Ultrasmall Ferrite Nanoparticles Synthesized via ... - ACS Publications

Apr 3, 2017 - Department of Anatomy, Yong Loo Lin School of Medicine, National University ... in vivo high-resolution blood pool and liver-specific MR...
0 downloads 0 Views 6MB Size
Ultrasmall Ferrite Nanoparticles Synthesized via Dynamic Simultaneous Thermal Decomposition for High-Performance and Multifunctional T1 Magnetic Resonance Imaging Contrast Agent Huan Zhang,†,⊥ Li Li,‡,⊥ Xiao Li Liu,† Ju Jiao,§,⊥ Cheng-Teng Ng,∥ Jia Bao Yi,Δ Yan E Luo,† Boon-Huat Bay,∥ Ling Yun Zhao,# Ming Li Peng,† Ning Gu,¶ and Hai Ming Fan*,† †

Key Laboratory of Synthetic and Natural Functional Molecule Chemistry of the Ministry of Education, College of Chemistry and Materials Science, Northwest University, Xi’an, Shaanxi 710069, China ‡ State Key Laboratory of Oncology in South China, Imaging Diagnosis and Interventional Center, Sun Yat-sen University Cancer Center, Guangzhou 510060, China § Department of Nuclear Medicine, The Third Affiliated Hospital of Sun Yat-sen University, 600 Tianhe Road, Guangzhou, Guangdong 510630, China ∥ Department of Anatomy, Yong Loo Lin School of Medicine, National University of Singapore, 4 Medical Drive, MD10, 117594, Singapore Δ School of Materials Science and Engineering, University of New South Wales, Kensington, NSW 2052, Australia # State Key Laboratory of New Ceramics and Fine Processing, Key Laboratory of Advanced Materials, School of Material Science & Engineering, Tsinghua University, Beijing 100084, China ¶ State Key Laboratory of Bioelectronics, Jiangsu Key Laboratory for Biomaterials and Devices, School of Biological Science and Medical Engineering, Southeast University, Nanjing 210096, China S Supporting Information *

ABSTRACT: Large-scale synthesis of monodisperse ultrasmall metal ferrite nanoparticles as well as understanding the correlations between chemical composition and MR signal enhancement is critical for developing next-generation, ultrasensitive T1 magnetic resonance imaging (MRI) nanoprobes. Herein, taking ultrasmall MnFe2O4 nanoparticles (UMFNPs) as a model system, we report a general dynamic simultaneous thermal decomposition (DSTD) strategy for controllable synthesis of monodisperse ultrasmall metal ferrite nanoparticles with sizes smaller than 4 nm. The comparison study revealed that the DSTD using the iron-eruciate paired with a metal-oleate precursor enabled a nucleation-doping process, which is crucial for particle size and distribution control of ultrasmall metal ferrite nanoparticles. The principle of DSTD synthesis has been further confirmed by synthesizing NiFe2O4 and CoFe2O4 nanoparticles with well-controlled sizes of ∼3 nm. More significantly, the success in DSTD synthesis allows us to tune both MR and biochemical properties of magnetic iron oxide nanoprobes by adjusting their chemical composition. Beneficial from the Mn2+ dopant, the synthesized UMFNPs exhibited the highest r1 relaxivity (up to 8.43 mM−1 s−1) among the ferrite nanoparticles with similar sizes reported so far and demonstrated a multifunctional T1 MR nanoprobe for in vivo high-resolution blood pool and liver-specific MRI simultaneously. Our study provides a general strategy to synthesize continued... Received: November 14, 2016 Accepted: April 3, 2017 Published: April 3, 2017 © 2017 American Chemical Society

3614

DOI: 10.1021/acsnano.6b07684 ACS Nano 2017, 11, 3614−3631

Article

www.acsnano.org

Article

ACS Nano

ultrasmall multicomponent magnetic nanoparticles, which offers possibilities for the chemical design of a highly sensitive ultrasmall magnetic nanoparticle based T1 MRI probe for various clinical diagnosis applications. KEYWORDS: ultrasmall ferrite nanoparticles, dynamic simultaneous thermal decomposition, T1 MR contrast agent, chemical composition effect, magnetic resonance imaging, liver-specific MRI,

T

ernary compound nanoparticles with superior composition-tunable properties have long been of scientific and technological interest.1−6 In particular, the metal ferrite nanoparticle MFe2O4 (M = Mn, Co, Ni, etc.) represents an important class of magnetic ternary compound nanoparticles, which have found wide biomedical applications, mostly as highly sensitive magnetic resonance imaging (MRI) nanoprobes for in vivo and noninvasive detection of clinically important biological targets.2,7−10 Recent advances in this field have revealed that the ultrasmall magnetite nanoparticles with sizes smaller than 5 nm can exhibit remarkable and counterintuitive T1 MR enhancement due to the strong size-related effects,11,12 in stark contrast to the conventional superparamagnetic iron oxide based contrast agent (CA) such as Feridex, which is the representative T2 contrast agent. Despite its apparent advantages over the commercial gadolinium-based T1 MRI contrast agent, results from currently available ultrasmall ferrite nanoparticle-based T1 MRI probes have not been optimal. Artificially molecular engineering of chemical composition in magnetic ferrite nanoparticles has been frequently employed to optimize their application performance.2,7,13 For example, the MR relaxation and biochemical properties of magnetic MFe2O4 nanoparticle can be altered by adjusting the chemical indentity of M2+.2,7 However, synthesis of ultrasmall metal ferrite nanoparticles with well-controlled size and composition simultaneously remains a challenging task due to the complex dynamic nature of the nucleation and growth process in the multicomponent system. For instance, currently reported MFe2O4 nanoparticles synthesized by hydrolytic/ nonhydrolytic process either have large particle size or show poor monodispersity and crystal quality, which hampers their applications as ultrasensitive T1 MR contrast agents.14−22 Herein, taking monodisperse ultrasmall MnFe2O4 nanoparticles (UMFNPs) as a model system, we report a dynamic simultaneous thermal decomposition strategy to couple the multicomponent chemical doping process with the nucleation process, which allows us to achieve good control in particle size and distribution of ultrasmall metal ferrite nanoparticles for highly sensitive and multifunctional T1 MR contrast agents. Co-thermal decomposition of mixed metal precursors to prepare monodisperse magnetic MFe2O4 nanoparticles through the heating-up process has been well-established17−20,23,24 and has also been applied to synthesize other ternary compound nanoparticles.25 However, due to the intricate growth dynamics arising from the thermal decomposition process of mixed precursors,26 the control of size, morphology, composition, and crystal quality is thus quite difficult. In terms of precursor decomposition behavior, two different scenarios, dynamic nonsimultaneous thermal decomposition (DNSTD) and dynamic simultaneous thermal decomposition (DSTD) processes, commonly occur under certain synthetic conditions. Figure 1 schematically illustrates the growth dynamics of DSTD and DNSTD synthesis of MnFe2O4 using a modified LaMer diagram. We presume that the intermediate species generated by the dynamic dissociation of the metal complexes act as monomer. Stage I depicts the generation of monomer by on-site dissociation of molecule precursors, the accumulation of the

