Synthesis, Characterizations, and 9.4 Tesla T2 MR Images of Polyacrylic Acid-Coated Terbium(III) and Holmium(III) Oxide Nanoparticles

Polyacrylic acid (PAA)-coated lanthanide oxide (Ln2O3) nanoparticles (NPs) (Ln = Tb and Ho) with high colloidal stability and good biocompatibility were synthesized, characterized, and investigated as a new class of negative (T2) magnetic resonance imaging (MRI) contrast agents at high MR fields. Their r2 values were appreciable at a 3.0 T MR field and higher at a 9.4 T MR field, whereas their r1 values were negligible at all MR fields, indicating their exclusive induction of T2 relaxations with negligible induction of T1 relaxations. Their effectiveness as T2 MRI contrast agents at high MR fields was confirmed from strong negative contrast enhancements in in vivo T2 MR images at a 9.4 T MR field after intravenous administration into mice tails.


Introduction
Biomedical imaging is a rapidly growing field in nanomedicine [1][2][3]. Among various imaging techniques, magnetic resonance imaging (MRI), which detects proton spin relaxation signals is highly sensitive and its image spatial resolution is very high because of ample protons in living objects [1,2]. T 2 MRI detects transverse proton spin relaxation signals and the image contrast becomes darker in the presence of contrast agents because transverse proton spin relaxations are accelerated by the contrast agents. The excellent physicochemical properties of nanoparticles (NPs) make them suitable for use as advanced imaging agents in diagnosing diseases [4][5][6][7][8][9][10]. Among the NPs, lanthanide oxide (Ln 2 O 3 ) NPs can play an important role in biomedical imaging because of their exceptionally large paramagnetic moments at room temperature, making them suitable for use as MRI contrast agents even at ultrasmall particle sizes [8][9][10]. Such Ln 2 O 3 NPs include those with Ln as gadolinium (Gd) ( 8 S 7/2 ), terbium (Tb) ( 7 F 6 ), dysprosium (Dy) ( 6 H 15/2 ), holmium (Ho) ( 5 I 8 ), and erbium (Er) ( 4 I 15/2 ), which all possess high atomic magnetic moments [11].

One-Pot Polyol Synthesis of PAA-Coated Ln 2 O 3 NPs (Ln = Tb and Ho)
To synthesize the PAA-coated Ln 2 O 3 NPs (Figure 1), 2 mmol of Ln 3+ -precursor salt and 2.0-2.5 mmol of PAA were added to 20 mL of TEG in a three-necked round-bottom flask. The mixture solution was magnetically stirred at room temperature under atmospheric conditions until the precursor salt dissolved in the TEG. In a separate beaker, a NaOH solution consisting of 10 mmol of NaOH in 15 mL of TEG was prepared. The NaOH solution was slowly added to the precursor solution until the pH of the mixture solution reached 10. At high pH values (pH > 8.0), more stable Ln(OH) 3 is mainly formed rather than the Ln 3+ -PAA complex [38]. The Ln(OH) 3 is then further converted into Ln 2 O 3 NPs at elevated temperatures [39]. After the pH became nearly constant, the reaction temperature was slowly increased to 110 • C using a heating oil bath and maintained at that temperature for 15 h with constant magnetic stirring. The reaction mixture was then cooled to room temperature and transferred to a 500-mL beaker. The obtained PAA-coated NPs were washed with 400 mL of ethanol three times to remove Na + , OH − , TEG, PAA, and the unreacted precursors. The product solution was diluted with 400 mL of ethanol and then stored in a refrigerator (~4 • C) for several days until the product NPs precipitated. The upper transparent portion of the solution was decanted and the remaining product solution was washed with ethanol again. To remove ethanol from the product NPs, 400 mL of triple-distilled water was added to the product solution, which was then concentrated to a volume of 20-30 mL using a rotary evaporator that removed the ethanol. This washing process was repeated three times. The production yield was 60-70%. This good yield and low production cost suggest that the present method is economically effective. The product solution was split into two parts: one half volume was subject to a powder sample by freeze-drying it under vacuum, and the other half volume was diluted with triple-distilled water to prepare a suspension sample.

