From Structure to Function: Zn/Mn-Modified Maghemite as an Advanced Nanoplatform for Magnetic Hyperthermia and Radionuclide Therapy

Abstract

The development of nanoplatforms capable of efficient heat generation and stable radionuclide delivery is essential for effective bimodal cancer therapy. In this study, binary (Fe–M) and ternary (Fe–M–M′) metal oxide nanoparticles were synthesized via a polyol method optimized to produce flower-like γ-Fe₂O₃ (maghemite) structures, with M and M′ representing Zn and/or Mn. Comprehensive structural and magnetic characterization was conducted to explain the relationship between composition, defect structure, and hyperthermic performance. The analyses revealed that cation substitution induced an Fe-site vacancy, primarily at octahedral positions, leading to local structural distortions, as confirmed by powder X-ray diffraction and pair distribution function analysis. The optimized composition, with Zn/Mn/Fe = 0.040:0.182:1, exhibited the highest concentration of vacancies and structural disorder. These vacancies altered the bonding environment, enhancing magnetic interactions at tetrahedral sites while weakening those at the octahedral positions. The resulting multicore nanoflowers (20–63 nm; core size 13–18 nm) displayed strong heating performance, with intrinsic loss power ranging from 0.34 to 5.77 nHm² kg^–1. The optimized sample achieved a temperature increase of 30 °C within 2 min and a specific absorption rate of 369 W g^–1. This composition was further coated with citrate (CA) and successfully radiolabeled with ¹⁷⁷Lu, achieving a radiolabeling yield of 92.7% and excellent stability, thus forming a robust nanoplatform for combined magnetic hyperthermia and radionuclide therapy. Biological evaluation of the optimized S5 composition revealed selective cytotoxicity toward HeLa and LS174 cells, while toxicity was significantly lower to A549, A375, and normal MRC-5 cells. Citrate coating of S5 nanoparticles (S5@CA) drastically reduced their cytotoxicity across all tested cell lines (IC₅₀ > 200 μg mL^–1), confirming their enhanced biocompatibility for therapeutic applications. In HeLa cells subjected to magnetic hyperthermia, the viability decreased to approximately 84% after 30 min and 61% after 60 min of treatment, demonstrating the sustained hyperthermic effect at a controlled working temperature of 48 °C. These results underscore the effectiveness of cation substitution and vacancy engineering in tailoring the functional properties of maghemite-based nanomaterials for advanced multimodal cancer therapies.

1. Introduction

Magnetic nanoparticles (MNPs) have attracted significant attention in nanomedicine and cancer therapy due to their unique ability to generate localized heat when exposed to an alternating magnetic field (AC). This property makes them ideal candidates for magnetic hyperthermia (MH), a minimally invasive treatment that selectively raises tumor temperature to induce apoptosis while sparing surrounding healthy tissue. Additionally, MNPs have been explored as radionuclide carriers, enabling synergistic bimodal cancer therapy by integrating hyperthermia with radionuclide therapy. Among various MNP systems, iron oxide nanoparticles, particularly magnetite (Fe₃O₄) and maghemite (γ-Fe₂O₃), have attracted considerable attention due to their biocompatibility, high saturation magnetization (M _S), and chemical stability. ,

To further improve the heating efficiency of MNPs in MH, one widely adopted strategy is the substitution of Fe³⁺ ions in the spinel lattice of iron oxides with other cations, including 3d and 4f metal ions. Such substitution allows for fine-tuning of structural, microstructural, and magnetic properties, leading to enhanced magnetic anisotropy, spin canting, and heating performance. ,− Doped maghemite nanoparticles, such as those containing cobalt, zinc, or gadolinium, have been extensively studied for their tunable magnetic properties. − For example, Yang et al. demonstrated that controlled Mg²⁺ doping in γ-Fe₂O₃ significantly enhanced AC magnetic heating efficiency, with Mg_0.13-γ-Fe₂O₃ nanoparticles exhibiting intrinsic loss power (ILP) values nearly 100 times higher than those of commercial MNPs (Feridex). Similarly, Jang et al. showcased the profound impact of precise Zn²⁺ doping in (Zn _x Mn_1–x )­Fe₂O₃ nanoparticles, achieving exceptionally high M _S (up to 175 emu g^–1) at 300 K, which correlated with superior MRI contrast (r ₂ = 860 mM^–1 s^–1) and highly efficient magnetic hyperthermia (specific absorption rate; SAR = 432 W g^–1). These findings highlight the potential of compositional engineering to tailor the functional properties of iron oxide nanoparticles. Concurrently, surface coating of iron oxide-based nanoparticles with metallic nanoshells such as gold has also emerged as an effective strategy to modulate heat transfer from the magnetic core to the surrounding medium and to precisely control the temperature profile during magnetic hyperthermia, depending on the thickness of the metallic shell.

In addition to compositional tuning, cation substitution frequently induces Fe-site vacancies, particularly at octahedral positions within the iron oxide crystal lattice, which can significantly affect magnetic behavior. These vacancies disrupt magnetic interactions, alter local symmetry, and influence spin dynamics, ultimately modifying the heating efficiency under an AC magnetic field. − For instance, the removal of Fe³⁺ or Co²⁺ ions in spinel ferrites such as CoFe₂O₄ results in altered magnetic anisotropy and saturation magnetization. Moreover, substituting Fe³⁺ with Zn²⁺ in iron oxide can drive cation redistribution between tetrahedral and octahedral sites within the spinel structure, further influencing the magnetic performance. Studies employing positron annihilation spectroscopy and Doppler broadening have demonstrated that Zn²⁺ incorporation significantly modifies vacancy distribution and magnetic interactions. These findings underscore the critical role of defect engineering and precise control of the vacancy chemistry in optimizing nanoparticle performance for magnetic hyperthermia and other functional applications.

In addition to compositional and vacancy-related effects, the shape and internal structure of MNPs are also critical determinants of their magnetic behavior. Notably, flower-like nanoparticles (nanoflowers), composed of aggregated primary nanocrystals, exhibit enhanced collective magnetic interactions, which significantly increase their SAR values. Reported SAR values for such iron oxide nanoflowers can reach up to 3000 W g^–1, ranking them among the most efficient heat-inducing MNPs to date. The formation mechanism of maghemite nanoflowers, driven by the oriented growth, has been thoroughly investigated. Furthermore, small-angle neutron scattering analyses have shown that these structures undergo additional levels of aggregation, resulting in multiscale architectures that further amplify magnetic coupling and hyperthermic response.

In this study, we synthesized binary and ternary oxide nanoparticles with the general formula (Zn _x Mn _y Fe _z V _q )₂O₃, where V denotes Fe-site vacancy, using a polyol-mediated synthesis route optimized to yield flower-like γ-Fe₂O₃ structures. Our primary objective was to investigate whether the synthesis approach could maintain both the maghemite phase and its characteristic morphology upon partial substitution of Fe³⁺ ions with Zn²⁺ and Mn²⁺ across a wide compositional range. We hypothesized that introducing multivalent cations into the iron oxide lattice would facilitate the formation of vacancy-rich structures, thereby modulating the structural, magnetic, and functional properties critical to MH and radionuclide therapy. Precursor solutions were prepared in stoichiometric ratios corresponding to nominal (Zn,Mn,Fe)₂O₃ compositions. The actual elemental composition (cation molar ratio) of each sample was determined using inductively coupled plasma optical emission spectroscopy (ICP–OES). Based on the crystal symmetry confirmed through structural analysis, the general formula of the synthesized nanoparticles is expressed as (Zn _x Mn _y Fe _z V _q )₂O₃.

