Comprehensive magnetic resonance characteristics of carbon-encapsulated iron nanoparticles: a new frontier for the core-shell–type contrast agents

The development of carbon-encapsulated iron nanoparticles (CEINS) is of considerable interest in many areas of cancer nanotechnology, ranging from basic tumor biology to early detection and treatment of cancer. To meet these challenges, the present study was undertaken to determine the magnetic and relaxometric performance of CEINS used as a new contrast agent for magnetic resonance imaging (MRI) in preclinical phantom models. CEIN samples were synthesized using a carbon arc discharge route, and the as-synthesized nanoparticles were purified and functionalized with surface acidic groups. The presence of various Fe-bearing metallic phases reflecting the net magnetic properties of CEINS was characterized by powder X-ray diffraction (XRD), thermogravimetry (TGA), and vibrating sample magnetometry. The morphological and surface chemistry features were characterized by electron microscopy (transmission electron microscopy [TEM] and scanning electron microscopy [SEM]), Raman spectroscopy, and Fourier transform-infrared (FT-IR) spectroscopy. The textural properties of CEINS, including porosity, surface total charge density, and zeta potentials, were also measured. The as-synthesized different CEIN samples were finally examined as a potent MRI contrast drug candidate. Magnetic resonance relaxation measurements were performed in bovine gelatin-based phantom models by using a 1.5-T MRI scanner equipped with a standard radiofrequency “birdcage” type head coil. To obtain data, T1- and T2-weighted MR images were acquired using the inversion recovery spin echo (SE) and the SE protocol with multiple time of echo (TE), respectively. Chemical characterization showed similarity in morphology and textural properties between as-synthesized CEINS, purified CEINS, and CEINS functionalized with acidic groups. The as-synthesized CEINS had significantly higher Fe content and higher saturation magnetization. The analysis of the relaxometric properties of CEINS revealed that all the CEIN samples decreased T2 relaxation times on the T2-weighted images. The relaxation rate (1/T2) showed some differences between the as-synthesized, purified, and surface-functionalized CEINS containing surface carboxylic groups. Both the as-synthesized and purified CEINS slightly decreased the T1 relaxation times, which was evident through increase in the relaxation rates (1/T1). This study concludes that CEINS may represent a novel “core-shell”–type negative contrast drug candidate for MRI. It should be emphasized that all the studied CEIN samples have acceptable r2 relaxivities at the field strength of 1.5 T where most of the MRI systems operate in clinical radiology.


Introduction
There have been an increasing number of investigations on various aspects of magnetic nanomaterials in the past two decades (Jasmin de Souza et al. 2017;Lee et al. 2013;Pankhurst et al. 2009;Shubayev et al. 2009;Sun et al. 2008;Vallabani et al. 2019;Wang et al. 2001). Among the different magnetic nanoparticles studied to date, iron oxide nanoparticles have received much attention as contrast agents for use in magnetic resonance imaging (MRI) (Chung et al. 2011;Maurea et al. 2014;Vasanawala et al. 2016;Wang 2011;Wang and Idée 2017). According to the hydrodynamic size of iron oxide nanoparticles, these iron species are classified into three categories of particles: micro-size iron oxide (MPIO; > 1000 nm), superparamagnetic iron oxide (SPIO; 60-180 nm), and ultra-small superparamagnetic iron oxide (USPIO; < 50 nm). Subsets of USPIO are also identified as very small nanoparticles of iron oxide (VSOP; 7-9 nm) and monocrystalline iron oxide nanoparticles (MION; 10-30 nm). Most of the iron oxide nanoparticles used as contrast agents in MRI are composed of both Fe(II) and/or Fe(III) cations and O 2 − and/ or OH − anions forming a group of oxides, hydroxides, and oxy-hydroxides (Laurent et al. 2008). In the presence of an external magnetic field, these magnetic particles tend to align with the field and exhibit net magnetization in the direction of the magnetic field (Di Marco et al. 2007). It should be emphasized that a major disadvantage of iron oxide nanoparticles is related to their relatively moderate magnetic performance (e.g., in comparison to bare Fe nanoparticles), which is considered to be an important parameter in molecular MRI. To elucidate this in a better way, one