General Characterizations
The particle diameter of the PAA-coated Ln 2 O 3 NPs (Ln = Tb and Ho) was measured using a high-resolution transmission electron microscope (HRTEM) (Titan G2 ChemiSTEM CS Probe, FEI, Hillsboro, OR, USA) operated at 200 kV acceleration voltage. Each sample dispersed in ethanol was dropped onto a carbon film supported by a 200-mesh copper grid (Pelco No. 160, Ted Pella Inc., Redding, CA, USA) using a micropipette (Eppendorf, 2-20 µL). The copper grid with the sample was allowed to dry in air for an hour at room temperature before mounting inside the HRTEM for measurement. The Ln-concentration in each suspension sample was measured using an inductively coupled plasma atomic emission spectrometer (ICPAES) (IRIS/AP, Thermo Jarrell Ash Co., Waltham, MA, USA). Each sample was pretreated with HCl/HNO 3 (1:3 mole ratio) acids to completely dissolve NPs in solution before measurement. The hydrodynamic diameter was measured using a dynamic light scattering (DLS) particle size analyzer (Zetasizer Nano ZS, Malvern, Malvern, UK). For measurement, the NP suspension samples with 0.01 mM Ln were used. The zeta potentials (Zetasizer Nano ZS, Malvern, Malvern, UK) were measured using the NP suspension samples with 0.01 mM Ln. The colloidal stability was also investigated by measuring the transmission (T) and backscattering (BST) of thee near infrared (NIR) beam (880 nm) as a function of h (h = 5 to 10 mm from the vial bottom containing the sample solution) and t for t = 0 to three days using a Turbiscan (Turbiscan AGS, Formulaction, Paule Raymondis, France). The crystal structure of the powder samples before and after thermogravimetric analysis (TGA) was measured using a multi-purpose x-ray diffraction (XRD) spectrometer (X'PERT PRO MRD, Philips, The Netherlands) with unfiltered CuKα radiation (λ = 1.54184 Å) and 2θ = 15-100 • . A Fourier transform-infrared (FT-IR) absorption spectrometer (Galaxy 7020A, Mattson Instrument Inc., Madison, WI, USA) was used to investigate the surface coating of the NPs. For measurement, pellets of powder samples in KBr were prepared. The amount of PAA-coating on the NP surfaces was estimated by recording TGA curves (SDT-Q600, TA Instrument, New Castle, DE, USA) using powder samples between room temperature and 900 • C while air flowed over the powder sample. The amount of PAA-coating was quantified in wt.% from the mass drop in the TGA curve after taking into account the initial mass drop between room temperature and~105 • C due to water desorption. A vibrating sample magnetometer (VSM) (7407-S, Lake Shore Cryotronics Inc., Westerville, OH, USA) was used to characterize the magnetic properties of the powder samples by recording the magnetization (M) versus applied field (H) (or M−H) curves (−1.8 T ≤ H ≤ 1.8 T) at 300 K. The measurements were carried out using powder samples of 20-30 mg. The net M value of each sample (i.e., only the Ln 2 O 3 NPs without the PAA coating) was estimated using the net mass of the Ln 2 O 3 NPs extracted from the TGA curve.

In Vitro Cellular Cytotoxicity Measurements
The in vitro cellular toxicity of the PAA-coated Ln 2 O 3 NPs (Ln = Tb and Ho) was assessed using a CellTiter-Glo Luminescent Cell Viability Assay (Promega, Madison, WI, USA). In this assay, a luminometer (Synergy HT, BioTek, Winooski, VT, USA) was used to quantify intracellular adenosine triphosphate. Human prostate cancer (DU145) and normal mouse hepatocyte (NCTC1469) cell lines were used as test cells. These two cell lines were used because it would be better to use one cancer and one normal cell line rather than to use two cancer or two normal cell lines. In addition, they are common cell lines. Both cells were seeded on a 24-well cell culture plate and incubated for 24 h (5 × 10 4 cell density, 500 µL of cells per well, 5% CO 2 , and 37 • C). Five test NP suspension samples were prepared by diluting each concentrated NP suspension sample with a sterile phosphate-buffered saline solution. Then, 2 µL of each test suspension sample was added to the cells to have 10, 50, 100, 200, and 500 µM Ln in the treated cells, and the treated cells were incubated for 48 h. The viabilities of the incubated cells were measured and normalized with respect to the control cells with 0.0 M Ln. The measurements were repeated in triplicate to obtain the average cell viabilities.