Figure 1. Diagram for DSTD and DNSTD synthesis of manganese ferrite nanoparticles.

monomer, and the burst of nucleation, while stage II describes the diffusion-limited growth. The temperature continuously increases during the heat-up process until reaching the final reactive temperature. For the DNSTD process, one metal precursor will be dissociated first due to the large difference in decomposition kinetics of the two metal precursors, and the rapid accumulation of a single-metal-containing monomer results in the formation of a non-dopant-containing nucleus in stage I. Upon the decomposition of another metal precursor in stage II, these nuclei serve as the host to grow ternary MnFe2O4 nanoparticles. In order to achieve high-quality nanoparticles, the final reaction temperature should be sufficiently high to activate solid−solid ion diffusion.27 As a result of the high reaction temperature and growth-doping process,28 the obtained MnFe2O4 nanoparticles via the DNSTD process usually have sizes larger than 6 nm and show a strong T2 MR signal.12,13 Mixing the dopant and host at stage I may further reduce the particle size and improve the quality of the nanoparticle. The DSTD is thus defined as a special co-thermal decomposition reaction where the precursors have an approximate on-site dissociation temperature (the offset within ΔT), and a multiplemetal-containing monomer can be formed before the nucleation. Unlike the hot-injection process, in which nucleation- or growthdoping can be decoupled by altering the reaction temperature,28 the regulation and control of growth dynamics for DNSTD/ DSTD is realized by tuning the thermal stability of the precursors. In this context, seeking suitable precursors is of utmost importance for successfully achieving the DSTD process. 3615

DOI: 10.1021/acsnano.6b07684 ACS Nano 2017, 11, 3614−3631

Article

ACS Nano

ultrasmall metal ferrite nanoparticles in a controllable manner, which undoubtedly offers the possibilities for designing highly sensitive ultrasmall magnetic nanoparticle based T1 MRI probes for various clinical diagnosis applications.

However, variations of on-site dissociation temperatures of currently available metal precursors such as iron/manganese acetylactonate and carboxylate complexes, are usually too large to be paired. With due credit to Song and Zhang for demonstrating the feasibility of artificial engineering of the precursor thermal stability for the synthesis of high-quality MnFe2O4 nanoparticles,20 we were inspired to consider the possibility of DSTD synthesis of utlrasmall metal ferrite nanoparticles by devising green and inexpensive precursors. Thermal stability of widely used precursors of coordination compounds in a given metal cation can be easily modulated by altering the ligands.20,29 Unfortunately, such modification may also affect the growth kinetics of the nanoparticle by diverse ligand effects,30 leading to derivation of an uncertain product and complicated optimization of reactive parameters mediated by trial and error. Hence, tailoring the molecular structure of currently available coordination compound precursors to a slight extent is a viable solution to reduce possible side effects. Among the various coordination compound precursors used for the preparation of metal ferrite nanoparticles, metal-oleates are favored because of their relative nontoxicity and usability. Furthermore, the formation kinetics of the binary ferrite (Fe3O4) nanoparticles with high monodispersity and tunable size obtained by iron-oleate precursors in the heating up process has been well-studied,19,23 which may provide clues on the understanding of the growth dynamics of metal ferrite nanoparticles in the DSTD process. The current issue for DSTD synthesis of UMFNPs is the distinct difference between the on-site dissociation temperature of metal-oleate precursors, which is 215 °C for iron-oleate and 311 °C for Mn-oleate. Therefore, we prepared Fe-eruciate, a new type of coordination compound precursor with improved thermal stability, where a short-chain ligand of oleate (18-carbon chain) is substituted with long-chain eruciate (22-carbon chain) (Figures S1−S4). Both thermogravimetric analysis (TGA) and temperature-dependent FTIR spectra revealed that Fe-eruciate decomposes at around 326 °C, which is very close to that of Mn-oleate (Figure S5). By using this green Fe-eruciate precursor, we were able to achieve a DSTD process and attain ultimate control of the UMFNPs with regard to uniformity of size, single crystallinity, and stoichiometry. The size of UMFNPs can be easily tuned from 2 nm to 4 nm. A comparison study of DSTD and DNSTD synthesis of MnFe2O4 nanoparticles revealed that the growth behavior of UMFNPs in the DSTD process was analogous to that of binary iron oxide nanoparticles using single precursors, while the DNSTD process in the same conditions may prevent effective nucleation and growth separation, resulting in nonuniform particles. As a proof of concept, the Fe-eruciate precursorpromoted DSTD has been successfully applied to prepare other ultrasmall ferrite nanoparticles such as CoFe2O4 and NiFe2O4 with sizes of 3 nm. Moreover, the success in DSTD synthesis of ultrasmall metal ferrite nanoparticles allows us to investigate the correlations between the composition and MR T1 signal enhancement in such a system. The molecularly engineered UMFNPs have exhibited the lowest r2/r1 ratio (2.49) at 3 T that has been reported thus far for ultrasmall ferrite nanoparticles, as well as a large r1 relaxivity (up to 8.43 mM−1 s−1) that is 2.09 times higher than a commercial gadolinium complex based contrast agent (Omniscan). In vivo MRI imaging carried out on a rat model verified that the UMFNPs could be used as ultrasensitive and multifunctional T1 MR contrast agents for high-resolution MR imaging of blood pools and liver. The study aims to provide a general DSTD strategy for synthesizing