may compare the saturation magnetization of the magnetite (Fe 3 O 4 ) and alpha iron (Fe) nanoparticles. The maximum achievable magnetization for the first and second phases is estimated to be ca. 80 and 224 emu/g (both values refer to room temperature), respectively (Pankhurst et al. 2009;Tartj 2006). In other words, one can easily predict that the mitigation of signal intensity observed on postcontrast T2-and T2*-weighted MR images can be achieved at the lower concentration, when pure Fe nanoparticles are used to obtain data. Hence, the nanoparticles comprising pure iron (Fe) phases have substantially better magnetic characteristics for application in cancer diagnostics based on molecular MRI. It should be noted that magnetic nanoparticles composed of different bare metals (i.e., Fe, Co, Ni, or their alloys) have been found to have some limited resistance properties to oxidation and agglomeration/aggregation processes (El-Gendy et al. 2009). These drawbacks can be substantially overcome by encapsulation of pure metallic nanoparticles in carbon layers, leading to production of the so-called carbon-encapsulated iron nanoparticles (CEINS) (Bystrzejewski et al. 2007). Such heterogeneous nanoparticles are composed of the metallic core (generally of the spherical shape), which is covered by few layers of graphene coating, leading to the "core-shell"-type magnetic nanostructures (Bystrzejewski et al. 2007). The carbon coating is mainly formulated of highly crystalline and curved graphene monolayers, which protect the core against any oxidization processes, preserve its specific magnetic properties, and further endow the encapsulated metal nanoparticles with the features of biocompatibility and stability in many organic and inorganic media, including animal and human fluids (Bystrzejewski et al. 2011). The encapsulation of iron nanoparticles in graphene-like shells was found to be a well-accepted method to increase their stability and preserve specific magnetic properties. The carbon coating is also considered as the optimal encapsulating material because it is light and tightly covers the magnetic core in nanoparticles (Bystrzejewski et al. 2011).
Our previous studies showed that CEINS have a high corrosion resistance (Bystrzejewski et al. 2006) and are readily susceptible to various functionalization processes (Kasprzak et al. 2015). It should also be noted that the mean diameter distribution of CEINS can be controlled by changing a very simple macroscopic parameter, namely grain size of the starting Fe powder used in synthesis processes or the chemical composition of the anode ). This phenomenon opens up new possibilities to control the magnetic characteristic of CEINS, leading to increase in the magnetic performance. More recently, studies have shown that CEINS can be precisely functionalized with a large number of functional groups that can be successfully linked with some biological molecules (Kasprzak et al. 2016;Poplawska et al. 2014). Interestingly, carbonencapsulated iron oxide nanoparticles (Fe 3 O 4 ) and carbon-coated gadolinium particles have been also tested as novel carbon-shell-type contrast drug candidates in MRI (Atabaev et al. 2015;Bae et al. 2012 (Bae et al. 2012;Lee et al. 2014;Khramtsov et al. 2019;Park et al. 2011).
In the present study, we report the comprehensive and systematic characteristics of carbon-encapsulated iron nanoparticles examined as a potent contrast drug candidate for MRI by testing in the phantom gelatin models (in vitro). The CEINS were synthesized using a carbon arc discharge route, and the as-synthesized CEINS were purified and functionalized with acidic groups, including carboxylic moieties. The size and hydrodynamic properties of the CEIN samples were characterized by transmission electron microscopy (TEM), scanning electron microscopy (SEM), and dynamic light scattering (DLS). Further structural details were evaluated from powder X-ray diffraction (XRD), thermogravimetry (TGA), Raman spectroscopy, and Fourier transform-infrared (FT-IR) spectroscopy. The global surface charge density and zeta potential measurements were also performed to obtain data. The magnetic properties of CEINS were measured using a vibrating sample magnetometer. Finally, the spin-lattice relaxation (T1) and spin-spin (T2) processes were examined in the aqueous gelatin media of CEINS by using a magnetic resonance scanner operating at 1.5 T. Our results revealed that CEINS can function as a new T2 (negative) contrast agent in MRI.