We systematically explored the effects of Zn and Mn substitution, coupled with controlled defect engineering, on the structural, magnetic, and functional properties of the nanoparticles. Comprehensive characterization was conducted, including X-ray powder diffraction (XRPD), pair distribution function (PDF) analysis, transmission electron microscopy (TEM), dynamic light scattering (DLS), superconducting quantum interference device (SQUID), and SAR measurements. Additionally, the nanoparticles were radiolabeled with the therapeutic ¹⁷⁷Lu, and their biological performance was assessed through MTT assays on both cancer and normal cell lines. In vitro MH experiments on HeLa cells confirmed that the CA-coated nanoparticles effectively reduced the cell viability, demonstrating their therapeutic potential. In addition to their use as heating agents, the citrate-coated nanoparticles were also assessed as carriers for radioactive ¹⁷⁷Lu to facilitate bimodal therapy.

The central innovation of this work lies in demonstrating that Fe-site vacancy, induced by Zn/Mn substitution, markedly enhances magnetic hyperthermia efficiency while maintaining excellent radiolabeling stability of the nanoparticles. These findings position the synthesized nanoflowers as promising multifunctional nanoplatforms for combining magnetic hyperthermia and radionuclide cancer therapy.

2. Experimental Section

2.1. Sample Preparation and Coating

2.1.1. Reagents

All chemicals used in this study were analytical grade and were not subjected to additional purification. Iron­(III) chloride hexahydrate (FeCl₃·6H₂O), iron­(III) nitrate nonahydrate (Fe­(NO₃)₃·9H₂O), zinc chloride (ZnCl₂), manganese­(II) chloride tetrahydrate (MnCl₂·4H₂O), diethylene glycol (DEG), N-methyldiethanolamine (NMDEA), sodium hydroxide (NaOH), sodium chloride (NaCl), nitric acid (HNO₃), dimethyl sulfoxide (DMSO), and trypsin–EDTA solution were purchased from Sigma-Aldrich (St. Louis, USA) and Merck (Kenilworth, USA), respectively. Ultrapure Milli-Q water (Millipore Co., Billerica, MA, USA) was used throughout all of the experiments.

2.1.2. Synthesis of (Zn _x Mn _y Fe _z V_q)₂O₃ Samples

An optimized polyol method was employed to synthesize (Zn _x Mn _y Fe _z V _q )₂O₃ nanoparticles. Iron chloride served as the iron precursor in an aminated polyol environment. To synthesize (Zn _x Fe _z V _q )₂O₃, a mixture of 0.2181 g of ZnCl₂ (1.6 mmol) and 0.8656 g of FeCl₃·6H₂O (3.2 mmol) was added to 64 g of a solvent medium composed of DEG and NMDEA in a 50:50 weight ratio. This mixture was vigorously stirred magnetically for 60 min. In a separate beaker, 0.512 g of NaOH (12.8 mmol) was stirred ultrasonically together with 32 g of the same solvent mixture (DEG and NMDEA) for 1 h while heating at 80 °C. After ensuring complete dissolution of sodium hydroxide, the solution was carefully transferred to a beaker containing the zinc chloride and iron chloride. The combined mixture was then stirred for another hour. The resulting mixture was sealed in a 100 mL Teflon reactor, placed in a preheated oven, and allowed to react for 12 h at 220 °C. Following the synthesis, the reactor was cooled to room temperature (RT) inside of an oven. The product was precipitated using an external magnet and washed multiple times with an ethanol and ethyl acetate solution mixed in a 50:50 weight ratio. To enhance the colloidal stability and ensure complete oxidation of the samples, an acidic treatment was performed. The mixture was mechanically agitated with 10 mL of 10% HNO₃, and then 1.374 g of Fe­(NO₃)₃·9H₂O (34 mmol) was added to the suspension and heated to 80 °C for 1 h. After being cooled to RT, the mixture was washed again with nitric acid. Finally, the nanoflowers were washed with deionized water, separated magnetically, and redispersed in pure Milli-Q water. The rest of the (Zn _x Mn _y Fe _z V _q )₂O₃ series was synthesized through a consistent process, varying only the relative amounts of ZnCl₂ and MnCl₂·4H₂O. The ZnCl₂ was systematically replaced by MnCl₂·4H₂O in different molar ratios (see Section ), ultimately resulting in a composition containing solely 0.3166 g (1.6 mmol) of MnCl₂·4H₂O for (Mn _y Fe _z V _q )₂O₃.

To ensure colloidal stability and biocompatibility, selected sample S5 (see Section ) was further surface-modified with citric acid (S5@CA), a ligand known to significantly enhance the long-term colloidal stability of ferrite nanoparticles. This effect has been previously demonstrated for citrate-coated cobalt ferrite nanostructures, which remained stable for over a year without signs of agglomeration or precipitation. Although CA was used as the modifying agent, under the experimental pH conditions (pH ∼ 7), it exists predominantly in its deprotonated form (citrate), which can act as an electrostatic stabilizer and coordinate metal cations such as Lu³⁺ through its carboxylate groups.

2.2. Instrumental Techniques

ICP–OES was used to determine the elemental composition of (Zn _x Mn _y Fe _z V _q )₂O₃ samples using a Thermo Fisher Scientific iCAP some Duo ICP instrument (Cambridge, UK). Samples were subjected to total acid digestion prior to analysis. Measurements were taken at the following emission wavelengths: 238.204 nm for Fe, 267.716 nm for Mn, and 213.856 nm for Zn.

The crystal structure of the (Zn _x Mn _y Fe _z V _q )₂O₃ samples was determined by XRPD data. Dried powder samples were examined using a high-resolution Smart Lab diffractometer (Rigaku, Japan) equipped with a Cu Kα radiation source (λ = 1.5406 Å) operating at 40 kV and 30 mA. Diffraction patterns were collected within a 2θ range of 10–110°. X-ray diffraction data was acquired at a scanning rate of 0.5° per minute with a step size of 0.02°. Phase identification of the synthesized materials was conducted using the Halder–Wagner method implemented in the PDXL2 integrated X-ray powder diffraction software (ver. 2.8.40, Rigaku Corporation, Tokyo, Japan).

For PDF analysis of subnanometer structures, synchrotron X-ray total scattering data were collected at the 28-ID-1 (PDF) beamline of the National Synchrotron Light Source II (NSLS-II) at Brookhaven National Laboratory. Fine powders of studied nanoparticle samples described in detail later, ∼100 mg each, were encapsulated in 1.0 mm diameter Kapton capillaries and sealed at both ends with epoxy glue. The beamline was configured for PDF data collection mode, utilizing a PerkinElmer 2D image plate detector (sample-to-detector distance: 206 mm, calibrated against a Ni polycrystalline standard) for rapid data acquisition with a direct-space r-resolution of 0.12 Å. However, the modest Q-space resolution of this mode limits the PDF field of view in r-space to approximately 6 nm. Data were acquired at 300 K by using an incident X-ray energy of 74 keV (λ = 0.1675 Å). The raw 2D experimental data were converted into 1D intensity vs momentum transfer (Q) patterns using the Fit2D program. These patterns were further processed and corrected following standard protocols, including the use of PDFgetX3 software, before being Fourier transformed with Q _max = 25 Å^–1 to obtain the final PDFs, G(r).