RESULTS AND DISCUSSION Synthesis of the UMFNPs. Recent studies on ultrasmall ferrite nanoparticle-based T1 MR contrast agents have revealed that the MR signal enhancement is highly dependent on the particle size and uniformity, as well as composition and surface chemistry of the nanoparticles.31−33 The ability to synthesize metal ferrite nanoparticles with extremely small size, narrow size distribution, and stoichiometric composition is thus vital for high-performance MRI applications. Among various ferrite nanoparticles, the UMFNPs are an ideal model system to investigate the controllable growth mechanism because the T1 relaxivity of ultrasmall magnetic nanoparticles may be significantly enhanced by molecular engineering of the Mn dopant.34 Herein, the UMFNPs were synthesized by DSTD of mixed ironeruciate and manganese-oleate complexes in the presence of oleic acid and oleyl alcohol in benzyl ether or 1-octadecene. The oleyl alcohol is used to further reduce the on-site decomposition temperature of the precursors, facilitating the formation of small-sized nanoparticles.31 By varying the ratio of oleyl alcohol to oleic acid and the final reaction temperature, the particle size could be finely tuned from 2 nm to 3.9 nm. Detailed reaction parameters are summarized in Table S2. The low-magnification TEM images of as-prepared UMFNPs with average sizes of 2, 3, and 3.9 nm are shown in Figure 2a−c, demonstrating that UMFNPs could be synthesized in a large scale. Corresponding high-magnification TEM images (Figure 2d−f) demonstrate that the UMFNPs were fairly uniform in size with a narrow distribution (Figure S8a−c). The lattice fringes shown in the high-resolution TEM images (Figure 2g−i) are an indication of the high crystallinity of the UMFNPs. The observed lattice spacings of 2.53, 2.45, and 3.01 Å are for 2, 3, and 3.9 nm sized UMFNPs, correspond to the d spacings of the (311), (222), and (220) lattice planes, respectively. Inductively coupled plasma− atomic emission spectroscopy (ICP-AES) reveals that the Mn/Fe ratios of all UMFNPs are about 0.5, which is consistent with their stoichiometric ratios. The powder X-ray diffraction (XRD) patterns of as-prepared UMFNPs (Figure 3a) were in good accord with the standard spinel MnFe2O4 powder diffraction data (JCPDS card no. 10-0319), and no other secondary phases such as manganese oxide or ferrous oxide could be traced. The broadening of diffraction peaks with decreased particle size is due to the small crystallite size.35 The mean sizes estimated on the basis of the strongest (311) peaks using the Debye−Scherrer equation were 2.07, 2.98, and 4.29 nm, which are basically consistent with the observations under TEM imaging. Figure 3b and c show the Fe 2p and Mn 2p X-ray photoelectron spectroscopy (XPS) of the UMFNPs. The core level binding energies at 711.6 and 724.7 eV for 3.9 nm UMFNPs, 712.7 and 725.8 eV for 3.0 nm UMFNPs, and 712.5 and 726.1 eV for 2.0 nm UMFNPs are attributed to the characteristic doublets of Fe 2p3/2 and Fe 2p1/2 of Fe3+, respectively. The binding energies of Mn 2p3/2 and Mn 2p1/2 were 642.4 and 653.7 eV for 3.9 nm UMFNPs, 642.9 and 654.7 eV for 3.0 nm UMFNPs, and 642.9 and 654.7 eV for 2.0 nm UMFNPs, respectively. All the peaks from the Fe 2p and Mn 2p spectra are basically in good agreement with previously reported values for Fe3+ and Mn2+.36−38 The slight shifts toward high energy can be found for 2 and 3 nm UMFNPs with respect to that for 3.9 nm UMFNPs, 3616

DOI: 10.1021/acsnano.6b07684 ACS Nano 2017, 11, 3614−3631

Article

ACS Nano

Figure 2. TEM and high-resolution TEM images of UMFNPs: (a, d, g) 2 nm UMFNPs, (b, e, h) 3 nm UMFNPs, and (c, f, i) 3.9 nm UMFNPs.

of the breaking of bonds in the carboxylate group and dissociation of these complexes. The approximate dissociation temperature for Fe-eruciate and Mn-oleate complexes suggests that the formation of UMFNPs could undergo a DSTD process, while the DNSTD process may take place when co-thermal decomposition of Fe-oleate and Mn-oleate complexes occurs due to the large difference (>20 °C) in the on-site dissociation temperature. A comparison study of the DSTD and DNSTD process was thus performed to further understand the dynamic nature in the nucleation and growth of the UMFNPs. The DSTD and DNSTD processes were established using Fe-eruciate/ Mn-oleate complexes and Fe-oleate/Mn-oleate complexes as the precursors, respectively. The initial Mn/Fe ratio of the mixed precursors is 0.5, and the reaction parameters are the same as those for the synthesis of 3 nm UMFNPs. The reaction temperature rising profile for DSTD and DNSTD is shown in Figure 5a. The final reaction temperature in this study was 265 °C, and it took about 48 min to reach this temperature. The temporal evolution of particle size, size distribution, and composition ratio for DSTD and DNSTD growth is presented in Figure 5b−d. The representative TEM images of the products obtained at 35, 39, 48, and 78 min during the DSTD and DNSTD process are shown in Figure 5e−l. For a better understanding, the whole nanoparticle growth process has been divided into three regimes marked as dashed lines. Regime I starts from the beginning of the heat-up process and ends at a

which could possibly be due to a significantly increased surface state in small-sized particles.39 We posited that the DSTD process is essential for the synthesis of highly monodisperse UMFNPs, although the exact mechanism remains to be elucidated. Hence, we first investigated the dynamic thermal decomposition behaviors of the coordination compound precursors by using temperature-programmed FTIR in the presence of oleyl alcohol. Figure 4 shows the FTIR spectra of Fe-eruciate, manganese-oleate, and iron-oleate complexes at various temperatures, respectively. The characteristic bands for the iron-eruciate complex appeared at 1590, 1531, and 1445 cm−1 (also found in Figure S4a). The bands at 1590 and 1531 cm−1 arose from the asymmetric vibration of the carboxylate group, while the band at 1445 cm−1 is attributed to the carboxylate symmetrical stretching vibration.40 Mn-oleate and Fe-oleate complexes showed three similar vibration bands. While the bands at 1611 and 1553 cm−1 (Mn-oleate) and 1595 and 1526 cm−1 (Fe-oleate) are assigned to the asymmetrical vibration of carboxylate, the bands at 1425 cm−1 (Mn-oleate) and 1443 cm−1 (Fe-oleate) are attributed to the symmetrical vibration of carboxylate.40 Noticeably, the intensity of asymmetrical vibration of carboxylate for all complexes decreased with an elevated temperature. This vibration band for the Fe-oleate complex finally disappeared at a temperature of about 190 °C (Figure 4c), but remained visible for both Fe-eruciate and Mn-oleate complexes until the temperature was higher than 220 °C (Figure 4d). The vanishing of the band is a characteristic 3617

DOI: 10.1021/acsnano.6b07684 ACS Nano 2017, 11, 3614−3631

Article

ACS Nano

Figure 3. (a) XRD patterns of the different sized UMFNPs. (b) Fe 2p and (c) Mn 2p XPS spectra of the UMFNPs.

