePt@IONP.
The information on interaction behavior can be obtained from AC susceptibility measurements χ"(ω) by using empirical models.The symmetric χ"(ω) can be conveniently represented by the Cole-Cole expression: [6,7] .
where χ 0 =NV 2 M 2 /K B T is the static susceptibility of the sample comprising N monodisperse particles of volume V with a saturation magnetization M. The single relaxation time indicates the monodisperse particle.The AC susceptibility curve of FePt @IONP exhibits a symmetric fitted relaxation time isτe = 1.3×10 -5 sec.On the other hand, the AC susceptibility of IONP exhibits a structure with features fitted by two relaxation times (τe1=2×10 -5 sec. andτe2=1.8×10 - sec.).
According to ref. [8], the effective relaxation time of ~10 -5 sec corresponds to the particle size in the range of 15 to 20 nm.The larger particle or cluster size leads to the longer relaxation time.Our TEM images shown in Figure 2 of manuscript reveal that both of IONPS and FePt@IONPs have the particle sizes in this range and no obvious difference of size distribution for those two samples.Therefore, the longer relaxation time of IONPs (τe2=1.8×10 - sec) may suggest that a strong dipole interaction exists among IONPs, leading to lager magnetic clusters.(FePt:7×10 6 J/m 3 ) were used for simulation parameters.The damping parameter of α= 0.5 was used to reach the equilibrium remanent state rapidly.The direction of the magnetic induction is indicated by black arrows.The OOMMF simulations for the remanent state reveal that the domain structure of core-shell FePt@IONP nanocubes exhibits a vortex state in each nanoparticle, analogously to the observed electron holography fringe patterns [9,10,11] .The induction state of IONPs shows strong interaction between adjacent nanocubes.

Figure S10
. Shape evolution of magnetic nanoparticle assembly with or without DC magnetic field of 0.47 T by using optical microscopy.
Under the application of DC magnetic field, the magnetic particles can be assembled into different configurations [12,13,14,15] .In our case, we set the DC magnetic field equal to 0.47 T.
For IONPs: (i) well-dispersed nanoparticles without magnetic field; (ii) when the magnetic field of 0.47T was applied for 30 seconds, the chain clusters are formed and aligned along the magnetic field direction; (iii) when the magnetic field was turned off, the chain disappeared and a few clusters appeared in the aligned directions; (iv) at 30 seconds after turning off the magnetic field, the IONPs are re-dispersed.If water is slightly disturbed, IONPs can be welldispersed in the solution, as shown in (i).
For FePt@IONPs: (v) the well-dispersed nanoparticles without magnetic field; (vi) when the magnetic field of 0.47T was applied for 30 seconds, the chain clusters are formed and aligned in the same direction as the field; (vii) when the magnetic field was turned off, the chain was still present but with distorted shapes; (viii) at 30 seconds after turning off the magnetic field, short-range chains remained.If water is slightly disturbed, FePt@IONPs can be welldispersed in the solution, as in (v).
S 11: The AC susceptibility spectra for IONPs and FePt@IONPs in solution with or without a magnetic field Figure S11.AC susceptibility spectra at 25 KHz of cube IONPs and FePt@IONPs with and without the DC magnetic field of 0.47T.The concentration of the sample is 0.1mg/ml.Here, the DC field is turned on and off at the 20 th and 30 th sec, respectively.With the applied filed, the nanoparticles aggregated, leading to reduced susceptibility.The reversible spectra indicate that both IONPs and FePt@IONPs can be re-dispersed after removing the DC field.
When the DC field is applied, the AC susceptibility of IONPs and FePt@IONPs in solution is reduced, as shown in Figure S11.When the field is removed, in both cases the AC susceptibility is raised to their original values.AC susceptibility can be reduced due to the formation of magnetic clusters [16] .When DC magnetic field was applied, aggregation of nanoparticles may occur, leading to a significant reduction in AC susceptibility.After removal of the DC field, the raised AC susceptibility indicates that the nanoparticles are well dispersed in the solution again.We consider a number of protons (Np) moving in a system with cubic Ns (Fe 3 O 4 or FePt@Fe 3 O 4 ).The dephasing of the proton moment S(t) is given by [17,18] : is the angle with respect to the z direction.