Relaxometric Property Measurements
To obtain longitudinal (r 1 ) and transverse (r 2 ) water proton spin relaxivities, longitudinal (T 1 ) and transverse (T 2 ) water proton spin relaxation times were measured using a 3.0 T MRI scanner (MAGNETOM Trio Tim, Siemens, Munchen, Germany) and a 9.4 T MRI scanner (9.4 T MRI System, Varian, Palo Alto, CA, USA). A series of NP suspension samples (1, 0.5, 0.25, 0.125, 0.0625, and 0.0 mM Ln) (Ln = Tb and Ho) were prepared by diluting the original concentrated NP suspension samples with triple-distilled water. T 1 relaxation times were measured using an inversion recovery method. T 2 relaxation times were obtained using a multiple spin-echo method with a Carr-Purcell-Meiboom-Gill pulse sequence. Then, the r 1 and r 2 values were estimated from the slopes in the plots of 1/T 1 and 1/T 2 versus Ln-concentration, respectively.

In Vivo T 2 MR Image Measurements
In vivo animal imaging studies were performed in accordance with the rules and regulation and permission of the animal research committee of the Korea Institute of Radiological and Medical Science. A 3.0 T MRI scanner (MAGNETOM Trio Tim, Siemens, Munchen, Germany) and a 9.4 T MRI scanner (9.4 T MRI System, Varian, Palo Alto, CA, USA) were used to obtain the in vivo T 2 MR images. Two or three Institute of Cancer Research (ICR) mice weighing~30 g were used for each sample. The mice were anesthetized using 1.5% isoflurane in oxygen. Measurements were made before and after the injection of a NP suspension sample prepared in triple-distilled water into mice tail veins. The injection dose was approximately 0.1 mmol Ln/kg (Ln = Tb and Ho) (injection volume = 70 µL). During measurements, body temperature of the mice was maintained at 37 • C using a warm water blanket. After measurements, the mice were revived from anesthesia and placed in a cage with free access to food and water. The spin echo sequence was used to obtain T 2 MR images. The typical measurement parameters were as follows: H = 3.0 (or 9.4) T, echo time (TE) = 30 ms, repetition time (TR) = 2000 ms, frequency encoding step = 320 Hz, phase = 320, number of acquisitions (NEX) = 4, field of view (FOV) = 70 mm, slice thickness = 1.0 mm, and slice gap = 1.1 mm.