Figure 4. Temperature-programmed Fourier transform infrared spectroscopy (FTIR) of (a) iron-eruciate complex, (b) manganese-oleate complex, and (c) iron-oleate complex in the presence of oleyl alcohol. The asymmetric vibration of the carboxylate group is marked by red arrows. (d) Plot of the temperature dependence of the peak intensity of the asymmetric vibration of the carboxylate group in the FTIR spectra.

reaction time of 39 min, at which the reaction temperature reaches the dissociation temperature of Fe-eruciate/Mn-oleate complex precursors (220 °C). Regime 2 covers the time slot from 39 to 78 min, during which the reaction is terminated for real UMFNP growth by rapid cooling. Regime 3 is the additional aging process from 78 to 168 min. As shown in Figure 5b, for the DSTD process, no nanoparticle was formed in regime I (Figure 5e), as the reaction temperature was below the dissociation temperature of Fe-eruciate/ Mn-oleate complexes. In regime II, the particle size was observed to rapidly increase from 1.68 nm (39 min, Figure 5f) to 2.58 nm (Figure 5g), and then the particle size slowly increased to 3 nm from 48 to 78 min. The slow increase in particle size can still be

observed in regime III. A similar trend in particle size variation has been observed for the DNSTD process except that the nanoparticle was formed in regime I (Figure 5i), which is earlier than the DSTD process. The particle sizes in the DNSTD process were always much larger than that in the DSTD process at a designated reaction time. The size distribution for the DSTD process was “focused” from 17.31% to 15.68% and then maintained throughout regime II (Figure 5c), before gradually increasing up to 30% in regime III. In contrast, the size distribution observed in the DNSTD process started out with a standard deviation of 30.49%, which is much larger than that observed in the DSTD process. There was an initial slow decrease from 30.49% to 28.53%, before a sudden jump to 35.36% at 3618

DOI: 10.1021/acsnano.6b07684 ACS Nano 2017, 11, 3614−3631

Article

ACS Nano

Figure 5. Comparison study of DSTD and DNSTD synthesis of manganese ferrite nanoparticles. (a) Temperature profile, (b) particle size, (c) standard deviation, and (d) Mn/Fe ratio evolution in the DSTD and DNSTD processes. TEM images of the representative products taken at reaction times of (e and i) 35 min, (f and j) 39 min, (g and k) 48 min, and (h and l) 78 min during DSTD and DNSTD processes.

initial products obtained at 39 min were higher than 0.42, indicating the high Mn/Fe ratio in the nucleus. This ratio rapidly decreased from 0.42 to 0.16 in regime II and increased slightly in regime III. But for the DNSTD process, the Mn/Fe ratios of the initial products were quite low (0.06), suggesting that the nucleus contained mainly iron oxide. It then escalated to 0.19 at 39 min of regime I, followed by a rapid decrease to 0.11, and was maintained over the entire regime II. Similar to the DSTD process, a slight increase in the Mn/Fe ratio during the aging process was observed.

around 39 min in regime I (Figure 5c). The size distribution in the DNSTD process was “refocused” to 29.14% in regime II and then gradually increased to 38.78% in regime III. In comparison with nearly spherical nanoparticles obtained in the DSTD process, the nanoparticles generated by the DNSTD process were nonuniform and irregularly shaped with a relatively large size distribution, as shown in Figure 5i−l. The variation of Mn/Fe ratio in the nanoparticles during the growth was also determined analytically for each aliquot using ICP-AES as shown in Figure 5d. As for the DSTD process, the Mn/Fe ratios of the 3619

DOI: 10.1021/acsnano.6b07684 ACS Nano 2017, 11, 3614−3631

Article

ACS Nano

hot-injection process.41 Unfortunately, the multistep sequential decomposition in this case can prevent effective separation of nucleation and growth, as evidenced by the irregular shape and broad size distribution of the obtained nanoparticles. Although such defects may be largely eliminated in the growth of large nanoparticles by elaborately designed synthetic conditions such as reducing the temperature rising rate and elevating the final reaction temperature,42 our results clearly show its limitation in the synthesis of highly monodisperse UMFNPs. In addition, the variation in particle size and size distribution during the aging process is an indication that nanoparticle growth in both DSTD and DNSTD undergoes a classic Ostwald ripening process.43 The elongated reaction time will not affect the Mn/Fe ratio significantly, as the slight increase possibly originated from the uncompleted decomposition of the Mn-oleate complex. It would appear that the findings obtained from the comparison study indicate that the control of particle size and size distribution of UMFNPs in DSTD follow the same rules constructed for binary iron oxide nanoparticles; however, due to the different decomposition kinetics of the two precursors, the appropriate optimization of their initial ratio is necessary for the synthesis of stoichiometric UMFNPs. To demonstrate its generality, the principle of DSTD has been applied to synthesize other ultrasmall ferrite nanoparticles such as NiFe2O4 and CoFe2O4. The Ni-oleate and Co-oleate complexes were paired with an Fe-eruciate complex for DSTD synthesis because of their approximate thermal decomposition temperatures (Figure S7, Ni-oleate 324.7 °C and Co-oleate 340.9 °C). Figure 6a and b show the TEM images of the as-prepared ultrasmall NiFe2O4 and CoFe2O4 nanoparticles. As shown in Figure 6a and b, the particles sizes were 3.1 and 3.4 nm

On the basis of the above experimental results, it is clear that the metal ferrite nanoparticles synthesized through DSTD and DNSTD will undergo different nucleation and growth kinetics, where DSTD is crucial for the synthesis of high-quality UMFNPs. As illustrated in Figure 5, after rapid nucleation in the DSTD process, the fast growth of the nanoparticles and the narrowing of their size distribution occurred simultaneously. This behavior is essentially the same as that previously observed in the thermal decomposition synthesis of Fe3O4 binary ferrites using single Fe-oleate precursors,19,23 which suggests the effective separation of nucleation and growth in the DSTD process. Further evidence can be found in the evolution of the composition ratio. The high Mn/Fe ratio in the nucleus indicates that the dual-metal mixed monomer was formed before the nucleation. Moreover, the subsequent decrease of the Mn/Fe ratio in the growth stage could be ascribed to the faster decomposition rate of the Fe-eruciate complex than the Mn-oleate complex (Figure S5d). Benefiting from both the effective separation of nucleation and growth and the diffusionlimited growth, highly monodisperse UMFNPs could be synthesized via the DSTD process by lowering the reaction temperature and shortening the growth time simultaneously. As for the DNSTD process, the extremely low initial Mn/Fe ratio implies that the nucleation of the iron oxide host occurred first. The sudden increase in both particle size and size distribution at 39 min (220 °C) accompanied by a rapidly increased Mn/Fe ratio suggests explicitly that the growth-doping process was induced by the subsequent decomposition of the Mn-oleate complex. This late decomposition leads to narrowing of the size distribution with an inevitable increase in particle size to some degree, similar to that of “size refocusing” observed in the second