Materials and methods
Synthesis and surface modifications CEINS were synthesized by a carbon arc discharge route. The detailed procedure is described elsewhere (Borysiuk et al. 2008;Bystrzejewski et al. 2005). The synthesis method is based on the simultaneous sublimation of a heterogeneous anode containing iron (Fe) and graphite (C) powders. The anode material is transformed to the vapor phase congruently due to the very high temperature of the carbon arc (5000-6000 K). Next, the as-formed metal-carbon gas undergoes rapid cooling, which leads to the nucleation of Fe-C vapor and subsequent solidification to form iron nanoparticles encapsulated in few graphene monolayers. Briefly, the carbon arc discharge was ignited between the graphite cathode and the graphite anode, which was doped with Fe (45% wt.). The discharge was maintained at the current of 70 A under Ar-H 2 atmosphere (50:50 vol.%) at the total pressure of 60 kPa. The as-obtained raw product contained CEINS, and nonencapsulated iron particles were designated as Fe@C/Fe. A fraction of the as-synthesized Fe@C/Fe unpurified sample was further subjected to a purification process to remove the nonencapsulated Fe particles (Fe). The purification procedure included 24 h refluxing in 3 M HCl, washing in excess of water and ethanol, and drying in air at 70°C. The as-purified product was designated as Fe@C. The raw CEINS were also chemically oxidized to introduce surface acidic groups, mainly carboxylic, which are covalently bound to carbon coatings. The corresponding product was designated as Fe@C-COOH. In a typical run, 2 g of the raw Fe@C/Fe product was suspended in a mixture containing 80 ml of concentrated H 2 SO 4 and 26 ml of concentrated HNO 3 . The sample was sonicated in a water bath at 25-30°C for 3 h (80 W). Then, the suspension was diluted by 1 l of distilled water and allowed to cool down for 2 h. The suspended particles were recovered on a membrane filter under reduced pressure and washed with excess of water and ethanol. All reagents used in the CEINS synthesis were obtained from Sigma-Aldrich and used without further purification.

Physicochemical characterizations
Morphological details of CEINS were studied by TEM (Zeiss Libra 120 operated at 120 kV). The diameter distribution was obtained by analyzing at least 300 objects on TEM images. The qualitative phase composition was evaluated by powder XRD (Bruker D8) using Cu K-α radiation in a 2θ range between 10 and 70 with a step of 0.02. Raman spectra were acquired using a dispersive spectrometer (Jobin Ivon T-64000) equipped with a 515-nm excitation laser with the spectral resolution of 4 s cm −1 . FT-IR spectra were recorded in a transmission mode with a Perkin Elmer System 2000 spectrophotometer. The amount of the surface acidic groups in CEINS was also determined by Boehm titration (Boehm 1994). Magnetic measurements were carried out at 25°C using a vibrating sample magnetometer (Lake Shore 668). The measured magnetization referred to the total mass of the investigated sample. The Fe content was evaluated by TGA using a TA Q-50 instrument. The TGA measurements were carried out under N 2 -O 2 (60/40) atmosphere at the constant heating rate of 10°C · min −1 . The hydrodynamic diameter and zeta potential were measured using the Malvern Zetasizer instrument.

Phantom relaxometry measurements
In the present study, a purified (Millipore) water solution of carboxymethyl cellulose (CMC; Sigma-Aldrich) at the concentration of 0.1 mg/ml (viscosity 0.957 mPa s) was used as a stabilizing agent to increase the sedimentation stability of water suspensions of CEINS. To obtain data, the sedimentation studies of CEINS in different CMC-water media were described elsewhere in details (Grudzinski et al. 2013). In the phantom relaxometric studies, the CMC-water media of CEINS were embedded in the bovine gelatin type B (Sigma-Aldrich) and placed into a plastic centrifugation tube (16 mm in diameter, 100 mm long) to perform MRI measurements. Each phantom was filled with 5.0% gelatin type B containing respective CEINS at the final concentration ranges of 0.0001, 0.001, 0.01, 0.1, 0.5, 1, 5, 10, 25, and 50 μg/mL. Water and gelatin B media without CEINS served as controls. For preparing phantom gels, the gel suspension of 5.0% (w/v) low-meltingpoint gelatin gel type B was preliminary blended in pure water and heated first in a thermostat bath up to 65°C. The contents were blended in a Vortex mixer and then cooled down to a bath temperature of between 43 and 40°C just before the addition of CEINS. The suspensions of CEINS were prepared in the aqueous solution of CMC (0.1 mg/ml). The prepared gel solution was mixed thoroughly with vigorous shaking with different CEINS suspensions at 40°C to prevent gelation while mixing and poured in plastic centrifuge tubes. The resultant mixture was quickly cooled down to prevent CEINS sedimentation. The plastic tubes containing CEINS were finally placed in a plastic rack and used for MRI measurements. Additionally, the gelatin phantoms containing CEINS were also studied by DLS (Malvern Zetasizer, 173 deg scattering angle) to evaluate the mean hydrodynamic diameter. The as-prepared phantom samples were melted and immediately transferred to polystyrene cuvettes for MRI studies.