The microstructure and morphology of the synthesized samples were examined using a high-resolution transmission electron microscope (FEI, Talos F200X, Thermo Fisher Scientific, US) at an accelerating voltage of 200 kV equipped with an EDS analyzer. Sample preparation involved placing a drop of particles suspended in water onto a carbon-coated copper grid and allowing it to dry at RT for TEM observations. The TEM images were manually analyzed by using ImageJ software. The mean nanoflower particle sizes (D _NF) and their distribution were determined by measuring the largest internal dimension of 200 nanoflowers per sample, while the nanoflower core size (D _C) was estimated by measuring at least 100 cores inside of each sample. Subsequently, the data were fitted to a log–normal function to obtain the mean size, standard deviation, and index of polydispersity (PdI).

The hydrodynamic diameters of 0.1 mg mL^–1 suspensions containing the investigated samples dispersed in deionized water were measured at 25 ± 0.1 °C in disposable cells. These measurements were determined by DLS using a NanoZS90 (Malvern, UK) instrument equipped with a 4 mW He–Ne laser source (λ = 633 nm). The samples were allowed to thermally stabilize at the set temperature for 10 min prior to measurement.

To evaluate the magnetic properties, around 10 mg of pure (Zn _x Mn _y Fe _z V _q )₂O₃ powder samples were packed into polycarbonate capsules and secured with cotton wool. Hysteresis loops were then measured at 5 and 300 K using a SQUID magnetometer (MPMS-5, Quantum Design, USA) with applied magnetic fields reaching up to 50 kOe. The magnetization dependence on temperature was measured in the range from 5 to 300 K, using both zero-field-cooled (ZFC) and field-cooled (FC) regimes under a direct current (DC) magnetic field of 50 Oe.

The M _s, average particle size (D _avg), and standard deviation (σ) were calculated by fitting the hysteresis data to a weighted Langevin function, ∫g _n (d,σ,D)L(H) dD. Here, $g_n(\mathrm{d̅},\sigma ,D)$ represents a log–normal distribution (eq ).

$$g_n(\mathrm{d̅},\sigma ,D)=\frac{1}{\sqrt{2\pi }\sigma \frac{D}{\mathrm{d̅}}}\exp [-\frac{{\ln \left[\frac{D}{\mathrm{d̅}}\right]}^2}{2{\sigma }^2}]$$ 1

where D is the particle diameter, while $\mathrm{d̅}$ and σ represent distribution parameters. L(H) is the Langevin function

$$L(H)=M_{\mathrm{S}}(\coth ({\mu }_{0}H\mu /kT)-{({\mu }_{0}H\mu /kT)}^{-1})$$ 2

where μ₀ represents the magnetic permeability of vacuum, μ the magnetic moment of the particle, k the Boltzmann constant, and T the measurement temperature (eq ). The particles were represented by spheres (V = πD ³/6) with homogeneous magnetization (μ = M _S V), where M _S represents the saturation magnetization value.

Calorimetric measurements were conducted by using a commercial AC hyperthermia device (DM100 nB nanoScale Biomagnetic, Spain). Heating curves were obtained under a 252 kHz frequency with a 200 Oe AC field. For the measurements, 1 mL of colloidal (Zn _x Mn _y Fe _z V _q )₂O₃ dispersions in water (c = 5 mg mL^–1) was added to a measurement cuvette. The temperature increase was continuously tracked over 120 s using an optical fiber system.

SAR, expressed in W g^–1, is calculated using eq

$$\mathrm{S}\mathrm{A}\mathrm{R}=\frac{C}{m_{\mathrm{F}\mathrm{e}}}{\left(\frac{\mathrm{d}T}{\mathrm{d}t}\right)}_0$$ 3

where C is the heat capacity of the fluid per mass unit, in this case water, m _Fe is the concentration of the colloidal nanoparticle suspension (g L^–1 of iron), and (dT/dt)₀ is the measured temperature increase per time unit.

ILP, expressed in nHm² kg^–1, quantifies a magnetic nanoparticle’s inherent ability to absorb power when exposed to an alternating magnetic field. It is calculated using eq

$$\mathrm{I}\mathrm{L}\mathrm{P}=\frac{\mathrm{S}\mathrm{A}\mathrm{R}}{f{H}_0^2}$$ 4

where f is the frequency and H ₀ is the intensity of the applied magnetic field. Notably, ILP is an intrinsic property, meaning its value is independent of the magnetic field parameters and facilitates a comparison of MNP’s heating conversion efficiency across various magnetic fields.

2.3. In Vitro Studies

2.3.1. Cell Culture

The cervix adenocarcinoma cell line (HeLa), human colon carcinoma (LS174), human nonsmall cell lung carcinoma (A549), human malignant melanoma (A375), and a normal cell line, human fetal lung fibroblast cell line (MRC-5), were grown in RPMI-1640 medium (Sigma) at 37 °C. Media were supplemented with 10% fetal bovine serum, l-glutamine, and penicillin–streptomycin (Sigma). Media were supplemented with 10% heat-inactivated (56 °C) fetal bovine serum (FBS), 3 mM l-glutamine, 100 IU/mL penicillin, 100 μg mL^–1 streptomycin, and 25 mM HEPES adjusted to pH 7.2 by bicarbonate solution. All cell lines were obtained from the American Type Culture Collection (Manassas, VA, USA). The cells were grown at 37 °C in an atmosphere of 5% CO₂ in humidified air. RPMI-1640, DMEM, FBS, l-glutamine, and HEPES were products of Sigma-Aldrich, St. Louis, MO.

2.3.2. Treatment of Cell Lines

Based on the analyses of the physical and chemical properties of the synthesized samples, the sample exhibiting the highest heating efficiency under alternating magnetic fields was selected for in vitro studies and radiolabeling. This sample, denoted as S5, has a cation molar ratio of Zn/Mn/Fe = 0.040:0.182:1 (see Sections and 3.2).

In vitro assay for the cytotoxic activity of investigated nanoparticles (S5 and S5@CA) was performed when the cells reached 70–80% of confluence. The stock solutions (100 mg mL^–1) of the MNPs were dissolved in the corresponding medium to obtain the required working concentrations. Target cells HeLa (2000 cells per well), LS174 (5000 cells per well), A549 (5000 cells per well), A375 (5000 cells per well), and MRC-5 (5000 cells per well) were seeded into wells of a 96-well flat-bottomed microtiter plate. Twenty-four hours later, after the cell adherence, five different, double-diluted concentrations of investigated MNPs were added to the wells. The final concentrations of the MNPs were 12.5, 25, 50, 100, and 200 μ mL^–1, except for the control wells, where only nutrient medium was added. All investigated concentrations were set up in triplicate. Nutrient medium with corresponding concentrations of investigated compounds but without cells was used as a blank, also in triplicate. The cultures were incubated for 72 h.