Figure 6. TEM images of (a) ultrasmall nickel ferrite nanoparticles and (b) ultrasmall cobalt ferrite nanoparticles prepared by the DSTD process with the XRD pattern of (c) nickel ferrite nanoparticles and (d) cobalt ferrite nanoparticles. 3620

DOI: 10.1021/acsnano.6b07684 ACS Nano 2017, 11, 3614−3631

Article

ACS Nano 25K T

with quite small size distribution (Figure S9), respectively. The XRD patterns of the as-prepared ultrasmall ferrite nanoparticles are shown in Figure 6c and d, which are in good agreement with the standard cubic spinel NiFe2O4 and CoFe2O4 powder diffraction data (JCPDS card nos. 10-0325 and 22-1086). These results indicate that highly monodisperse ultrasmall NiFe2O4 and MnFe2O4 nanoparticles can be achieved by the DSTD process. Moreover, the successful synthesis of other ultrasmall ferrite nanoparticles not only suggests that the DSTD is a relatively robust method for the synthesis of ultrasmall ternary compound nanoparticles but also demonstrates the high versatility of the newly designed Fe-eruciate complex for the synthesis of various ultrasmall magnetic nanoparticles. Magnetic Characterizations of the UMFNPs. The magnetic properties of the UMFNPs were measured by using a superconducting quantum interference device (SQUID) magnetometer. The hysteresis loops of the UMFNPs are shown in Figure 7a. The saturation magnetization (Ms) for 2, 3, and 3.9 nm sized particles is 18.59, 24.91, and 29.18 emu g−1, respectively. The largely reduced magnetic moment of the UMFNPs compared to the bulk counterpart (∼80 emu g−1) can be attributed to the reduction in the volume magnetic anisotropy and the surface spin canting induced by their extremely small particle size.31 The coercivity and remanence of the UMFNPs are negligible at room temperature (Figure 7b), indicating their superparamagnetic behavior. The temperature-dependent magnetization curves after zero-field cooling and field cooling with an applied magnetic field of 20 Oe for these nanoparticles are shown in Figure 7c. The blocking temperatures (TB) of the 2, 3, and 3.9 nm sized particles were measured to be 6, 12.5, and 15 K, respectively (Figure 7c). The effective magnetic anisotropy constant (Keff)

was calculated by the equation Keff = VB B , where kB represents the Boltzmann’s constant and V, the volume of a single nanocrystal.19 As shown in Figure 7d, the Keff is found to increase with the decreasing particle size, similar to early observations for nanoscale ferrimagnetic particles.19,44 Since the significantly increased surface to volume ratio and surface spin disorder, surface anisotropy will play a leading role over volume magnetic anisotropy in the effective magnetic anisotropy in the UMFNPs.31,45,46 The smaller the particle, the higher the Keff and the larger the surface spin disorder. Because the T1 MR contrast effect is the result of protons of water molecules interacting with the electron spins of the contrast agent, the phenomenon that occurs in UMFNPs is believed to significantly promote the T1 MR effect by increasing the surface atoms with unpaired electrons and at the same time suppress the T2 MR effect by decreasing the magnetic moment. Surface Modification and Colloidal Stability of the UMFNPs. The as-synthesized UMFNPs are hydrophobic and hence do not dissolve in aqueous media. As such, it is essential to carry out appropriate surface modification of the UMFNPs to disperse them in an aqueous medium for MRI application. Herein, phosphorylated mPEG is used for surface modification by ligand exchange. As a result of the high surface energy of the UMFNPs, aggregation or dissociation may occur during the ligand exchange process in a harsh chemical environment.47 However, the ligand exchange with phosphorylated mPEG was very successful in our situation. The obtained hydrophilic UMFNPs show very good uniformity and unchanged particle size, as observed in Figure 8d−f. The measured hydrodynamic size was around 7.81 ± 1.72, 8.79 ± 1.06, and 12.50 ± 1.24 nm for 2, 3, and 3.9 nm UMFNPs in deionized (DI) water, respectively.

Figure 7. (a) Hysteresis loop of the UMFNPs with diameters of 2 nm (black ■), 3 nm (red ●), and 3.9 nm (blue ▲) at room temperature. (b) Magnified hysteresis loops of 2 nm (black ■), 3 nm (red ●), and 3.9 nm (blue ▲) sized UMFNPs. (c) Temperature-dependent magnetization curves for the manganese ferrite nanoparticles measured after zero-field cooling (ZFC) and field cooling (FC) with diameters of 2, 3, and 3.9 nm under an applied magnetic field of 20 Oe. For the sake of presentation, the magnetization data were normalized with respect to the value at the maximum of ZFC magnetization, M(TB), for individual samples. (d) Size dependence of the blocking temperature TB and the effective magnetic anisotropy constant Keff of the UMFNPs obtained from the temperature-dependent magnetization curve. 3621

DOI: 10.1021/acsnano.6b07684 ACS Nano 2017, 11, 3614−3631

Article

ACS Nano

Figure 8. Hydrodynamic diameters of the phosphorylated mPEG-modified UMFNPs of (a) 2 nm, (b) 3 nm, and (c) 3.9 nm in aqueous solution. The inset is a photograph of the aqueous UMFNP dispersions. TEM images showing modified UMFNPs of (d) 2 nm, (e) 3 nm, and (f) 3.9 nm. Hydrodynamic size variation of the modified UMFNPs of (g) 2 nm, (h) 3 nm, and (i) 3.9 nm as a function of time upon incubation in water, 10% FBS, normal saline, and 0.2, 0.4, and 1 M NaCl solution.

measurements. Figure 9a shows the qualitative comparison of T1weighted MR images of the different sized UMFNPs with respect to [Fe+Mn] concentration. The T1-weighted MR images of all the samples tend to be brighter with increasing [Fe+Mn] concentration, showing that the UMFNPs as positive contrast agents can effectively reduce the spin−lattice relaxation time of water protons. Furthermore, small-sized UMFNPs exhibited a more significant reduction in T1 relaxation time, as shown by the much brighter images at the designated [Fe+Mn] concentration. In addition, as the release of Mn2+ from Mn-containing nanoparticles in acid conditions has been reported to brighten the images in T1-weighted MRI,48,49 we further measured the MR images of the UMFNPs dispersed in citrate buffer solution with pH 5.0, which simulates the pH microenvironment of the endosomes or lysosomes within the cells.49 As shown in Figure S12, there is no obvious change in brightness after dispersing the samples in citrate buffer for 1 h, suggesting that the T1 enhancement effect in our case was due to the UMFNPs rather than the free Mn2+ released from the particles. In fact, both T1 and T2 signal intensity of the UMFNPs can maintain their original values within 12 days in neutral conditions, but obvious