Particle Diameter (d), Hydrodynamic Diameter (a), Zeta Potential (ξ), and Crystal Structure
The estimated Ln-concentrations of the solution samples using an ICPAES were 28.42 mM Tb for the PAA-coated Tb 2 O 3 NPs and 30.44 mM Ho for the PAA-coated Ho 2 O 3 NPs. PAA has abundant COOH groups ( Figure 1) and thus can be tightly grafted on a Ln 2 O 3 NP surface through many electrostatic bonds between its COOH groups and the many Ln 3+ on the NP surface. Consequently, the NP suspension samples showed good colloidal stability, as shown in the NP suspension samples in Figure 2a (i.e., the NPs had not precipitated since they were prepared (>one year)). The colloidal stability was additionally checked in a 10% FBS in RPMI 1640 medium and a sodium acetate buffer solution (pH = 7.0). The NPs were stable in these media (i.e., no precipitation for 20 days) (Figure 2a). The observed good colloidal stability in 10% FBS in RPMI1640 medium was consistent with the fact that PAA negligibly interacts with proteins such as albumin at physiological pH [40]. Each vial containing the NP suspension sample exhibited laser light scattering (i.e., Tyndall effect) due to the collision of the NP suspension with laser light whereas no light scattering was observed for the vial containing the triple-distilled water (Figure 2b), proving the good colloidal dispersion of the NPs in aqueous solutions. HRTEM imaging was used to estimate the particle diameters of the PAA-coated Ln 2 O 3 NPs (Ln = Tb and Ho) (Figure 2c). The core Ln 2 O 3 NPs could be clearly seen through lattice fringes in HRTEM images (see images at 2 nm scale), but the shell PAA coating around the NP core could not be seen because it is almost impossible to see small polymers like PAA through HRTEM imaging. The remaining is the background amorphous carbon covering the copper grid onto which the NPs were dispersed. As can be seen in the HRTEM images, the NPs were ultrasmall and nearly spherical. Their average particle diameters (d avg ) were estimated to be 1.8 ± 0.1 nm for the PAA-coated Tb 2 O 3 NPs and 1.7 ± 0.1 nm for the PAA-coated Ho 2 O 3 NPs from log-normal function fits to the observed particle diameter distributions (Figure 2d and Table 1). As can be seen in Figure 2d, most of the NPs were less than 3 nm in diameter, making them amenable to renal excretion [17][18][19], while some of them were larger than 3 nm, but less than 5.5 nm. In addition, the PAA-coated Tb 2 O 3 NPs were slightly bigger than the PAA-coated Ho 2 O 3 NPs. DLS was used to estimate the hydrodynamic diameters of the PAA-coated Ln 2 O 3 NPs (Figure 2e). Applying log-normal function fits to the observed DLS patterns, the average hydrodynamic diameters (a avg ) were estimated to be 13.5 nm for the PAA-coated Tb 2 O 3 NPs and 12.7 nm for the PAA-coated Ho 2 O 3 NPs (Figure 2e and Table 1). For both samples, the difference between a avg and d avg was approximately 11 nm, indicating large hydration spheres for both samples and thus explaining the observed good colloidal stability for both samples. The zeta potentials (ξ) were measured overtime (Figure 2f). The observed highly negative and constant ξ values over time (inset in Figure 2f) supported the excellent colloidal stability of the NP solution samples. The average ξ values are provided in Table 1. Here, the negative surface charges of NPs can be balanced by Na + in solution because as described in the experimental section, the PAA-coated Ln 2 O 3 NPs were produced in basic solutions (pH =~10) using NaOH. The colloidal stability of the PAA-coated Ln 2 O 3 NPs was also confirmed through the transmission (T) and backscattering (BST) measurements of NIR beam as a function of time (t) for three days. The ∆T (t) and ∆BST (t), corresponding to average T (t) and average BST (t) minus average T (t = 0) and average BST(t = 0), respectively, in which the average T (t) and average BST (t) are the averages of transmitted and backscattered NIR beam intensities, respectively, for all NIR beam heights (h) (h was changed between 5 and 10 mm from the vial bottom containing the sample solution) at a scan time t, were plotted as a function of t (Figure 2g,h, respectively). Note that both the ∆T (t) and ∆BST (t) are zero for stable colloids. Therefore, the negligible deviations of ∆T (t) and ∆BST (t) from zero for both solution samples confirmed the stable colloidal suspensions for both solution samples. Therefore, all results confirmed the excellent colloidal stability for both solution samples. The leaching experiments of Tb 3+ and Ho 3+ ions from the samples were carried out by dispersing 4.5 mg of powder samples in 4.0 mL of triple-distilled water and dialyzing them against 1.5 L of triple-distilled water (MWCO = 1000 amu) for two days. After dialysis, the solutions outside the dialysis bag were concentrated to~1.8 mL using a rotary evaporator. The measured metal ion concentration was below the detection limit of the ICPAES for both samples, indicating negligible leaching of metal ions from the samples.
The crystal structures of both powder samples before and after TGA were determined by XRD analysis. Both samples showed very broad and amorphous XRD patterns before TGA (bottom XRD patterns in Figure 3a,b), indicating that most of the NPs were not fully crystallized due to their ultrasmall size [39]. After TGA, however, sharp peaks corresponding to cubic structure of bulk materials were observed for both samples due to particle size growth and crystallization (top XRD patterns in Figure 3a,b). All peaks after TGA could be assigned with (hkl) Miller indices but only intense peaks were representatively assigned in the XRD patterns. The estimated cell constants of 5.28 Å for the Tb 2 O 3 NPs and 10.61 Å for the Ho 2 O 3 NPs after TGA were consistent with previously reported values [41]. Table 1. Particle diameter (d), hydrodynamic diameter (a), zeta potential (ξ), surface-coating amount (P, σ, N PAA ), magnetic properties, and water proton spin relaxivities (r 1 , r 2 ) of the PAA-coated Ln 2 O 3 NPs (Ln = Tb and Ho).