All the MRI experiments were performed using a 1.5-T system (Magnetom Avanto, Siemens Medical Solutions, Erlangen, Germany), with explorer gradients (maximum gradient of 40, 40, and 45 mT/m along the x, y, and z axes, respectively, and slew rate of 200 mT m −1 ms −1 ) and a 32-multichannel system. The standard "birdcage" type radiofrequency (RF) coil (Siemens) was used throughout the experiments. The phantom gel aliquots were kept inside the coil and placed within the MRI (1.5 T) scanner for imaging. The temperature was controlled at 20°C in the magnet room of the scanner. The experimental setup is shown in Suppl. Fig. S1.
After acquiring the MR images, the mean signal intensity (INS) was measured within the manually drawn equal-sized regions of interest (ROI) for each tested sample. The relaxation times were calculated by mono-exponential curve fitting of the signal intensity vs. time (TI or TE) using the MATLAB software. The following equation was used to fit the curves: where S(T I ) is the signal intensity with inversion pulse of duration T I applied, T R is the repetition time, S 0 is the proton density, and T 1 is the longitudinal relaxation time. The absolute values account for the fact that the signal intensity in data is represented by the unsigned integer. T 1 was calculated numerically with the nonlinear regression model using the Levenberg-Marquardt algorithm (LMA) to fit the model curve to the experimental data. For the signal intensity function of T2 relaxation, the following equation was used to determine T2 relaxation times: where T E is the time of echo, T 2 is the transversal relaxation time, and M 0 is the proton density. T1 and T2 relaxation times were calculated for all the studied CEIN samples with different iron concentrations in the gelatin phantom. According to the relaxation theory (Dayie et al. 1996;Roch et al. 1999), the relaxation rates (1/T1, 1/T2) are linearly dependent on Fe concentration. Therefore, the relaxivities (r1 and r2) were calculated as the slope of the respective plot of the 1/T1 and 1/T2 rates vs. the Fe content in different gelatin phantoms of CEINS to compare the contrastenhancing effect of the raw (Fe@C/Fe), purified (Fe@C), and surface-functionalized (Fe@C-COOH) CEINS. A coefficient of determination (Rsquared) was calculated to assess the goodness of each curve fitted. Figure 1 shows the representative SEM images of the studied CEINSs. The as-obtained raw (Fe@C/Fe), purified (Fe@C), and surface-functionalized (Fe@C-COOH) samples nearly have the same morphology. They are composed of spheroidal nanoparticles with diameters below 100 nm. The nanoparticles in the solid state form fractal agglomerates. Further details can be obtained from TEM images (Fig. 2). The TEM inspection reveals the core-shell structure of the studied CEIN samples. The as-synthesized sample (Fe@C/Fe) contains two types of nanoparticles: CEINS with welldeveloped carbon coatings and some iron particles that are not fully encapsulated or encapsulated in noncontinuous carbon shells. The purified (Fe@C) and surfacefunctionalized (Fe@C-COOH) material also contain two types of nanostructures. The first type, similar to Fe@C/Fe, includes CEINS. The nanoparticles have a typical core-shell structure, i.e., the metallic (iron) cores are covered by carbon coatings. The diameter of individual pristine CEINS varies between 5 and 100 nm. Importantly, the greatest fraction of the sample has the size ranging between 10 and 50 nm (Fig. 2).