γ (S4)
γ is the gyromagnetic ratio of hydrogen (2.67×10 8 rad/sT).When calculating the net magnetic field B due to the presence of nearby nanoparticles, only particles within distance of the proton are considered.The net magnetic field B can be expressed as: where µ 0 is the permeability of free space (4π×10 -7 H/m), M is the magnetization of the particler is the distance from the particle center, θ is the angle with respect to the z direction and a is the size of the cubic nanoparticles.When the echo spacing is small, S(t) can be expressed by (S6) where t is the overall time from the excitation to a particular echo we are measuring.S(t) has an exponential decay form with time.
Because of computational time, the simulation results shown in Figure 5 were done for small structures with a limited number of nanoparticles (total ~1500 nanoparticels).Our results illustrate the importance of the local configuration not just the global packing fraction.The relaxation rate depends also on the number of particle per cluster, on the local cluster density and the spacing between clusters.We performed also simulation for larger structures to ensure magnetic nanoparticle configuration more similar to experimental samples.The results are shown in Figure S12.
The configuration is set for the discontinuous and continuous chain structure, with each structure containing 3636 particles (total ~ 363600 nanoparticles).Due to large structures, the calculations are limited to a packing density of 0.001.From the curve fitting in the equation S6, the R 2 (R 2 =1/T 2 ) of continuous chain is 868 s -1 , higher than discontinuous chain (188 s -1 ) in the same packing density (Figure S12).Qualitatively the behavior is the same as for small clusters, in terms that the continuous chain has a larger R 2 than the discontinuous chain.
Quantitatively, the values are smaller in comparison with the small structure system for packing density of 0.001 (Figure 5).By tuning the temperature of synthesis process, we can change the core-shell particle size.
With increased size (17.8 nm) of superparamagnetic core-shell structure, we can further improve the relaxivity to 411.3 mM -1 s -1 , as shown in Figure S13.Although when the size is increased, the uniformity of NP becomes more difficult to control, our approach can still reach even higher r 2 by further optimization.

Figure S1 .
Figure S1.MR images for kidney and Liver of mice before and after intravenous injection of FePt@IONP.

Figure S2 .
Figure S2.The XRD data of (a) IONPs with randomly packed, (b) IONPs with cube surface parallel to the substrate and (c) FePt@IONPs with cube surface parallel to the substrate, X-ray photoelectron spectrum (XPS) of (d) Fe 3 O 4 nanocubes, and (e) FePt@Fe 3 O 4 nanocubes

Figure S4 .
Figure S4.Effect of different concentration of mPEG-coated FePt@IONPs on cell viability

Figure S7 .
Figure S7.Exchange-coupled properties of core-shell FePt@IONPs were investigated by using M-H curve measurements with a superconducting quantum interference device (SQUID).(a) M-H curve of FePt@IONP at 300K and 5 K and M-H curve of IONP at 5 K .(b) M-H curve of FePt nanoparticles at 5 K.The red curve shows its superparamagnetic nature with zero coercivity at 300K.FePt@IONPs (grey curve) show larger coercevity than IONPs (green curve) at 5 K.

S 9 :
Figure S9.The OOMMF simulation for three-dimensional visualization of the magnetic moment within an assembly of four nanocubes.(a) Four Fe 3 O 4 nanocubes form a flux closure state among nanocubes and (b) FePt@IONPs show that the presence of the core stabilizes the vortices in each cube.

Figure S12 .
Figure S12.Proton dephasing signal S for two systems: continuous chains (blue curve) and discontinuous chains (red curve).The proton dephasing signal for large structures of continuous chain of 3636 magnetic nanoparticles (6x1x606 particle in xyz direction) and discontinuous chains of 101 magnetic clusters with each composed of 36 magnetic nanoparticles.
To compare the simulation results with the experimental data shown in Fig1a, we extrapolate the R 2 value to 2mM Fe concentration (corresponding to packing density of 0.001) with the same linear relationship, the estimated R 2 values are 727 s -1 , 260 s -1 for FePt@IONPs continuous chains and IONPs discontinuous chains.On the other hand, the simulated values for the small structure are 1231s -1 and 415s -1 , respectively.For larger structures, the simulation results are closer to the real case.