EPR and Structural Characterization of Water-Soluble Mn2+-Doped Si Nanoparticles.

Water-soluble poly(allylamine) Mn2+-doped Si (SiMn) nanoparticles (NPs) were prepared and show promise for biologically related applications. The nanoparticles show both strong photoluminescence and good magnetic resonance contrast imaging. The morphology and average diameter were obtained through transmission electron microscopy (TEM) and high-resolution transmission electron microscopy (HRTEM); spherical crystalline Si NPs with an average diameter of 4.2 ± 0.7 nm were observed. The doping maximum obtained through this process was an average concentration of 0.4 ± 0.3% Mn per mole of Si. The water-soluble SiMn NPs showed a strong photoluminescence with a quantum yield up to 13%. The SiMn NPs had significant T1 contrast with an r1 relaxivity of 11.1 ± 1.5 mM-1 s-1 and r2 relaxivity of 32.7 ± 4.7 mM-1 s-1 where the concentration is in mM of Mn2+. Dextran-coated poly(allylamine) SiMn NPs produced NPs with T1 and T2 contrast with a r1 relaxivity of 27.1 ± 2.8 mM-1 s-1 and r2 relaxivity of 1078.5 ± 1.9 mM-1 s-1. X-band electron paramagnetic resonance spectra are fit with a two-site model demonstrating that there are two types of Mn2+ in these NP's. The fits yield hyperfine splittings (A) of 265 and 238 MHz with significant zero field splitting (D and E terms). This is consistent with Mn in sites of symmetry lower than tetrahedral due to the small size of the NP's.

Introduction

There is a critical need for clinical probes that can simultaneously assist in early detection, diagnosis, imaging, and therapeutic treatment of disease.[1−3] Silicon NPs have been shown to have minimal toxicity and are easily passivated and inexpensive to produce.[4−7] Unlike the II–VI or III–V quantum dot (QD) materials, Si is a bioinert semiconductor that is electrochemically stable. Si nanoparticles have visible luminescence properties caused by a radiative recombination of an exciton (electrons and holes) in Si.[3,8,9] Free-standing luminescent Si nanoparticles have been prepared using a variety of physical and chemical techniques, including ultrasonic dispersion of electrochemical etching,[10] laser-driven pyrolysis of silane,[11] ion implantation,[12] synthesis in supercritical fluids,[13] metal–organic chemical vapor deposition (MOCVD),[14] pulverizing and sonicating p-Si,[15,16] and solution-based precursor reduction.[17−20] Significant advances in the preparation of Si nanoparticles have been realized in the past two decades.[8] However, two important issues that remain are the reliance of many of the existing methodologies on specialized reagents or reactors and the need for hazardous HF to afford efficient control over particle size.[21] To date, only a few studies have reported the colloidal synthesis of water-dispersible Si nanoparticles,[12,22−28] and most of the studies report Si nanoparticles used solely for fluorescence imaging.[9,29]

Manganese-doped group IV semiconductors have attracted attention because of their relevance in current device technology.[30−33] Mn2+ ions have been proven to be a useful magnetic resonance (MRI) contrast agent for functional imaging[31] as a noninvasive diagnostic for detection of disease. The addition of paramagnetic Mn2+ to Si nanoparticles would allow combinations of optical detection with magnetic resonance imaging techniques[33] or magnetic separation.[32] Mn2+ ions are similar to other paramagnetic ions since they are capable of shortening the T1 of water protons. In addition, Mn2+ can also have a T2 effect, influencing the signal intensity by shortening the transverse relaxation time to produce negative contrast.[34,35] Mn2+ has potential as useful contrast agents and can be a valuable tool in obtaining more information to get detailed physiological and biochemical information from MRI.[36] Recently, we have reported that for biological applications dual-modality imaging methods have shown potential for significantly increased diagnostic accuracy compared to stand-alone imaging.[33] This has led to increasing interest in multifunctional materials that can be detected by both modalities.