Physicochemical characterizations
The water-CMC suspensions of Fe@C/Fe, Fe@C, and Fe@C-COOH nanoparticles were subjected to DLS and zeta potential measurements. The hydrodynamic diameter range and mean hydrodynamic diameter are listed in Table 1. The diameter range covers relatively broad limits and exceeds the size of pristine nanoparticles evaluated from TEM images (Fig. 2). The same conclusion applies to the mean hydrodynamic diameter, which is ca. 10 times larger than that of the pristine CEINS. This finding clearly demonstrates that the individual nanoparticles exist in an agglomerated form in the water-CMC suspension. Interestingly, the largest mean hydrodynamic diameter is found for the raw Fe@C/Fe sample. This is an unexpected result, because this sample is highly hydrophobic as it contains both uncoated Fe nanoparticles and nonfunctionalized carbon encapsulates. Surface-functionalized Fe@C-COOH have more hydrophilic characteristic and form (with the aid of CMC surfactant) smaller agglomerates. The zeta potential values of CEINS suspensions are between − 24.7 and − 25.4 mV. This finding shows that the obtained suspensions have a fair stability in the experimental media. Therefore, the bovine gelatin type B phantoms of CEIN samples were prepared for MRI measurements.
The qualitative analysis of the magnetic core using an X-ray microprobe (EDX) indicates the presence of iron and carbon. Some Cu signals are derived from the microscopic mesh, and traces of oxygen are also evidenced, probably as an air pollutant (Fig. 3). It should be noted that the carbon shell of CEINS nanoparticles is composed of well-crystallized graphene layers, ensuring the tightness of the encapsulated iron nanoparticles (please see the insert in Fig. 3).
The phase composition of the metallic core was studied by powder XRD. The diffractograms are shown in Fig. 4. The samples have nearly the same qualitative   (110), bcc Fe (200), and C (002) crystal planes in the XRD patterns were matched with those of the ferrite and graphite crystalline structures (fcc Fe-C 41506-ICSD, bcc Fe 52258-ICSD, and C (graphite) 76767-ICSD). Figure 5 shows the Raman spectra of raw (Fe@C/ Fe), purified (Fe@C), and functionalized with acidic groups (Fe@C-COOH) nanoparticles. The bands G and D are clearly evidenced. The ratio of G/D in the raw and purified samples is 1.89 ± 0.10. The oxidized nanoparticles have a lower value for this parameter (1.56 ± 0.07). The decrease in the H/D ratio is due to the introduction of carboxyl groups and the amorphic graphene planes in the nanoparticles.
The surface functionalization of CEINS was confirmed by FT-IR spectra and quantitative evaluation of surface functional acidic groups by Boehm titration (Fig. 6). The FT-IR spectra of Fe@C-COOH have several bands, which are absent in the raw Fe@C/Fe and purified Fe@C samples. The band located at 1710 cm −1 is due to the C=O vibration in carboxylic groups. The strong feature at 1580 cm −1 is typical for the C=C stretching vibration. For the Fe@C material, this band is very weak and is upshifted to 1620 cm −1 , probably due to more symmetric structure of carbon sheets in the nonfunctionalized carbon encapsulates. The broad band at 1180 cm −1 corresponds to the C-O vibrations. These results undoubtedly prove that the surface functional groups were successfully introduced on the carbon surface. The presence of surface functional acidic groups  Fig. 3 Energy-dispersive X-ray spectroscopy of the Fe@C sample and representative TEM image of this sample (insert) was also confirmed by Boehm titration. The raw (Fe@C/Fe) and purified (Fe@C) nanoparticles have very low content of surface carboxylic groups (below 0.02 mmol/g). The surface-functionalized samples (Fe@C-COOH) have substantially higher surface carboxylic groups (up to 0.53 mmol/g). The relative fractions of various other functional acidic groups (e.g., phenolic and lactonic) have also been found on a similar level for the surface-modified CEINS (ca. 0.3 and 0.28 mmol/g, respectively). The raw (Fe@C/Fe) and purified (Fe@C) samples have very low content of the total surface acidic groups such as carboxylic, phenolic, and lactonic (below 0.2 mmol/g). The surface-functionalized CEINS have substantially higher total acidity at the surface (ca. 1.11 mmol/g).