Surface-Coating Results
The surface coating was examined by FTIR absorption spectroscopy using powder samples. Characteristic absorption peaks of PAA [42] such as the C-H stretch at 2936 cm −1 and the C=O stretch at 1697 cm −1 were observed in the FTIR absorption spectra of the samples (Figure 4a). However, the C=O stretches for PAA in the samples were red-shifted by~154 cm −1 from that of free PAA. This is due to electrostatic bonding of the COO − groups in PAA to Ln 3+ (Ln = Tb and Ho) on the NP surfaces, which has been observed previously in various molecules with carboxylic groups bonded to metal oxides [43,44]. The two peaks in the sample spectra at 1555-1535 cm −1 and 1395-1400 cm −1 with a frequency difference of~147 cm −1 were the asymmetric and symmetric stretching frequencies of COO − , respectively, due to the bridge-bonding of the COO − of PAA with Ln 3+ on a NP surface [45], as schematically drawn in Figure 4b. Because each PAA (M w = ∼1800 Da) has ∼25 carboxylic groups, it is expected that many of them per PAA were conjugated to each NP. Because NaOH was used in the synthesis, free (or unconjugated) carboxylic groups of PAA in the PAA-coated NPs were in the form of COO − Na + (not COOH), which may appear at 1555 cm −1 [46], slightly higher than that (= ∼1545 cm −1 ) of conjugated COO − to Ln 3+ of the NP. Therefore, it was difficult to see the peaks of free carboxylic groups (i.e., COO − Na + ) of PAA in the PAA-coated NPs due to their overlap with the peaks of COO − conjugated to Ln 3+ of the NP. The average amount (P) of PAA-coating on the NP surface was estimated in wt.% from the mass drop in the TGA curves (Figure 4c). The initial mass drop between room temperature and~105 • C was attributed to water desorption. The following mass drop (i.e., P) was due to the combustion of PAA due to its oxidation reaction with flowing hot air. The remaining mass was due to Ln 2 O 3 NPs (Ln = Tb and Ho). The average number of PAA polymers coating a unit NP surface area (i.e., grafting density (σ) [47]) was calculated using the molecular mass of PAA (=~1800 Da), the d avg estimated from HRTEM imaging, and the bulk density (7.91 g/cm 3 for Tb 2 O 3 and 8.41 g/cm 3 for Ho 2 O 3 [48]). The average number (N PAA ) of PAA polymers coating the NP surface was estimated by multiplying σ by the NP surface area (πd avg 2 ). The surface-coating results are summarized in Table 1.