Incorporating dopants into nanostructures is difficult due to their small size and large surface-to-volume ratio which is an effective barrier to impurity doping.[37] Also, the doping of either n- or p-type impurities can cause the quenching of the photoluminescence (PL) because of the efficient Auger interaction between photoexcited electron–hole pairs and carriers supplied by the dopant element.[38,39] Typically, for group IV semiconductors, homogeneous doping of Mn2+ is difficult, largely due to the extremely low solubility of Mn in crystalline group IV semiconductors near room temperature.[40]

Water-soluble Mn-doped silicon (SiMn) and Fe-doped silicon (SiFe) nanoparticles have been prepared and characterized for biological applications.[7,33,41,42] In the case of SiMn nanoparticles, a major challenge to synthesis has been the low molar ratio of Mn in the nanoparticles (≤0.05%).[42] In an effort to increase the concentration of Mn in the Si nanoparticle, we have employed manganese(II) acetylacetonate as the source of Mn. In this paper, we focus on the synthesis and characterization of the NPs and EPR characterization of the Mn site in water-soluble, paramagnetic SiMn nanoparticles. The concentration of Mn in SiMn NPs afforded by using the organic manganese salts is improved 10-fold over our previous reports.[33] Electron paramagnetic resonance (EPR) illuminates the location of Mn in the SiMn NPs. These SiMn NPs were passivated with dextran in order to reduce their overall surface charge and improve solubility in anticipation of in vivo bioapplications.[33,43,44] These dextran-coated SiMn NPs were shown to have potential as an effective T2 contrast agent.

Experimental Section

Materials

Sodium hydride powder (NaH, 95%), silicon powder (Si, 99%), anhydrous manganese(II) acetylacetonate powder (Mn(acac)2, 99.99%), and allylamine (C3H7N, 98%) were purchased from Sigma-Aldrich and were used without further treatment. Ammonium bromide (NH4Br, ≥99.99%) and N,N-dimethylformamide (DMF, 99.8%) were purchased from Sigma-Aldrich and dried before use. Water was purified using a Nanopure analytical UV water system (18.0 MΩ cm, Barnstead). All manipulations were carried out under dry N2 or Ar gas, in a glovebox, a tube furnace, or Schlenk line, using standard anaerobic and anhydrous techniques.

Preparation of Mn-Doped Na4Si4 Precursors

Different Mn-doped Na4Si4 precursors were prepared according to a modified previously published procedure from Na4Si4.[45] Mixtures of NaH, Si, and Mn powders with the molar ratios (mNaH:mSi:mMn = 1.9:(1 – x):x, and x = 1%, 3%, 5%, and 7%) and NaH, Si, and Mn(acac)2 powders with the molar ratios (mNaH:mSi:mMn = 1.9:(1 – x):x, and x = 1%, 3%, and 5%) were mixed in a Spex 8000M mill. A tungsten carbide vial, two tungsten carbide balls (diameter of ∼1 cm), and one tungsten carbide ball (diameter of ∼5 mm) were used to mill the mixtures. The milling was performed for 1 h (using 15 min on, 15 min off intervals for a total of 30 min) to obtain a homogeneous powder, which was transferred back into the glovebox.

The mixtures (NaH, Si, and Mn powder and NaH, Si, and Mn(acac)2) were placed into separate 1 mL alumina crucible boats using a second 1.5 mL alumina boat as a cover. The crucibles were placed into a quartz tube with stopcocks on both ends. The quartz tube was removed from the glovebox and placed in a horizontal tube furnace and connected to Ar flowing at 30 mL/min. The quartz tube was rapidly heated (180 °C/h) to 420 °C and maintained for 48 h followed heating at 180 °C/h to 500 °C and dwelled for an additional 24 h. After cooling to room temperature, the tube was transferred back into the glovebox where the black solid was extracted and stored.