Magnetic measurements
The magnetic hysteresis loops are shown in Fig. 7. All the studied CEIN samples have typical ferromagnetic characteristics with a relatively low coercive field. The

Phantom relaxometry measurements
The relaxation study of the CEIN samples was performed at the magnetic field of 1.5 T using a typical clinical scanner (the experimental setup is shown in Suppl. Fig. S1). In the present study, T1weighted images were acquired using the inversion recovery SE sequence with eight inversion times (TI 50, 100, 200, 400, 800, 1200(TI 50, 100, 200, 400, 800, , 1600(TI 50, 100, 200, 400, 800, , and 2400. For the T2-weighted images, the SE sequence with 16 variable echo times (TE 22,44,66,88,110,132,154,176,198,220,242,264,286,308,330,and 352 ms) was acquired. The signal intensity versus the time of inversion for different concentrations of CEINS is shown in Fig. 8. The curves clearly evidence the returning process to the equilibrium magnetization after inversion in absolute values at the field strength of 1.5 T. The longitudinal magnetization almost returns to the point of origin. This effect, however, is not observed at two higher concentrations of 25 and 50 μg/ml for both the as-synthesized (Fe@C/Fe) and purified (Fe@C) CEIN samples, respectively. As shown in Fig. 8 (top and middle), the signal intensities of the CEIN samples are completely diminished. This effect is not observed for the Fe@C-COOH sample (Fig. 8 bottom). The representative T1-weighted images are presented in Fig. 9. The as-synthesized raw and purified nanoparticles slightly decrease the T1 time at higher concentrations, i.e., only at 25 and 50 μg/ml, but this effect is not observed for the functionalized CEINS containing the surface carboxylic groups (Suppl. Fig. S2A). The relaxation rates (r1 = 1/T1) calculated for the studied CEIN samples show a quite opposite direction (Suppl. Fig.  S2B). This is also confirmed in expressing the 1/T1 values versus iron (Fe) concentrations in different CEIN samples. As shown in Fig. 10, the r1 = 1/T1 values exhibit a linear dependence on Fe concentration in the phantom gelatin media. In the inversion recovery T1 studies, the fit (R-squared) is found to be good for the Fe@C/Fe (R 2 = 0.9626, p < 0.05), Fe@C (R 2 = 0.9122, p < 0.05), and Fe@C-COOH (R 2 = 0.9628, p < 0.05) samples measured at the magnetic field strength of 1.5 T; hence, the r1 relaxivity value is also determined accurately (please see the left panel of Fig. 10). The r1 relaxivities are calculated as the slope of the plot of 1/T1 vs. iron concentration to compare the contrastenhancing effect due to purification and surface functionalization processes for the as-synthesized CEINS. The r1 relaxivity values are shown in Table 2. It is interesting to note that the raw (Fe@C/Fe) and purified (Fe@C) CEINS have a higher contrastenhancing effect than the surface-functionalized (Fe@C-COOH) samples (Table 2).
The T2 relaxation curves of CEINS studied in the phantom gelatin media with different iron concentrations are shown in Fig. 11. The representative T2weighted images are presented in Fig. 12. The signal intensity expressed as arbitrary units for different echo times at the field strength of 1.5 T is steeply decreased with increasing CEINS concentration, but remains roughly unchanged at two higher concentration doses (25 and 50 μg/ml), thereby achieving the minimal value as compared to that of the background intensities (Fig. 11). In general, a similar profile of signal intensity is observed for both the naked (raw and purified CEINS) and surface-functionalized encapsulates containing acidic groups (Fe@C-COOH). This behavior was also confirmed on T2 relaxation times (Suppl. Figs. S3A, S3B). At the magnetic field strength of 1.5 T, the relationship between the 1/T2 and Fe content is nearly linear for all carbon-encapsulated magnetic nanoparticles studied here, thus achieving the R-squared values of 0.9711 (p < 0.05), 0.9585 (p < 0.05), and 0.9539 (p < 0.05) for Fe@C/Fe, Fe@C, and Fe@C-COOH, respectively (Fig. 10, right panel). The r2 relaxivities, calculated from the slopes 1/T2 vs. Fe concentration, obtained from different CEIN samples are shown in Table 2. This study evidences that the as-synthesized raw Fe@C/Fe sample has the highest r2/r1 ratio among all the studied CEIN samples.