2.3.3. Determination of Cell Survival

The effect of the investigated nanoparticles (S5 and S5@CA) on survival of the specified cell lines was determined by the microculture tetrazolium test (MTT) 72 h after addition of the compounds according to Mosmann with modification by Ohno and Abe. After incubation, the magnetic fluid was removed, and the cells were washed thrice with phosphate buffered saline (PBS) to ensure complete removal of MNPs from the wells. Thereafter, 300 μL of media solution and 20 mL of MTT solution (5 mg mL^–1 phosphate-buffered saline) were added to each well. Samples were incubated for an additional 4 h at 37 °C in a humidified atmosphere of 95% air/5% CO₂ (v/v). Then, 100 μL of 100 g L^–1 sodium dodecyl sulfate was added to dissolve the insoluble product formazan, resulting from conversion of the MTT dye by viable cells. The absorbance (A) at 570 nm was measured 24 h later. The number of viable cells in each well was proportional to the intensity of the absorbance of light, which was read in an enzyme-linked immunosorbent assay (ELISA) plate reader. To determine cell survival (%), the A of a sample with cells grown in the presence of various concentrations of the investigated compounds was divided by the control optical density (the A of control cells grown only in nutrient medium) and multiplied by 100. In each experiment, the A of the blank was always subtracted from the A of the corresponding sample with target cells. IC₅₀ is defined as the concentration of an agent inhibiting cell survival by 50% compared to a vehicle-treated control. All experiments were done in triplicate.

2.3.4. Magnetic Hyperthermia In Vitro

To examine the effect of magnetic hyperthermia, the HeLa (200,000 cells/culture dish) cells were grown on 35 mm culture dishes and then subjected to 24 h incubation with or without S5@CA, as well as with or without the presence of a magnetic field, for a duration of 60 min using an AC hyperthermia device (DM100 nB nanoScale Biomagnetic, Spain), module DM2 under a 236 kHz frequency with a 200 Oe AC field. A concentration of 1.1 mg mL^–1 of S5@CA was applied to achieve a satisfactory effect of hyperthermia under an AC magnetic field. Therefore, the cytotoxicity of this high concentration of S5@CA on HeLa cells was tested at the same time.

Briefly, the method involved seeding cells in four culture dishes numbered as (1) control, cells without S5@CA but with 60 min MH; (2) control, cells with S5@CA and without MH; and (3) cells with S5@CA and 30 and (4) 60 min MH. When the cells reached approximately 80% confluence, S5@CA was added to dishes 2, 3, and 4. Subsequently, the cells from dishes 1, 3, and 4 were exposed to an alternating magnetic field, with treatment durations of 30 min (dish 3) and 60 min (dish 4). All cells were then allowed to grow for 24 h. The samples were placed in duplicate in dishes to examine the effects after 24 h. Subsequently, the cells were washed in PBS, trypsinized, and further assessed for viability, and cell survival was determined using the MTT assay.

2.3.5. Radiolabeling of Citrate-Coated Magnetic Nanoparticles with ¹⁷⁷Lu

To achieve a stable colloidal suspension and provide functional groups for binding the radionuclide lutetium-177 (¹⁷⁷Lu), the nanoparticles with the highest hyperthermic efficiency (S5) were coated with CA. The resulting sample, S5@CA, was radiolabeled following a previously established procedure. Here, we provide a concise description of the methodology. A direct labeling approach was employed, where S5@CA was incubated with ¹⁷⁷LuCl₃ under controlled conditions. The labeling reaction was performed after the addition of 7.4 kBq of ¹⁷⁷LuCl₃ in 0.05 M HCl to 0.5 mL of a suspension of S5@CA at pH 5 and with continuous mixing at RT for 60 min. The binding of ¹⁷⁷Lu³⁺ to nanoparticles occurs through surface functional carboxyl groups of citrates, resulting in a stable radiolabeled product. Due to the high stability of magnetic nanoparticles in suspension, determination of the radiolabeling yield (RLY) using magnetic separation was not feasible. Instead, the RLY was determined by determination of the radiochemical purity of the ¹⁷⁷Lu-labeled S5@CA using thin-layer chromatography with 0.1 M acetate buffer as the mobile phase and silica gel plates as the stationary phase. In this system, unbound ¹⁷⁷Lu³⁺ migrates with the solvent front (R _f = 0.8–0.9), whereas ¹⁷⁷Lu-S5@CA remains at the origin (R _f = 0.0–0.1). After development, the strips were air-dried and placed on a radiochromatography scanner (LabLogic Systems Limited, UK) to quantify the radioactivity. The in vitro stability of ¹⁷⁷Lu-S5@CA was assessed at RT in saline (0.90% w/v of NaCl) and human serum albumin solution (HSA) designed to mimic physiological conditions over a period of 72 h postlabeling.

3. Results and Discussion

3.1. Chemical Composition Analysis

The precursor solutions were prepared using stoichiometric ratios corresponding to the nominal (Zn _x Mn_1–x )­Fe₂O₄ compositions, where x ranged from 0 to 1. However, due to the known redox flexibility of Fe and Mn ions, as well as synthesis-specific conditions, deviations from ideal stoichiometry were anticipated. The actual elemental composition of the synthesized samples was determined using ICP–OES. Significant deviations were observed between the nominal and measured cation ratios, suggesting the formation of nonstoichiometric spinel structures incorporating Fe^2+,3+ and Mn^2+,3+ ions, along with Zn²⁺. These deviations are commonly accompanied by the formation of cationic vacancies, particularly at Fe octahedral sites, as a mechanism for maintaining charge neutrality. These findings are consistent with results previously reported in the literature, confirming the stable oxidation state of Zn²⁺ and the multivalence of Fe and Mn ions. Based on ICP–OES results and structural analyses, the empirical formulas of the synthesized nanoparticles, denoted as S1–S6, can be expressed as (Zn _x Mn _y Fe _z V _q )₂O₃, where V represents an Fe-site vacancy. The measured cation ratios for the samples are as follows: S1–Zn/Fe = 0.271:1; S2–Zn/Mn/Fe = 0.160:0.056:1; S3–Zn/Mn/Fe = 0.126:0.098:1; S4–Zn/Mn/Fe = 0.074:0.139:1; S5–Zn/Mn/Fe = 0.040:0.182:1; and S6–Mn/Fe = 0.263:1.

3.2. Structural and Microstructural Analysis of (Zn _x Mn _y Fe _z V_q)₂O₃

To investigate the influence of cation substitution on the structural integrity and crystallization behavior of the synthesized nanoparticles, comprehensive structural and microstructural analyses were performed. The crystal structures of the samples were confirmed through XRPD. The diffraction patterns (Figure ) showed that all Bragg reflections could be indexed to the magnetite structure (space group Fd $\mathrm{3̅}$ m) or maghemite structure (space group P4₃32 or P4₁32). Notably, the (210) and (211) reflections typical of maghemite were not observed, which is a common characteristic of nanostructured maghemite reported in the literature. , Furthermore, no reflections indicating the presence of other phases or impurities were detected. The crystal structures were refined using XRPD data with PDXL2 software. Due to the similar atomic scattering factors of zinc, iron, and manganese, the occupancy of cations in nonequivalent crystallographic positions was not analyzed, as the refinement procedure is insensitive to these minimal differences in scattering contrast using XRPD data. Notably, Da Silva et al. demonstrated that the combined application of Mössbauer and Raman spectroscopy serves as a powerful method for evaluating cation site occupancy in nanosized ferrites. Implementing this dual spectroscopic approach in a future study could enable a detailed assessment of cation distribution and provide deeper insights into the structural organization and functional properties.