Caution: NaH and Mn-doped Na

Preparation of Water-Soluble Mn-Doped Si Nanoparticles

A two-necked 500 mL Schlenk flask was placed into the glovebox, and Na4Si4–Mn (0.25 g, 4.9 mmol) and NH4Br (0.50 g, 5.1 mmol) were added and taken to a Schlenk line. 250 mL of degassed DMF was added by cannula to the solid. The mixture was allowed to reflux overnight under an inert atmosphere. After the reaction mixture was cooled to room temperature, 2 mL of allylamine was added, and the reaction mixture was allowed to reflux vigorously overnight while stirring. This mixture was allowed to cool to room temperature, and the mixture was centrifuged. The supernatant, a yellow solution, was decanted into a round-bottom flask. The solvent was exchanged with nanopure water using a rotavap followed by dialysis against nanopure water to purify the product of any unreacted and excess salts present. The resulting orange viscous product was filtered through a syringe filter to remove any solid impurities and was utilized for further characterization and studies.

pH Stability Study

Sodium tetraborate decahydrate was used to prepare a borate buffer for the pH stability study. Hydrochloric acid and sodium hydroxide were added to adjust the pH from 3 to 10 for a total of eight samples. Thirteen microliters was added to 3 mL of each acid and base from pH 3 to pH 10. The particles were allowed to sit for 2, 4, 6, 8, 24, 48, and 96 h, and at each time point the absorbance and emission were recorded. The eight samples were placed in an incubator for 2, 4, and 48 h at 37 °C. Following the incubation the samples were placed in a sandbath at approximately 90 °C for 4 h. The absorbance and emission were recorded for these time points.

Characterization

The X-ray powder diffraction data were collected with a Bruker D8 Advance diffractometer using Cu Kα radiation (40 keV, 40 mA). The samples were ground into powders and measured using a commercial sample holder supplied by Bruker for air-sensitive materials or a quartz plate for the precipitates from the reactions. Data were collected on a rotating sample (15 rpm) from 20° to 80° (2Θ), with a step size of 0.019 44° and a step time of 20 s for the air sensitive and 2.4 s for the non-air-sensitive samples, respectively. Data acquisition was performed using the Bruker Command Center software. No standard was added to the measurements. MDI Jade was used for phase identification; full pattern fitting for the present and identified phases was performed using the FullProf Suite.[46] Elemental analysis was performed by ICP quadrupole mass spectrometers (ICP-MS, Agilent Technologies 7500ce). Transmission electron microscopy (TEM) was performed on a Philips CM-12, operating at 80 kV. TEM samples were prepared by dipping holey-carbon-coated, 400-mesh electron microscope grids into the aqueous solution of the product and drying them in air. High-resolution transmission electron microscopy (HRTEM) was carried out using a JEOL 2500SE Schottky emitter microscope operated at 200 kV with a Gatan multiscan camera associated with it. The images were captured using a Digital Micrograph software provided by Gatan Inc. Infrared spectra were obtained by dropping the liquid onto the plate of a Bruker optic GmbH alpha ATR-FTIR. Photoluminescence (PL) spectra were measured on a FluoroMax-3P fluorometer, where the Mn-doped Si nanoparticles were dispersed in water.

Longitudinal (T1) and transverse (T2) relaxation times were measured at 60 MHz (1.4 T) and 37 °C on a Bruker Minispec mq60 (Bruker, Billerica, MA). The solutions of Mn-doped Si nanoparticles were prepared by dissolving appropriate amount of samples in pH 7.0 deionized water. T1 values were measured using an inversion recovery sequence with 10–15 data points, and T2 values were measured using a Carr–Purcell–Meiboom–Gill (CPMG) sequence with τ = 1 ms and 200 data points. Each solution was incubated at 37 °C for 10 min before measurement and measurements performed at 37 °C. The longitudinal (r1) and transverse (r2) relaxivity were determined as the slope of the line for plots of 1/T1 or 1/T2, respectively, against increasing manganese concentration with a correlation coefficient greater than 0.99.