The insets of the photographs in Figure 8a−c show that there was no observable aggregation in the water-dispersed UMFNPs. Since high colloidal stability of the UMFNPs in various conditions is essentially mandatory for biomedical applications, the colloidal stability of the hydrophilic UMFNPs was further examined by monitoring their hydrodynamic sizes in DI water, normal saline, different concentrations of NaCl solution, and cell culture medium containing 10% FBS. The time-dependent size curves of the UMFNPs were plotted in Figure 8g−i. Preliminary studies have shown that the UMFNPs exhibited excellent colloidal stability in all media after a 12-day incubation at room temperature, as there was no apparent change in size and polydispersity index (PDI). The results also imply that these hydrophilic UMFNPs can maintain the good colloidal stability when they are used as MRI contrast agents under physiological conditions. MR Relaxation Properties of the UMFNPs. In order to assess the MR T1 relaxivity of the UMFNPs, in vitro MRI measurements were performed using a 3 T MRI scanner. The samples were dispersed in deionized water with an [Fe+Mn] concentration from 0.07 to 1 mM determined by ICP-AES 3622

DOI: 10.1021/acsnano.6b07684 ACS Nano 2017, 11, 3614−3631

Article

ACS Nano

proton relaxation entails the direct bonding of water molecules to paramagnetic ions, while the outer-sphere relaxation is linked with the dynamics and diffusion of nearby water molecules beyond the directly bonding ones. T1 relaxivity is therefore the sum of inner-sphere and outer-sphere contributions that can be given by33,34 ⎛1⎞ ⎛1⎞ 1 =⎜ ⎟ +⎜ ⎟ T1 ⎝ T1 ⎠inner sphere ⎝ T1 ⎠outer sphere

(1)

⎛1⎞ 128π 2γI2M n ⎛ ⎞3 2 1 ⎜ ⎟ Ms = ⎜ ⎟ 405ρ ⎝ 1 + L /a ⎠ ⎝ T1 ⎠outer sphere τDJA ( 2ωIτD )

⎛1⎞ qPM = ⎜ ⎟ T1M + τM ⎝ T1 ⎠inner sphere Figure 9. (a) T1-weighted phantom images of the 2, 3, and 3.9 nm sized UMFNPs. (b) Plot of 1/T1 over [Fe+Mn] concentration of the UMFNPs of 2, 3, and 3.9 nm. The slope indicates the specific relaxivity (r1).

Table 1. Comparison of Relaxivity of UMFNPs, 3 nm γ-Fe2O3, 6 nm MnFe2O4, 7 nm MnO, and Clinical Omniscan at 3 T

2 nm UMFNPs 3 nm UMFNPs 3.9 nm UMFNPs 3 nm γ-Fe2O3 6 nm MnFe2O4 7 nm MnO Omniscan

Ms at 300 K (emu g−1)

r1 (mM−1 s−1)

r2 (mM−1 s−1)

r2/r1

18.59 24.91 29.18 31 50

8.43 8.23 6.98 4.96 1.39 1.57 4.04

21.02 21.97 27.08 28.59 64.08 7.82 4.19

2.49 2.67 3.88 5.76 46.10 4.98 1.04

(3)

where γI is the proton gyromagnetic ratio and Mn and ρ are the molar mass and density of the UMFNPs. Ms and a are the saturation magnetization and the radius of the UMFNPs, respectively. D is the diffusion coefficient of water molecules, L is the thickness of an impermeable surface coating, τD is the translational diffusion time (τD = r2/D), r is the effective radius of the particles (r = a + L), and JA represents Ayant’s s pectral density function, which is related to Z = 2ωIτD . PM is the mole fraction of surface metal ions, q is the number of water molecules bound per metal ion (i.e., the hydration number), and τM is the residence lifetime of the bound water. The “m” subscript refers to the shift or relaxation rate of the solvent molecule in the inner-sphere. T1M is the relaxation time of the bound water. In terms of the boundary condition associated with a high magnetic field, the outer-sphere mechanism is usually assumed to dominate proton relaxation for superparamagnetic nanoparticle based contrast agent systems, while the contribution from the inner-sphere mechanism is omitted. In this case, r1 is proportional to the square of Ms and τDJA ( 2ωIτD ) according to eq 2, where τDJA ( 2ωIτD ) monotonically increases when the particle size is less than 4 nm (Table S4). Therefore, the UMFNPs with larger particle size and higher Ms should have shown a larger r1 theoretically. But quite to the contrary, as shown in Table 1, the larger sized UMFNPs gave rise to smaller r1 values, which strongly suggests that the outer-sphere mechanism alone cannot explain our experimental observations because some important contributions are missing. More evidence can be found in the r1 comparison of the same sized UMFNPs and γ-Fe2O3. Both of them have the same τDJA ( 2ωIτD ) and the Ms of the γ-Fe2O3 nanoparticles is even slight higher than that of the UMFNPs; however, the UMFNPs exhibit a higher r1 value than γ-Fe2O3 nanoparticles. In fact, the UMFNPs can be treated as a core/shell structure comprising a ferromagnetic core and paramagnetic shell (spin disorder layer), where the thickness of the spin disorder layer for manganese iron nanoparticles is about 0.95 nm.50 With the dramatically decreasing size, the paramagnetic to ferromagnetic composition ratio in the UMFNPs is significantly increased.31 Therefore, they can be conceived to some extent as “pseudo-macromolecules”, in which the inner-sphere contribution cannot be ignored.

changes will be observed after 12 h in acidic conditions probably due to the degradation of UMFNPs (Figures S15 and S16). The r1 relaxivity of the UMFNPs was extracted from the measured T1 data and plotted in Figure 9b. The measured r1 relaxivities were 8.43, 8.23, and 6.98 mM−1 s−1 for the 2, 3, and 3.9 nm UMFNPs, respectively. The MR relaxivity comparison of the UMFNPs, 3 nm γ-Fe2O3 nanoparticles, 6 nm MnFe2O4 nanoparticles, 7 nm MnO nanoparticles, and clinical contrast agent Omniscan is presented in Table 1. The highest r1 values of the UMFNPs are

sample

(2)

double those of 3 nm γ-Fe2O3 and commercial Omniscan, 5.4 times higher than that of 7 nm MnO nanoparticles, and 6 times higher than that of the large-sized MnFe2O4 (6 nm), for which the T2 MR effect is dominant. Notably, this r1 value is the highest measured among the ferrite nanoparticles with similar size reported so far, where the r2/r1 ratio is the smallest one as shown in Table S3. As compared to γ-Fe2O3 nanoparticles with the same size and surface modification, it can be easily deduced that the large r1 value of the UMFNPs is attributed to the doping of Mn2+ ions. Nevertheless, the mechanism of T1 enhancement by molecular engineering of ultrasmall ferrite nanoparticles remains unclear. According to theoretical considerations, the relaxation enhancement of a contrast agent is known generally to follow the inner-sphere and outer-sphere mechanisms.34 The inner-sphere 3623