1.

Comparison of observed (dots) and calculated (solid line) intensities for S1–S6 samples. The difference pattern appears below. The difference pattern appears in blue at the bottom.

The structural refinement was initially performed using the magnetite model, which yielded reasonable agreement between the experimental and calculated values. However, certain Bragg reflections, particularly (200) and (222), were not well described by the magnetite model. These discrepancies can be attributed to the presence of cation vacancies in the octahedral sublattice, leading to local distortions and symmetry reduction characteristic of the maghemite phase.

Refining the structural model to maghemite resulted in an excellent match between experimental and calculated data for all reflections (Figure ), suggesting that the synthesized samples mainly crystallize in the maghemite structure. This structural preference is strongly influenced by the formation of Fe vacancies, which stabilize the maghemite phase and induce specific local distortions. This observation is consistent with the polyol method, which typically favors the formation of maghemite-like structures in iron oxides.

The results indicate that the maghemite structure is stabilized despite the incorporation of zinc and manganese ions, which partially substitute for iron. The partial substitution likely leads to the appearance of cationic vacancies, predominantly in octahedral sites, to maintain charge neutrality. These vacancies, along with the specific ionic radii and oxidation states of Zn²⁺, Mn^2+,3+, and Fe^2+,3+, contribute to the observed structural preferences. This finding underscores the versatility of the maghemite framework in accommodating mixed cation compositions while preserving its defect spinel character.

The refined lattice parameter values (a) are presented in Table . These values depend on the distribution of cations across crystallographic sites, where the same ion can have different radii in tetrahedral and octahedral sites. The possible presence of Fe vacancies leads to a slight contraction of the lattice parameter, particularly in samples with higher Zn and Mn contents, where substitution-induced defects further enhance this effect. Due to these factors, as well as variations in chemical composition and cation distribution, a linear dependence of the lattice parameter on ion concentration is not expected (Table ). The crystallite sizes of the samples, determined through refinement, are also provided in Table , with values ranging from 13.4 to 18.4 nm.

1. (Micro)­structural Parameters of the Investigated Samples.

	sample
(micro)structural parameter	S1	S2	S3	S4	S5	S6
lattice parameter, a [Å]	8.4255(4)	8.4272(6)	8.4247(7)	8.4217(6)	8.4199(6)	8.4333(7)
crystallite size [nm]	13.4(1)	13.6(1)	17.2(2)	14.5(1)	18.4(1)	15.1(1)
core size, D C [nm]	10 ± 3	9 ± 2	10 ± 1	10 ± 2	11 ± 3	12 ± 2
PdI of D C	0.263	0.186	0.097	0.129	0.228	0.168
flower size, D NF [nm]	23 ± 2	67 ± 6	58 ± 6	57 ± 7	49 ± 6	64 ± 6
PdI of D NF	0.197	0.093	0.083	0.125	0.131	0.093
Z-average hydrodynamic diameter, DH [nm]	43.91	88.43	71.58	88.51	99.97	73.03
PdI of D H	0.194	0.092	0.114	0.201	0.234	0.072

The nanometer and subnanometer structures of the studied particles were analyzed using the PDF approach. Raw data, collected with Q _max = 25 Å^–1, were fitted using the PDFgui structure modeling software suite. Structural models of magnetite (space group Fd $\mathrm{3̅}$ m) and maghemite (space group P4₃32) were employed in the analysis and applied across variable length scales under the assumption of a spherical nanoparticle shape. The magnetite model, which assumes full stoichiometry, provides a good fit to the data across the entire PDF field of view (1 Å < r < 60 Å), particularly in the mid-to-long range (20 Å < r < 60 Å), where local structural correlations are minimized. However, significant deviations from this model, indicative of local distortions, are clearly evident in the differential ΔG within the subnanometer range (r < 10 Å, Figure ). This discrepancy is further supported by an increased oxygen thermal parameter (U _iso) and enlarged fit residuals, R _w, both of which indicate a decline in fit quality when this data range is included in the analysis (Table ).

2.

Fits of the magnetite Fd $\mathrm{3̅}$ m model revealing local distortions for samples S1–S6, as indicated in figure panels. Blue open symbols represent PDF data, red solid lines represent the model, and green solid lines are the difference between the data and the model, which are offset for clarity.

2. Assessment from PDF Analysis Using the Fd $\mathrm{3̅}$ m Model in the 20–60 Å Range.

	sample
model parameter	S1	S2	S3	S4	S5	S6
lattice parameter, a [Å]	8.435(1)	8.421(2)	8.425(2)	8.422(2)	8.418(2)	8.431(3)
coherent nanoparticle diameter [nm]	17.3(6)	12.2(3)	12.2(3)	12.5(4)	13.4(4)	12.1(4)
oxygen thermal, U iso [Å2]	0.019(5)	0.017(9)	0.018(9)	0.019(9)	0.02(1)	0.02(1)
fit residual 20–60 Å fit, R w [%]	7.5	6.6	6.8	6.9	7.4	8.5
fit residual 1–60 Å fit, R w [%]	13.6	12.8	13.3	14.1	14.6	15.7

Focusing on the intraunit-cell length scale structure (r < 10 Å, Figure ), the misfit of the magnetite (Fd $\mathrm{3̅}$ m) model reveals distinct peak shifts in the experimental PDF data (blue open symbols) compared to the model (red solid line). Specifically, the first PDF peak in the data, located at ∼1.98 Å, appears at a distance shorter than that of the corresponding feature in the model (∼2.02 Å), while the second peak at ∼3.00 Å is shifted to a longer distance relative to that of the model (∼2.96 Å). These shifts indicate stronger bonding at the tetrahedral sites and weaker bonding at the octahedral sites. Such local structural distortions are consistent with the presence of vacancies at the Fe octahedral sites. To verify this, we employed the maghemite (P4₃32) model to fit the data within this range, which effectively resolved the discrepancies observed with the higher-symmetry magnetite model. From this refined model, we extracted the octahedral Fe site occupancies for all samples (Figure ), revealing a clear increase in vacancy concentration (i.e., a decrease in Fe occupancy) for the doped samples with a pronounced extremum observed in sample S5. These vacancies appear to be uncorrelated beyond a length scale of one-to-two-unit cells. This is evidenced by the suppression of the residual signal (ΔG) below the uncertainty threshold, marked by the red rectangle in Figure . This threshold is typically defined as the greater of two contributions: the termination error arising from the finite size of the Fourier transform window and the overall level of noise propagated to the PDF data. In this case, we adopt the maximum of the latter at long interatomic distances.

3.

Rationalization of local distortions. The multipanel block (left) illustrates failures of the magnetite model (Fd $\mathrm{3̅}$ m) to explain local bonding environments. Block arrows indicate shifts of the data signal with respect to the model. Vertical ticks in the bottom panel mark apparent PDF peak maxima in the data and the model for selected features described in text. The multipanel block (middle) shows the fit of the same data with the maghemite model (P4₃32), which accounts for the deficiencies of the magnetite model. Featured specimens are as indicated in panels. The top right panel shows estimated occupation of octahedral Fe site occupancy as obtained from the maghemite model for S1–S6. The bottom right panel shows a full data range fit of the magnetite model (Fd $\mathrm{3̅}$ m), with the differential offset below. The red shaded rectangle is an estimated uncertainty. The residual signal, shown in green at the bottom (offset for clarity), props up above the uncertainty parapet on the subunit cell length scale. See text for details.