In Vitro Cellular Cytotoxicity Results
The in vitro cell viabilities of both DU145 and NCTC1469 cells treated with NP suspension samples were greater than 90% up to a 500 µM Ln-concentration for both samples (PAA-coated Tb 2 O 3 NPs in Figure 5a and PAA-coated Ho 2 O 3 NPs in Figure 5b), showing very low cellular toxicities for both samples. The mice also survived after in vivo MRI experiments (>1 year), proving good biocompatibility of both samples. The optical images of DU145 cancer cells incubated with PAA-coated Tb 2 O 3 NPs at various Tb-concentrations (Figure 5c) showed that the NPs seemed to be more concentrated at the cells and cell nuclei with increasing Tb-concentration.

Magnetic Properties
The magnetic properties of the powder samples were characterized by recording the M versus H (or M−H) curves (−1.8 T ≤ H ≤ 1.8 T) at 300 K ( Figure 6). Both samples were paramagnetic (i.e., no hysteresis, zero coercivity, and zero remanence in the M-H curves), similar to their corresponding bulk materials [26][27][28]. The measured M value of each sample was mass-corrected using the net mass of the NPs [i.e., the mass of the Ln 2 O 3 (Ln = Tb and Ho) only, without PAA] estimated from the TGA curves. From the mass-corrected M−H curves, the net M values at 1.8 T were estimated to be 3.8 emu/g for the Tb 2 O 3 NPs and 4.1 emu/g for the Ho 2 O 3 NPs (Table 1). The slightly higher M value (in emu/g) of the Ho 2 O 3 NPs than that of the Tb 2 O 3 NPs is due to the higher atomic magnetic moment (µ = 10.60 µ B ) of Ho than that (= 9.72 µ B ) of Tb [11], in which µ B is the Bohr magneton.

r 1 and r 2 Values
The r 1 and r 2 water proton spin relaxivities were estimated from the slopes in the plots of inverse T 1 and T 2 water proton spin relaxation times (i.e., 1/T 1 and 1/T 2 ) versus Lnconcentration (Ln = Tb and Ho), respectively, at H = 3.0 T (Figure 7a) and 9.4 T (Figure 7b). Both samples showed negligible r 1 values at all H, but their r 2 values were appreciable at H = 3.0 T and higher at H = 9.4 T ( Table 1).

In Vivo T 2 MR Images at 9.4 T MR Field
The effectiveness of both samples as T 2 MRI contrast agents at high MR fields was examined by taking in vivo T 2 MR images at a 9.4 T MR field (right images in Figure 8a,b). In addition, in vivo T 2 MR images at 3.0 T MR field were also taken for comparison (left images in Figure 8a,b). Each NP suspension sample prepared in triple-distilled water was intravenously administered into mice tails and T 2 MR images were acquired before and after administration. The T 2 MR images acquired 15 (or 16) min after administration showed negative contrast enhancements (i.e., darker images) in the liver and kidneys in both samples with respect to those before administration. The negative contrast enhancements were appreciable at a 3.0 T MR field and stronger at a 9.4 T MR field for both samples. The stronger negative contrast enhancements at a 9.4 T MR field than those at a 3.0 T MR field for both samples were attributable to their higher r 2 values at a 9.4 T MR field than those at a 3.0 T MR field (Figure 7 and Table 1). These stronger negative contrast enhancements at a 9.4 T MR field can be quantitatively seen in the plots of signal-to-noise ratio (SNR) of a region-of-interest (ROI) of the liver and kidneys with respect to those before administration (Figure 8c,d). These strong negative contrast enhancements at a 9.4 T MR field confirm that both samples are potential T 2 MRI contrast agents at high MR fields. Considering that ultrasmall NPs (d < 3.0 nm) are amenable to renal excretion [17][18][19], these results imply that the PAA-coated Ln 2 O 3 NPs (Ln = Tb and Ho) might be applied to the detection of diseases such as cancer at an early stage, which could be explored in the future.