DOI: 10.1021/acsnano.6b07684 ACS Nano 2017, 11, 3614−3631

Article

ACS Nano Along this line of thought, the enhanced r1 with the decreasing size in the UMFNPs could be due to the increased inner-sphere contributions originating from the increased amount of paramagnetic Fe3+ and Mn2+ ions on the particle surface. The high r1 value of the UMFNPs compared to γ-Fe2O3 nanoparticles can also be explained in terms of the inner-sphere mechanism. As described by eq 3, the r1 value is determined by qPM, T1M, and τM. The qPM is comparable for the UMFNPs and γ-Fe2O3 because of their same size. The T1M (Supporting Information part 5) is also the same as that of the Mn2+ and Fe3+, which have equal total electron spins.51 However, the τM for Mn(H2O)62+ was 4.3 × 104 ns, which is 2 orders of magnitude smaller than that of Fe(H2O)63+ (5.3 × 106 ns),34 suggesting that UMFNPs rather than γ-Fe2O3 will have a superior r1 relaxivity. Clearly, the artificial Mn2+ doping in ultrasmall iron oxide nanoparticles could effectively expedite spin−lattice relaxation and increase innersphere contribution by altering the residence lifetime of the bound water (τM). This is distinctly different from the early doping engineering of magnetic nanoparticles for ultrasensitive T2 MR contrast agents, where high Ms and large r2 relaxivity are pursued.2,7 It is to be noted that the outer-sphere mechanism may still be the dominant contribution to transverse spin−echo relaxation of the UMFNPs, not only because the r2 value in the UMFNPs increases with the increasing particle size and Ms but also because their r2/r1 ratio monotonically increases against τD, which are well in agreement with the prediction based on the outer-sphere mechanism. Our results indicate that both outersphere and inner-sphere proton relaxation contributions are responsible for the experimental relaxation enhancement in the UMFNP system. More significantly, the results also reveal that the strategy of metallic doping in ultrasmall iron oxide nanoparticles could modulate r1 and r2 relaxivity independently on some level by adjusting outer-sphere and inner-sphere contributions separately, which provides the possibility to develop the innovative nanoparticle contrast agent with desirable r1 and r2 relaxivity for specific MRI diagnostics. In Vivo Contrast-Enhanced MRI. In vivo MRI was carried out for Sprague−Dawley (SD) rats in the clinical 3.0 T Siemens Trio MRI scanner using the 3d-flash and a turbo spin echo (TSE) sequence for imaging the blood vessels and liver, respectively. Due to the good colloid stability and the high r1 relaxitivity, the 3 nm UMFNPs were employed as a model system for contrast-enhanced

MRI for angiography and liver imaging with a dosage of 2.5 mg [Fe+Mn]/kg body weight (0.045 mmol [Fe+Mn]/kg body weight). The clinical MR angiography contrast agent Omniscan (gadodiamide) and liver-specific contrast agent Primovist (Gd-EOB-DTPA) were set as controls for imaging blood vessels and the liver with dosages of 0.2 and 0.025 mmol [Gd]/kg body weight, respectively. All the contrast agents were intravenously administrated. Figure 10 shows the T1-weighted MR images acquired at preinjection and postinjection (20 min) of the UMFNPs and Omniscan. Quite notably, the blood vessels and liver were brightened in the T1-weighted MR images after the injection of the UMFNPs, demonstrating that the UMFNPs can effectively shorten the spin−lattice relaxation time of water protons in the circulating system. The prominent vascular details including the external jugular vein, axillary vein, internal jugular vein, subclavian vein, superior vena cava, and aortic arch can be distinctly distinguished, which indicates that the resolution of the image is very high, such that even the 0.47 mm veins were able to be accurately imaged. In contrast, for Omniscan-enhanced images with a dosage 4.4-fold higher than the UMFNPs, the essential vascular details were ambiguous. Figure 11a shows UMFNPsenhanced MR images acquired at 0, 0.5, 1, 5, 10, 20, 30, 60, 120, and 180 min, respectively. The intense signal of the blood vessels can be maintained for 3 h on contrast-enhanced magnetic resonance angiography, while the Omniscan-enhanced images showed weak contrast in vascular nets and the vascular nets can be depicted only within 20 min (Figure S17). Such long-term blood pool imaging achieved by UMFNPs is ascribed to their optimal particle size, which is neither very large to prevent the uptake by the reticuloendothelial system52 nor so small as to be rapidly excreted through the kidney.53 Moreover, because the [Fe+Mn] content of the aorta is almost the same as that of the blank aorta (Figure S19), there is no noticeable uptake by endothelium cells in the blood vessels. As such, it is rather the relatively long-term blood circulation of the UMFNPs than endothelium cell uptake that is responsible for the increased signal duration. The contrast-enhancement ratios of MR signal intensity ΔSI (ΔSI = |SIpost − SIpre|/SIpre) for the UMFNPs in the region of interest of the subclavian vein were extracted and compared with that of Omniscan. As shown in Figure 11b, the ΔSIUMFNPs was always much larger than ΔSIomniscan. At 30 s, the

Figure 10. In vivo UMFNP-enhanced MR images obtained using the 3d-FLASH sequence. 3624

DOI: 10.1021/acsnano.6b07684 ACS Nano 2017, 11, 3614−3631

Article

ACS Nano

Figure 11. (a) Representative UMFNP-enhanced MR angiography images using the 3d-flash MR sequence acquired at 0, 0.5, 1, 5, 10, 20, 30, 60, 120, and 180 min after intravenous administration of 3 nm UMFNPs. (b) Enhancement ratio of the MR signal ΔSI of the subclavian vein after being contrast-enhanced by UMFNPs and Omniscan.