To determine the particle size distribution of the flower-shaped nanoparticles, TEM images were obtained, as shown in Figure . From these micrographs, a comprehensive data set of 200 individual particle sizes (N = 200) was collected. A particle size histogram was then generated by using the Sturges method. The bin width (W) was calculated using the relation W = (D _max – D _min)/k, where k = 1 + 3.322 log­(N). , The constructed histograms demonstrate a good fit to a log–normal distribution. Nanoparticles form multicore flower-shaped structures (nanoflowers), with average sizes ranging from 23 to 67 nm, depending on the chemical composition (Figure and Table ). A comparison between the crystallite size obtained from XRPD analysis and the core size of the nanoflowers observed in TEM micrographs reveals that the crystallite size exceeds the core size, as shown in Table . This is in contrast to typical single-core nanoparticles, where the crystallite size is generally smaller or comparable to the core size observed via TEM. The structural characteristics of nanoflowers have been extensively studied, and our results align with literature data. , In multicore iron oxide nanoflowers, individual cores are connected through oriented aggregation and epitaxial growth, causing XRPD to detect the entire structure as a single crystalline domain. High-resolution transmission electron microscopy (HR-TEM) images further confirm that the crystal lattice extends across multiple cores, demonstrating that these structures are not merely physical clusters but rather coherently interconnected crystalline domains. , A noticeable deviation was observed in the morphology of sample S1, where the flower-like structure was less pronounced and the nanoflowers were significantly smaller compared to those in samples with different compositions (Table ). However, it can be stated that the change in chemical composition does not affect the morphology induced by the synthesis method.

4.

TEM micrographs of (Zn _x Mn _y Fe _z V _q )₂O₃ samples.

The representative S5 nanoflowers and size distributions are presented in Figure b,c. The samples remained relatively stable in a colloidal form, likely due to molecules from the synthesis, such as DEG, adhering to the nanoparticle surfaces. The measured hydrodynamic diameters, presented in Table as Z-average values with the corresponding PdIs, show significant variation, particularly between sample S1 and the other samples. Z-average values range from about 44 nm (S1) to almost 100 nm for sample S5.

5.

(a) HR-TEM micrograph of the selected S5 nanoflower sample (D _NFnanoflower diameter, D _Ccore diameter), (b) nanoflower size distribution, and (c) core size distribution.

3.3. Magnetic Properties and Heating Efficiency of (Zn _x Mn _y Fe _z V_q)₂O₃

ZFC/FC magnetization measurements were conducted in a 50 Oe field in the temperature range from 5 to 300 K (Figure a). Flat FC curves show the presence of significant interparticle interactions in all six samples. This is most likely due to the specific growth pattern of nanoflowers, where magnetic cores inside the particles are in close contact. Additionally, close packing of particles in measured powders promotes the presence of interparticle interactions.

6.

Magnetic measurements of (Zn _x Mn _y Fe _z V _q )₂O₃: (a) ZFC/FC curves under a DC field of 50 Oe, (b) hysteresis loops at 5 K, and (c) hysteresis loops at 300 K.

Isothermal field dependence of magnetization at 5 and 300 K are presented in Figure b,c. At 5 K, samples behave similarly, with narrow hysteresis, a saturation field of around 10 kOe, and an M _S value around 100 emu g^–1. At 300 K, the coercive field of all samples drops to zero, and their M _S values are around 70 emu g^–1 (Table ). Similar values have been reported in work by Lucaciu et al. The exception is sample S1, which has lower magnetization values and does not saturate up to 50 kOe.

3. Average Particle Diameter, Standard Deviation, Polydispersity Index, and Saturation Magnetization Values Acquired from Magnetization Measurements of (Zn _x Mn _y Fe _z V _q )₂O₃ .

sample	Davg [nm]	SD [nm]	PdI	Ms [emu g–1]
S1	13.7	4.5	0.11	50
S2	10.6	2.0	0.04	71
S3	11.2	2.7	0.06	69
S4	10.9	2.3	0.04	71
S5	10.7	0.5	0.002	69
S6	11.2	2.7	0.06	66

^aThe standard errors of the fits are of the order of 10^–2 for D _avg and SD, 10^–4 to 10^–3 for PdI, and 10^–1 for M _s and are omitted for the sake of clarity.

The results of fitting the weighted Langevin equation (eq ) to hysteresis data acquired at 300 K are presented in Table . With the exception of S1 and S5, all samples show similar behavior. Sample S1 could not be fitted by the Langevin equation alone, but a linear component proportional to the applied field had to be added to the model. This paramagnetic contribution could be explained by incomplete crystallization and the presence of a disordered layer at the surface of the particles. This is in agreement with the TEM analysis (Figure ). The sample shows a wide size distribution and features diminished saturation magnetization. In contrast, sample S5 shows a narrow size distribution of magnetic cores inside nanoflowers.

From the results presented in Tables and , one can infer that an increase in the Zn fraction leads to a smaller “flower” size, although the crystallite and magnetic core sizes remain similar. The reduction of M _S and nanoparticle size with Zn doping was previously reported by several groups. ,

The temperature change during hyperthermia measurements for all samples is shown in Figure a. The samples exhibit significantly different performances, with ΔT after 2 min ranging from 3 °C for sample S1 to over 30 °C for sample S5. Beyond the overall temperature increase, optimizing the heating efficiency in magnetic hyperthermia also inherently benefits from achieving the shortest possible heat transfer time (τ*). This parameter, crucial for efficient therapeutic application, can be quantitatively extracted by fitting the temperature–time data (as presented in Figure a,c) using the Box–Lucas model, a method well-established in the literature for analyzing thermal dynamics in magnetic nanoparticle suspensions.

7.

Magnetic hyperthermia measurements of (Zn _x Mn _y Fe _z V _q )₂O₃. (a) Heating curves of the samples at a frequency of 252 kHz and a field of 15.9 kA m^–1; (b) SAR values of (Zn _x Mn _y Fe _z V _q )₂O₃ and the ILP parameter of (Zn _x Mn _y Fe _z V _q )₂O₃. (c,d) Heating abilities of the S5@CA sample.

The graph of SAR values (Figure b) further highlights the significant differences in the heating performance of the samples. While S1 barely manages 22 W g^–1, S5 achieves 369 W g^–1. This performance is highly competitive and comparable to those of some of the most efficient magnetic nanoparticles reported in the literature. For example, Jang et al. demonstrated high SAR values of 432 W g^–1 for (Zn_0.4Mn_0.6)­Fe₂O₄ nanoparticles, underlining the critical role of optimized magnetic properties for effective hyperthermic treatment. The excellent SAR of the multicore nanoflower S5 sample (369 W g^–1) further supports the notion that both precise compositional tuning and an advantageous morphology contribute to superior heating capabilities. According to Liu et al., Zn content determines particle size and, consequently, the magnetization relaxation time (τ). τ depends strongly on particle size, and the optimal applied field frequency (f) should satisfy the condition 1 = 2πfτ. In our case, f = 252 kHz was appropriate for the S5 sample. The ILP values further highlight the superior performance of samples with an optimized Mn/Zn/Fe ratio (Figure b). These results suggest that careful balancing of the chemical composition can optimize the heating efficiency of the samples for magnetic hyperthermia applications.