Discussion
The major aim of this study was to perform the comprehensive magnetic characterization of CEINS used as a new contrast drug candidate for MRI. We synthesized three different batches of CEINS by using the arc plasma method and examined the as-synthesized raw (Fe@C/ Fe), purified (Fe@C), and surface-functionalized nanoparticles mainly containing carboxylic groups (Fe@C-  Fig. 9 Representative T1-weighted images of the gelatin phantoms composed of CEINS. T1-weighted images were acquired using the inversion recovery SE sequence with different inversion times (TI 50, 100, 200, 400, 800, 1200(TI 50, 100, 200, 400, 800, , 1600, and 2400 ms). The remaining spatial resolution parameters were acquired as follows: TR (3500 ms), TE (14 ms), echo spacing (14.1 ms), voxel size (1.1 × 1.1 × 2.0 mm), field of view (140 mm), flip angle (150 deg), slice thickness (2.0 mm), and averages (3) COOH). The study showed that the as-obtained CEINS have similar morphological features. They were composed of mainly spheroidal nanoparticles with a typical core-shell structure containing different amounts of iron phases. All CEINS samples had typical ferromagnetic characteristics with a relatively low coercive field. We also observed some differences between the coercive field obtained for the as-synthesized raw nanomaterials and two other CEIN samples; these differences were mainly related to different size distribution and plausibly larger superparamagnetic content in the as-synthesized raw material. It should be noted that saturation magnetization was also quite different for all the CEIN samples. Among the examined CEINS, the as-  Fe@C-COOH 2.10 ± 0.14 69.81 ± 7.50 33.2 synthesized raw material was found to have the highest magnetization on magnetic hysteresis loops.
In the present study, the effects of carbonencapsulated iron nanoparticles on the spin-lattice (T1) and spin-spin (T2) relaxation processes were studied using the phantom water-gelatin models (in vitro). The experiments were performed at the low magnetic field strength of 1.5 T by using an MRI clinical scanner equipped with the standard RF "birdcage" type head coil. In the experiments, both the longitudinal (T1) and transversal (T2) relaxation times and the r1 and r2 relaxivities were examined due to different iron phase composition and surface features of the studied CEIN samples. To obtain data, T1-and T2-weighted MR images were acquired using the inversion recovery SE and the SE with multiple TE, respectively. The present data showed that the raw and purified CEIN samples had significantly lower r1 relaxivities than the surfacefunctionalized CEIN samples. The as-synthesized raw Fe@C/Fe samples were noted to have the surface charge density (C/m 2 ) nearly equal to zero due to the absence of any surface acidic groups (Grudzinski et al. 2013). In contrast, the carbon surface density of the Fe@C (0.20 C/m 2 ) and Fe@C-COOH (1.12 C/m 2 ) samples were found to have some acidic sites implicated by the point of zero charge values, which were definitely lower than 7.0 in the CMC media (Grudzinski et al. 2013). Therefore, it seems reasonable that the exchange rate of hydrated protons was more pronounced in those CEINS where the protons are available in the surface acidic groups. As such, the proton exchange in the surface acidic groups, mainly carboxylic (-COOH), may play a significant role in affecting the r1 relaxivity in the examined CEIN samples. Note that different surface coatings can largely affect T1 relaxation times due to the inner, intermediate, and outer sphere relaxivities in nanoparticle-based MRI (Caravan et al. 2009;Roch et al. 2005). More detailed mathematical approaches and discussions on these processes have been recently presented (Zhang et al. 2018). The effect of CEINS on the T2 spin-spin relaxation is a c o n s i d e r a b l y di ff e r e n t p r o c e s s f r o m t h e abovementioned T1 spin-lattice relaxation. T2 relaxation occurs through three major mechanisms, which are related to (i) Curie spin relaxation, which originates from the dipolar interaction between water protons and a large static magnetic moment arising from electrons, (ii) dipole-dipole coupling between the metal ions and the water hydrogen nuclei, and (iii) scalar or contact relaxation processes (Caravan et al. 2001;Krishnan 2010). Our present study showed that the r2 relaxivity, as calculated from 1/T2 vs. iron concentration, resulted in the highest values for the Fe@C/Fe sample among the studied CEIN samples. This is a direct consequence of magnetic properties of this sample. In fact, the Ms value for the Fe@C/ Fe material significantly exceeded those of the two other studied samples, i.e., Fe@C and Fe@C-COOH. This observation is also in agreement with the