ΔSIUMFNPs reached its maximum of 426%, which was 2.6-fold higher than ΔSIomniscan, before decreasing to 178% at 180 min, while ΔSIomniscan was dramatically down to zero. The high contrast-enhancement ratio for the UMFNPs indicates that the UMFNPs have potential to be a superior T1 contrast agent for MR angiography, which is clinically important to detect myocardial infarction, kidney failure, atherosclerotic plaques, thrombosis, and tumoral angiogenesis.54−60 In addition to high-resolution MR angiography, the high contrast-enhancement in the liver by the UMFNPs has also attracted much interest. To further investigate its ability for liverspecific MRI, UMFNP-enhanced MR images of the liver were acquired using the TSE sequence, where the clinical Primovist was used as the control. Figure 12a shows the transverse-plane MR images of liver acquired at 0, 10, 20, 30, 45, 60, 120, and 180 min

after the injection of UMFNPs. The significantly brightened MR image suggests the effective accumulation of UMFNPs in the liver. The contrast-enhancement ratio ΔSI of liver images by the UMFNPs and Primovist are shown in Figure 12b. The highest ΔSIUMFNPs of 70.3% appeared at 10 min, which was almost 3 times higher than that of 3.3 nm Fe3O4.61 Over the entire imaging time window, the ΔSIUMFNPs was always higher than ΔSIPrimovist. Since the Mn2+ ion can be selectively taken up by hepatocytes via the vitamin B6 transporter,62 it is reasonable to assume that the UMFNP is a hepatocyte-specific contrast agent. This assumption is further confirmed by the in vivo UMFNP distribution in liver tissues, where many UMFNPs were found in the hepatocytes (Figure S20). In contrast, Resovist is localized mostly in Kupffer cells. Although the exact mechanism still needs further investigation, we believe that the presence of a large number of Mn2+ ions on the surface of the UMFNPs plays an important role in hepatocyte-specific uptake. All results obtained from the in vivo MRI verify that the UMFNPs are a highly sensitive and multifunctional T1 MR contrast agent for highresolution imaging of the blood pool and liver. Pharmacokinetics, Biodistribution, and Excretion of the UMFNPs. To investigate the in vivo behavior of the UMFNPs, the pharmacokinetics, biodistribution, and excretion of the 3 nm UMFNPs were then carried out with the same dosage (2.5 mg [Fe+Mn]/kg body) used for contrast-enhanced MRI. Inductively coupled plasma−mass spectrometry (ICP-MS) was used to quantify the related metal elements in the rats’ blood and tissue at designed time points after a single tail vein injection. Figure 13a shows the normalized [Fe+Mn] concentration−time curve in blood for the UMFNPs after a single intravenous (i.v.) injection. This curve shows a single-exponential disposition; therefore, the pharmacokinetic parameter was determined with a one-compartment pharmacokinetic model63 and the blood elimination half-life (t1/2) is about 0.31 h. Figure 13b shows the in vivo biodistribution of the UMFNPs at 24 h after i.v. injection. The results revealed that the UMFNPs mainly accumulated in the liver, spleen, and kidney. A similar biodistribution has also been found in 2 and 3.9 nm UMFNPs (Figure S21). The [Fe+Mn] concentration in both feces and urine of SD rats was also measured to examine the excretion pathway of the UMFNPs. Quantitative analysis showed that more than 27.04% of the UMFNPs was excreted from the body via the renal clearance pathway within 24 h and up to 30.33%

Figure 12. (a) T1-weighted MR images of transverse planes of the liver using the TSE sequence acquired at 0, 10, 20, 30, 45, 60, 120, and 180 min after intravenous administration of 3 nm UMFNPs. (b) Enhancement ratio of MR signal ΔSI of the liver after contrastenhanced by UMFNPs and Primovist. 3625

DOI: 10.1021/acsnano.6b07684 ACS Nano 2017, 11, 3614−3631

Article

ACS Nano

Figure 13. Pharmacokinetics, biodistribution, and excretion of 3 nm UMFNPs in rats. (a) Time−[Fe+Mn] concentration curves of blood and (b) biodistribution pattern at 24 h after i.v. injection of the UMFNPs. Percentage of (c) renal and (d) hepatobiliary excreted [Fe+Mn] versus time after i.v. injection of the UMFNPs. Data are reported as mean ± standard deviation (n = 3).

at a high Fe concentration (>50 μg/mL). At a low Fe concentration, there was no obvious change in the viability of the HepG2 cells in comparison with that of Chang liver cells and endothelium cells. We further investigated the cell localization of the UMFNPs, which is critical for their cytotoxicity. The uptake of the UMFNPs by Chang liver cells was visualized by TEM and is shown in Figure 15. According to the TEM observations, the small UMFNPs (the 2 and 3 nm samples) were observed to localize in the cytosol, cytoplasmic vesicles, and the nucleus (Figure 15a−d), while the large UMFNPs (the 3.9 nm samples) were mainly distributed in cytoplasmic vesicles and the cytosol (Figure 15e and f). It is not surprising that such small nanoparticles can enter the nucleus, as their hydrodynamic sizes (7.81 and 8.79 nm) are smaller than the nuclear pore size exclusion limit of ∼9 nm.66 A similar phenomenon has been observed for ultrasmall Au nanoparticles.67,68 We speculate that the increased toxicity of the small UMFNPs may be associated with their nuclear internalization; hence, their cytotoxicity may be largely reduced by loading them onto a relatively large and biocompatible carrier for practical applications in the future.

within 60 h (Figure 13c), indicating that the UMFNPs could be efficiently renally cleared from the body. About 22.42% of UMFNPs were excreted via the hepatobiliary pathway within 24 h and 67.40% of the UMFNPs were excreted within 60 h (Figure 13d). This efficient renal clearance of the UMFNPs is ascribed to their small hydrodynamic size (8.79 nm, slightly larger than the filtration-size threshold of glomerular capillary walls) and neutral hydrophilic mPEG coating.63,64 The rapid excretion of the UMFNPs within 60 h is of value to minimize systematic toxicity. In addition, the histological assessment of the excised heart, liver, lung, kidney, spleen, and brain was also performed to evaluate the acute systemic toxicity and the subacute systemic toxicity after intravenous exposure to the UMFNPs (Figures S22 and S23). According to the standard of biological evaluation of medical device testing for systemic toxicity (ISO 10993-11:2006), the acute systemic toxicity and the subacute systemic toxicity were tested at 24 h and 14 d, respectively. As shown in Figures S22 and S23, there were no obvious pathological changes in these organs, indicating the safe use of the UMFNPs for potential clinical trials. In Vitro Cytotoxicity Assay. Since cytotoxicity is an important issue for the clinical application of nanoparticlebased contrast agents, an in vitro cell cytotoxicity assay was thus carried out to assess the toxicity profiles of the UMFNPs using standard cytotoxicity tests. Figure 14 shows the viability of Chang liver cells, HepG2 liver cancer cells, and endothelium cells after 24 h of incubation with the UMFNPs at 37 °C. All UMFNP samples show insignificant toxicity with a cell viability of more than 80% at an Fe concentration of less than 100 μg/mL for Chang liver cells and endothelium cells. According to the relative growth rate and toxicity grade conversion,65 the toxicity of the UMFNPs is classified as grade 1, which is safe for the Chang liver cells and endothelium cells. For HepG2 cells, a slight decrease in cell viability was observed for smaller UMFNPs (2 and 3 nm samples)

CONCLUSIONS In summary, we have developed a general DSTD strategy for synthesis of ultrasmall metal ferrite nanoparticles with a wellcontrolled size of