Sample S5 was coated with citrate for use as an agent in in vitro magnetic hyperthermia. Figure c,d displays heating curves and SAR and ILP values of S5@CA. The high hyperthermic efficiency of S5@CA suggests that it is well-suited for use as a heat generator in studies investigating the heating effect on selected cell lines.

To summarize the key findings from Sections , we identified the optimal composition for biomedical application based on comprehensive structural, microstructural, and compositional analyses. All synthesized (Zn _x Mn _y Fe _z V _q )₂O₃ nanoparticles crystallized in a defective maghemite phase, with Fe-site vacancy, primarily in the octahedral sublattice, confirmed by PDF analysis. These vacancies were associated with increased local structural disorder and an enhanced magnetic performance. Among the investigated compositions, sample S5 (Zn/Mn/Fe = 0.040:0.182:1) exhibited the highest vacancy concentration and most significant local distortions, resulting in outstanding heating efficiency, with a SAR of 369 W g^–1 and an ILP of 5.77 nHm² kg^–1. Its structural integrity, multicore nanoflower morphology, and exceptional magnetic hyperthermia performance make S5 the most promising candidate for further development. Consequently, this composition was selected for surface functionalization, radiolabeling, and in vitro biological evaluation.

3.4. Cytotoxicity of Nanoparticles and In Vitro Magnetic Hyperthermia

Given the excellent hyperthermic performance of sample S5, subsequent biological experiments were focused exclusively on this composition. To preliminarily evaluate its cytotoxicity, the synthesized S5 nanoparticles were tested on multiple malignant cell lines (HeLa, A549, LS174, and A375) as well as a normal cell line (MRC-5).

The IC₅₀ analysis revealed that S5MNPs exhibited significant cytotoxicity toward HeLa (IC₅₀ = 20.41 μg mL^–1) and LS174 (IC₅₀ = 15.86 μg mL^–1) cells, whereas they were considerably less toxic to A549 (lung carcinoma) and normal fibroblasts (MRC-5), with IC₅₀ values exceeding 200 μg mL^–1 (Table ). These differences in cytotoxicity could be attributed to variations in the membrane structure and cell metabolism. Specifically, it is plausible that colorectal carcinoma (LS174) and cervical carcinoma (HeLa) cells may possess an enhanced capacity for nanoparticle uptake via endocytosis. This could potentially lead to nanoparticle accumulation within the cells and an increase of reactive oxygen species (ROS) production, possibly contributing to cytotoxic effects observed in this study. In A375 melanoma cells, the cytotoxicity of S5 nanoparticles was noticeable but lower compared to HeLa and LS174 cells (IC₅₀ = 147.38 μg mL^–1). This might be attributed to distinct nanoparticle uptake pathways specific to melanoma cells and their ability to efficiently detoxify oxidative stress. On the other hand, the minimal cytotoxicity observed in A549 cells suggests a relative resistance to nanoparticles, which may result from reduced nanoparticle internalization efficiency and stronger detoxification mechanisms. Further dedicated studies, including quantitative analysis of nanoparticle internalization and cellular ROS levels, would be necessary to experimentally validate these proposed mechanisms. It is also worth noting that mathematical modeling, such as the application of the Hill model, offers a powerful quantitative framework for understanding complex dose–response relationships and elucidating specific parameters like benchmark dose (BMD), lethal dose (LD₅₀), and cooperativity in cell viability studies. This approach can provide deeper insights into the intricate interactions between nanoparticles and biological systems, as effectively demonstrated in studies investigating various nanoplatforms for cancer therapy.

4. Cytotoxic Activity (IC₅₀ Values) of Samples against Human Cancer Cell Lines .

sample	HeLa IC50 [μg mL–1]	A549 IC50 [μg mL–1]	LS174 IC50 [μg mL–1]	A375 IC50 [μg mL–1]	MRC-5 IC50 [μg mL–1]
S5	20 ± 1	>200	16 ± 1	147 ± 2	>200
S5@CA	>200	>200	>200	>200	>200

^aIC₅₀ values as well as cell survival were expressed as the mean ± SD determined from the results of the MTT assay.

The specific stabilization mechanisms depend on the type of coating material. It has been shown that surface modification with bovine serum albumin and polyethylene glycol can significantly reduce the cytotoxicity of magnetic nanoparticles. By coating S5 nanoparticles with citrates (S5@CA), their cytotoxicity was reduced in all tested cell lines, with IC₅₀ values exceeding 200 μg mL^–1 for all tested samples. Citrate may act primarily as an electrostatic stabilizer, providing a surface charge that prevents nanoparticle aggregation through Coulombic repulsion. This stabilization could, in turn, reduce their interaction with the cell membrane and potentially decrease the nanoparticle uptake via endocytosis.

The cell survival results on HeLa cells, 24 h after exposure to an AC magnetic field in the presence of S5@CA nanoparticles as heat generators, are presented in Table . The nanoparticle concentration of 1.1 mg mL^–1 used in the experiment indicates their nontoxicity on HeLa cells even at such high concentrations. The group, which included cells treated with MNPs without the application of a magnetic field, was used as a control for the assessment of the effect of in vitro MH. The control showed 100% survival, serving as a reference value. The results of in vitro MH demonstrated a time-dependent reduction in cell survival, with viability percentages of (84 ± 2)% after 30 min and (61 ± 1)% after 60 min of treatment. The treatment duration was measured from the moment the temperature reached 48 °C (after 4–5 min), which was then maintained as the maximum operating temperature (working temperature). Importantly, the heating profile remained stable throughout both 30 and 60 min treatments, suggesting that the structural and functional properties of the S5@CA nanoparticles were preserved under repeated magnetic field exposure. This observation is in line with our earlier in vivo studies using similar γ-Fe₂O₃ nanoflowers, where multiple MH treatments over a two-week period yielded reproducible therapeutic effects, indirectly confirming the stability of their (micro)­structure properties. These data are also consistent with literature indicating that magnetic hyperthermia can reduce cell survival, but its efficiency depends on the type of nanoparticles and their hyperthermic efficiency, the coating type, which can influence stability and interactions with cells, and the cell line type, where certain lines, such as Saos-2 osteosarcoma cells, demonstrated higher sensitivity to hyperthermia compared to healthy cells, as well as the experimental protocol itself, including magnetic field strength, treatment duration, and temperature. An increase in field strength, frequency, and treatment duration enhances the therapeutic effect but also carries the potential risk of nanoparticle aggregation and damage to surrounding tissue.