chemical composition of the studied CEINS, i.e., the Fe content changes in the following order: Fe@C/Fe > Fe@C > Fe@C-COOH (Grudzinski et al. 2013). There was also a clear correlation between the Ms values and Fe content evaluated from TGA measurements in our study (the higher the Fe content, the higher was the Ms value). In general, Fe@C/Fe made of high Ms materials can more efficiently induce field inhomogeneity and can plausibly influence a greater volume of water. Therefore, it can be hypothesized that the T2 relaxation is mostly dominated by scalar or contact relaxation processes in these core-shell-type nanoparticles. More recently, the coating thickness was shown to influence the relaxivities of iron oxide nanoparticles (LE et al. 2017). The authors demonstrated that the thinner the stabilizing layer is, the larger is the r2 relaxivity. For CEINS, the carbon coating, which is composed of few graphene layers, has the same thickness (ca. 3-4 nm) in all CEIN samples (please see insert in Fig. 3); thus, it cannot be concluded that the carbon surface diameter would strongly affect the r2 relaxivity in the studied CEINS. Other recent studies showed that the r2  Fig. 12 Representative T2-weighted images of gelatin phantoms composed of CEINS. T2-weighted images were acquired using the SE sequence with TR (3570 ms) and variable echo times (TE: 22,44,66,88,110,132,154,176,198,220,242,264,286,308,330,and 352 ms). The remaining spatial resolution parameters were as follows: acquisition matrix of 256 × 256, voxel size (1.4 × 1.1 × 5.0 mm), field of view (270 mm), slice thickness (5 mm), flip angle (180 deg), and averages (1) relaxation rates are sensitive to some aggregation states, hydrodynamic size, and morphology of iron oxide nanoparticles (Ebert et al. 2015;Joos et al. 2017;Ta et al. 2017). Among the CEIN samples studied here, the raw nanoparticles were found to have the highest potential to undergo agglomeration and aggregation processes as compared to purified and surface-functionalized CEINS. This was also confirmed in the present study on the hydrodynamic size of CEINS measured with DLS performed on the water-CMC suspensions of Fe@C/Fe, Fe@C, and Fe@C-COOH samples (Table 1) and the gelatin type B phantoms of these nanoparticles (Suppl. Fig. S4A-D). Comparing the results obtained for T1/T2 relaxivity properties, it can be noted that carbonencapsulated magnetic nanoparticles result in a higher ratio (r2/r1) for the as-synthesized and purified samples than those calculated for the surface-functionalized CEINS containing some carboxylic groups. Note that the r2/r1 ratio for CEINS, ranging from 33.2 to 48.0, was higher than those measured at the same magnetic field strength for iron oxide nanoparticles in human fluids such as serum and whole blood, phosphate-buffered saline, and water (Ahmad et al. 2018;Jarrett et al. 2007;Knobloch et al. 2018). Because the r1 and r2 relaxivities are magnetic field-dependent, in general, the r1 value decreases with the increasing magnetic field, but the r2 value usually increases; thus, the application of 1.5 T seems to be reasonable because most of the MRI clinical scanners used worldwide operate in this field. Therefore, CEINS could be considered as an excellent negative contrast agent in the low magnetic field (1.5 T). Because purification and surface decoration processes are further required in CEINS to obtain nanoparticles that can recognize some specific molecular targets in animal and human tissues, one should be aware that such surface processes may affect the relaxivity of the final contrast drug candidate. Therefore, more detailed studies are required to elucidate the magnetic relaxation effects of CEINS decorated with different chemical and biological ligands. A more through approach on relaxation studies is also needed for CEINS studied in both cellular and preclinical animal models.

Conclusion
We synthesized CEINS by using a carbon arc discharge route and revealed their physicochemical properties and MRI contrast abilities at the field strength of 1.5 T. The experiments presented here showed that CEINS demonstrate strong transverse relaxivities as a potent highefficacy T2 (negative) contrast agent. Furthermore, the r2/r1 ratio for the studied CEINS samples was much higher than that of other reported iron oxide nanoparticles examined at the same magnetic field strength. The results showed that the hydrodynamic size, chemical composition, and surface characteristic of CEINS play a significant role in reflecting the relaxivity performance of these nanoparticles. The relevance of this study could be the level of its potent applicability because CEINS as a T2 contrast agent for MRI could open a new frontier for the core-shell (iron-carbon)-type theranostics.