Discussion
PAA-coated Ln 2 O 3 (Ln = Tb and Ho) NPs were synthesized through the one-pot polyol method (Figure 1). Compared to D-glucuronic acid as a surface-coating ligand [49], PAA enhanced the colloidal stability as well as biocompatibility of the NPs.
The observed negligible r 1 and appreciable r 2 values can be explained as follows. The magnitudes of the r 1 and r 2 values depend on the degree of T 1 and T 2 water proton spin relaxations induced by the NPs, respectively. According to the inner sphere model, T 1 water proton spin relaxation is mainly accelerated by the magnetic dipole-dipole interaction between the 4f-electron spins of Ln 3+ (Ln = Tb and Ho) on the NP surfaces and the water proton spins in contact with or close to the NPs [22,50]. This interaction is negligible because the fast 4f-electron motions of Tb 3+ and Ho 3+ do not match with the slow water proton spin motions [50], thus providing negligible r 1 values. On the other hand, according to the outer sphere model, T 2 water proton spin relaxation is induced by the fluctuation of local magnetic fields generated by the NPs [20,21]. The model suggests that r 2 ∝ M NP 2 , in which M NP (in emu/NP) is a magnetic moment per NP [20][21][22]. The M NP values of the Ln 2 O 3 NPs are appreciable at room temperature, as can be noticed from their M values in Figure 6, explaining the observed appreciable r 2 values at 3.0 T. In addition, the enhanced r 2 values of the NPs at 9.4 T were due to their higher M NP values at 9.4 T because the paramagnetic NP samples were far from saturation ( Figure 6) and consequently, M NP value increase with increasing H until M NP becomes saturated. As shown in Table 1, bigger r 2 values were observed for the PAA-coated Tb 2 O 3 NPs than the PAA-coated Ho 2 O 3 NPs. This is attributable to slightly larger particle diameters of the PAA-coated Tb 2 O 3 NPs (d avg = 1.8 nm) than those of the PAA-coated Ho 2 O 3 NPs (d avg = 1.7 nm). The r 2 value of lanthanide oxide NPs is very sensitive to the particle diameter because bigger NPs can have larger M NP values. The M NP can be approximately estimated assuming a spherical shape for NPs and using d avg as follows: M NP ≈ µ × [number (n) of Ln 3+ per NP] and n ≈ (2/5) × (d avg /w) 3  The observed r 2 values were compared with other values (Table 2) [49,53,54]. Compared to r 2 values of D-glucuronic acid-coated Ln 2 O 3 NPs (Ln = Tb and Ho) [49] with similar particle diameters as those of PAA-coated Ln 2 O 3 NPs, the PAA-coated NPs exhibited lower r 2 values due to their larger coating ligands, explained as follows. The hydrodynamic diameters (a) of the PAA-coated Ln 2 O 3 NPs (Ln = Tb and Ho) were approximately twice those [49] of the D-glucuronic acid-coated Ln 2 O 3 NPs at the similar core particle diameters and the r 2 values of the PAA-coated Ln 2 O 3 NPs were smaller than those [49] of the D-glucuronic acid-coated Ln 2 O 3 NPs. These surface-coating ligand effects are schematically explained in Figure 9. PAA can accommodate more water molecules through its long chain length and many carboxylic groups (∼25 groups per PAA) than small D-glucuronic acid: therefore, each PAA-coated NP can be swollen into a bigger size than each D-glucuronic acid-coated NP, thus allowing a larger hydrodynamic diameter. However, D-glucuronic acid with a smaller size than PAA can allow more water molecules to closely approach the NP, making more water molecules feel stronger magnetic fields (B) generated by the NP, consequently making the D-glucuronic acid-coated NPs have a higher r 2 value. On the other hand, large NPs and nanorods [53,54] exhibited high r 2 values even with a large ligand coating because of their enhanced paramagnetic magnetic moment per NP or per nanorod, which increases with increasing particle size and because the r 2 value is proportional to the square of the magnetic moment [14].

Conclusions
PAA-coated Ln 2 O 3 NPs (Ln = Tb and Ho) (d avg = 1.8 and 1.7 nm, respectively) were prepared via one-pot polyol synthesis, and investigated as a new class of T 2 MRI contrast agents at high MR fields.
(1) Both samples exhibited excellent colloidal stability and good biocompatibility resulting from PAA-coating on the NP surfaces.