5. In Vitro Magnetic Hyperthermia on HeLa Cells.

treatment	[%]
HeLa	100
HeLa + S5@CA	100 ± 2
HeLa + AC60 min	108 ± 2
HeLa + S5@CA + AC30 min	84 ± 2
HeLa + S5@CA + AC60 min	61 ± 1

Comparing these results with other studies on multicore iron oxide nanoparticles in cell cultures provides valuable context, although direct quantitative comparisons are challenging due to significant variations in experimental parameters across different reports. Key factors such as the applied magnetic field (amplitude and frequency), exposure time, nanoparticle concentration, and inherent properties of the cell lines (e.g., uptake efficiency, thermal sensitivity) vary considerably. For instance, Lartigue et al. investigated in vitro MH of multicore flower-like iron oxide structures (D _NF = 24.9 nm) on MCF-7 cells. Their application of a more intense AC magnetic field (29 kA·m^–1, 520 kHz for 1 h) and higher iron concentrations (0.2 and 2 mM) resulted in 20% to 60% cell necrosis. While their field conditions were more aggressive, our study shows a comparable reduction in cell viability under our specific, controlled hyperthermic conditions (maintaining temperature at 48 °C). More recently, García-Soriano et al. conducted a comprehensive study using multicore iron oxide nanoflowers across six different cancer cell lines. Their findings included a 12–14% reduction in MCF-7 cell viability (at 0.05 mg_Fe·mL^–1, 23.8 kA·m^–1, 202 kHz, 20 min), a significant 34% reduction in MB231 cells, a moderate effect on A549 cells, and no effect on HCT116 cells. Other examples include a 15% viability reduction in PC3 cells using CuS-coated nanoflowers and 80–90% cell death in U87 cells with 29.1 nm nanoflowers.

Despite the inherent variability across studies, our results demonstrate that S5@CA nanoparticles exhibit significant potential as effective agents for MH in cellular contexts, positioning them as promising candidates for further therapeutic development. It is worth noting that surface modification can sometimes alter the chemical composition of multicationic nanoparticles due to partial ion leaching, as previously observed for Mn- and Fe-based nanostructures during ligand exchange steps. Although the Zn/Mn/Fe ratio was not re-evaluated after citrate coating in our study, the excellent heating performance and retained biocompatibility of S5@CA nanoparticles (1.1 mg mL^–1) observed in vitro strongly suggest that any potential compositional changes did not compromise their functional properties relevant to magnetic hyperthermia.

3.5. Radiolabeling of S5@CA

Considering that the S5@CA sample demonstrated excellent performance as an agent in in vitro hyperthermia, it was further tested as a potential radionuclide carrier for radionuclide therapy. In these studies, we used the β-emitting radionuclide ¹⁷⁷Lu, which stands out as a particularly promising candidate due to its suitability for theranostic applications, especially in the treatment of neuroendocrine tumors and prostate cancer. The radiolabeling yield of ¹⁷⁷Lu-S5@CA at RT for 60 min was 92.67%, indicating efficient labeling under established radiolabeling conditions. The in vitro stability results, summarized in Table , demonstrate that ¹⁷⁷Lu-S5@CA retained high radiochemical purity over 72 h in both physiological saline and HSA, with a minimal release of ¹⁷⁷Lu³⁺.

6. In Vitro Stability of ¹⁷⁷Lu-S5@CA under Different Conditions.

	radiochemical purity
sample 177Lu-S5@CA	1 h	24 h	48 h	72 h
saline	92.67	89.95	86.49	85.72
HSA	90.56	88.16	86.11	83.39

The carboxyl groups coordinate with Lu³⁺, making citrate coatings useful molecules in chelation. The slight decrease in radiochemical purity over time suggests minimal dissociation of ¹⁷⁷Lu³⁺ from nanoparticles, which is within acceptable limits for potential biological applications. Stability in HSA is slightly lower compared to that in saline, likely due to the presence of different competing biomolecules. Nonetheless, the observed radiolabeling yield and stability of ¹⁷⁷Lu-S5@CA are sufficient to support further in vivo studies.

4. Conclusion

This study presents the successful design and comprehensive evaluation of novel (Zn _x Mn _y Fe _z V _q )₂O₃ nanoflowers as multifunctional agents for bimodal cancer therapy, combining magnetic hyperthermia and radionuclide delivery. Through controlled cation substitution and defect engineering, we established a direct correlation between nanoparticle composition, vacancy distribution, and functional properties relevant to biomedical applications. Structural analysis confirmed that all (Zn _x Mn _y Fe _z V _q )₂O₃ samples crystallize in the maghemite phase, exhibiting significant local structural distortions and a high concentration of vacancies at Fe sites. PDF analysis revealed systematic changes in interatomic distances, validating the structural modifications induced by cation substitution. These findings highlight the remarkable adaptability of the maghemite structure in accommodating mixed cation compositions while maintaining its defective spinel character. All synthesized nanoflowers exhibited superparamagnetic-like behavior at physiological temperatures, with M _S values around 70 emu·g^–1 at 300 K. Notably, the optimized sample S5 (Zn/Mn/Fe = 0.040:0.182:1) demonstrated exceptional heating capability, achieving a temperature rise of over 30 °C within 2 min and a SAR of 369 W·g^–1. The ILP further revealed an enhanced hyperthermic performance with increased Mn content. All nanoflower samples exhibited superparamagnetic-like behavior at physiological temperatures (M _s ≈ 70 emu·g^–1 at 300 K), with the optimized composition (S5: Zn/Mn/Fe = 0.040:0.182:1) demonstrating outstanding heating efficiency (ΔT > 30 °C in 2 min, SAR = 369 W·g^–1). Zn/Mn/Fe content correlated with SAR/ILP, underscoring the role of compositional tuning and controlled nanoflower morphology in achieving high therapeutic performance. Cytotoxicity assays revealed selective toxicity of the uncoated S5 nanoparticles toward cancer cell lines (HeLa and LS174), with significantly reduced toxicity toward normal fibroblasts and A549 cells. Surface modification with citrate (S5@CA) drastically improved the biocompatibility across all tested cell lines, eliminating cytotoxicity at concentrations up to 200 μg mL^–1. Under magnetic hyperthermia conditions, S5@CA nanoparticles triggered a time-dependent reduction in HeLa cell viability, with survival decreasing to 84% after 30 min and 61% after 60 min at a constant working temperature of 48 °C. These findings validate the therapeutic efficiency of S5@CA as a heat-generating agent for controlled hyperthermia. The S5@CA nanoparticles were efficiently radiolabeled with ¹⁷⁷Lu (with 92.67% yield) and retained high radiochemical purity (>83%) over 72 h in both saline and serum-like conditions. Citrate provided surface carboxylate groups that enabled stable Lu³⁺ coordination, supporting the suitability of S5@CA for theranostic applications. In conclusion, this study underscores the importance of precise chemical composition control, defect chemistry, and microstructure in optimizing nanomaterials for biomedical applications. Our findings provide a foundation for the development of next-generation therapeutic strategies for cancer treatment.

M.O.: Conceptualization, Formal analysis, Investigation, Methodology, Writingoriginal draft. M.B.: Investigation, Software, Validation, Writingreview and editing. T.S.: Formal analysis, Investigation, Software, Validation. B.D.: Methodology, Project administration, Supervision, Writingreview and editing. M.A.: Data curation, Formal analysis, Funding acquisition, Investigation, Writingreview and editing. A.T.: Investigation, Methodology, Writingreview and editing. D.H.: Investigation, Methodology, Validation, Writingreview and editing. S.VD.: Data curation, Formal analysis, Funding acquisition, Investigation, Project administration, Writingreview and editing. E.B.: Data curation, Formal analysis, Software, Investigation, Writingreview and editing. B.A.: Conceptualization, Supervision, Investigation, Methodology, Resources, Validation, Writingoriginal draft, Writingreview and editing. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.