Effect of Gd5Si4 Ferromagnetic Nanoparticle Sizes on T1, T2 and T2* Relaxation in MRI

Until now most contrast agents (CA) that are used in Magnetic Resonance Imaging (MRI) studies have been paramagnetic. However, ferromagnetic CAs are potentially more sensitive as T
2
CAs than T
1
paramagnetic compounds due to their large magnetic moments. Previous study has shown that ferromagnetic gadolinium silicide (Gd
5
Si
4
) nanoparticles (NP) could be useful as potential T
2
CA for MRI with significantly reduced echo time (TE) compared to Superparamagnetic Iron Oxide Nanopartilces (SPION) which are currently the most widely used T
2
CA [1]. Furthermore, the need for better MRI images without the need of upgrading to the higher magnetic field strength can be achieved using better CA such as Gd
5
Si
4
NP. The quality of the image contrast in MRI is improved by shortening T
1
and T
2
relaxation times at the site or close proximity to the CA. The efficiency of aT
1
CA is defined by its relaxivity, r
1
, which is field and temperature dependent. While, T
2
agents are defined by their relaxivity, r
2
, which is dependent on both the saturation magnetization (Ms) value and the effective radius of the NPs [1, 2]. In this study, effect of Gd
5
Si
4
NP of varying sizes and with different concentrations are investigated on T
1
, T
2
and T
2
* (effective/observed T
2
) relaxations times. Gd
5
Si
4
was synthesized by arc-melting of the stoichiometric mixture of gadolinium and silicon under Ar atmosphere and then the Gd
5
Si
4
NPs were prepared by high energy ball milling of the crushed ingot. The synthesis process is described in detail elsewhere [3, 4]. Ferromagnetic Gd
5
Si
4
is extracted with NdFeB grade N52 permanent magnets placed below the beaker containing the ball milled powder in ethyl alcohol before size separating the NPs through time sedimentation process that provided three fractions (named S1, S2 and S3) with average sizes of 586 nm, 287 nm and 135 nm respectively as analyzed from SEM images (Fig. 1). XRD analysis on pre-separated sample show that Gd
5
Si
4
is the major phase while GdSi and Gd
5
Si
3
is the minor phases present in all fractions (Fig. 1). Magnetic properties measured in VSM reveal that the Curie temperature (Tc) decreases for Gd
5
Si
4
phase from 312 K for S1 to 304 K for S2 and is undetectable in S3. Another Tc observed at 105 K can be attributed to Gd
5
Si
3
phase. The M-H curves at 300 K exhibits ferromagnetic behavior descending to paramagnetic as we move from S1 to S3 fraction (Fig. 1). Prior to MRI measurements, NPs are diluted in solution with low-temperature 2% agarose with the following dilutions - 1:20, 1:200, 1:2000 and 1:20000. The high dilution factors were chosen based on solution MRI with lower dilution factors (data not shown) that exhibited extremely strong contrast at 21.1 T and unquantifiable results. Each nanoparticle layer was separated with a 1% agarose layer. MR images were acquired on the 21.1 T (900 MHz) magnet at the National High Magnetic Field Laboratory (NHMFL) in Tallahassee, FL. The magnet is equipped with Bruker Avance III console and Paravision 6.0.1 (Bruker, Ettlingen Germany). For all acquisition a 10-mm birdcage coil was used. Measurements were performed to quantify T
1
, T
2
and T
2
* relaxation times for each sample and dilution. For T
1
measurements, a turbo spin echo (TSE) sequence was used with two rare factors. The echo time (TE) was 8.8 ms and ten incrementing (12000 – 26 ms) repetition times (TR) were used. T
2
relaxation were acquired with a multi slice multi echo (MSME) sequence using a TR = 5000 ms and 20 incrementing echo time (7.5 – 150 ms). For T
2
*, a 2D gradient echo (GRE) sequence were used with TR = 5000 ms and eight incrementing TE (1.5 – 28.5 ms). Common acquisition parameters for T
1
and T
2
sequences were 2 averages, matrix = $110 \times 200$, FOV = $1.1\times2.0$ cm resulting in a $100\times100$ mm in plane resolution using a 1-mm slice while the 2D T
2
* sequence were acquired with 2 averages and a matrix of 100x55 resulting in a 200x200 mm in-plane resolution. Magnitude images were analyzed in Paravision using region-of-interest (ROIs) to cover each agarose layer as well as spacing layers. The average signal intensities were extracted and analyzed in Matlab using the Levenberg-Marquardt algorithm. For T
1
a three-parameter exponential growth function were used while for T
2
and T
2
* a three-parameter exponential decay function were employed. The results shown in Table 1 indicate higher concentrations of NPs shortens the T
2
and T
2
* relaxation times and the contrast disappears rapidly with any higher dilutions. Fraction S2 at 1/20 dilution show notable shortened T
1
and T
2
relaxation times compared to the other two fractions. Although S1 has more Gd
5
Si
4
phase volume fraction and larger average particle size compared to S2, further investigation is needed inorder to establish the cause in shortened relaxation times compared to S1 fraction.