Magnetic resonance imaging (MRI) was performed on a Bruker vertical bore 9.4 T (400 MHz) microimaging system, 21 °C, Bruker, Billerica, MA). The magnet was equipped with the standard gradient set (95 mT m–1 maximum gradient) and 25 mm internal diameter (i.d.) volume coil. Parameters for the images were TR = 500 ms and TE = 5 ms. For all images a flash-2D sequence was used with a field view (FOV) of 4.40 × 4.45 cm2, slice thickness 1.0 mm, and a 128 × 128 matrix.

The electron paramagnetic resonance (EPR) spectra for the samples were collected on a Bruker ECS106 X-band spectrometer, equipped with an Oxford Instruments liquid helium cryostat. Typical experimental conditions were frequency 9.68 GHz, temperature 10 K, modulation amplitude 10 G, microwave power 63 μW, conversion time 40.96 ms, time constant 40.96 ms, resolution 2048 pts, and average of 10 scans. Samples were dissolved in dimethylformamide (DMF).

Results and Discussion

Varying concentrations of Mn2+-doped Si (SiMn) NPs were made by the reaction of the Mn-doped precursor Na4Si4 and NH4Br in DMF and capped with poly(allylamine). Adapting a previously published procedure for M-doped Na4Si4, Na4Si4M (M = Mn, P, Fe),[41,45,47] mixtures of NaH, Si, and Mn metal powders or Mn(acac)2 powders with different molar ratios were explored to determine if the Mn powder or a Mn complex would provide an increase of the Mn doping concentration thus leading to differing physical properties. The reaction equations are provided in Figure . Mn metal powder had been employed previously,[33] and in this study the Mn(acac)2 complex is investigated as a potentially better starting material for Mn2+ incorporation. Three different starting concentrations of Mn(acac)2 complex (1, 3, and 5%, noted as A, B, and C) were employed to produce Mn2+-doped sodium silicide (Na4Si4Mn).[33]Figure b shows the reaction equation using the metal complex Mn(acac)2. After reaching the final temperature of 500 °C, the main phase formed is Na4Si4Mn.

Figure 1

(a) Schematic representation of Mn-doped Zintl phase Na4Si4Mn with Mn metal powder. (b) Schematic representation of Na4Si4Mn with Mn(acac)2. (c) Schematic representation of synthesis of poly(allylamine)-terminated SiMn NPs.

Powder X-ray diffraction was used to characterize Na4Si4Mn precursors (Supporting Information). Table provides the lattice parameters for the main phase, Na4Si4Mn. As the amount of Mn(acac)2 is increased, thermally stable binary manganese silicide phases are formed, and for the highest Mn content, a clathrate phase is observed in addition to Na4Si4Mn. The lattice parameters obtained for the three different preparations of Mn doping are consistently larger than that for pristine Na4Si4 (monoclinic, C12/c1 (no. 15), a = 12.1536(5) Å, b = 6.5452(5) Å, c = 11.1323(6) Å, β = 118.9(1)°).[48] The volume increases from 780.66 Å3 for preparation A to 784.18 Å3 for preparation C; thus, we concluded that Mn2+ is incorporated into the structure of Na4Si4.

Table 1

Lattice Parameters of the Na4Si4Mn Samples

		lattice parameters
preparation	% Mn added	a (Å)	b (Å)	c (Å)	β (deg)	V (Å3)
ref [48]	0	12.1536(5)	6.5452(5)	11.1323(6)	118.9(1)	775.26
A	1	12.1874(2)	6.5607(1)	11.1650(3)	119.018(1)	780.66
B	3	12.1999(2)	6.5691(2)	11.1706(4)	119.105(2)	782.20
C	5	12.2903(4)	6.5750(3)	11.0851(3)	118.905(2)	784.18

SiMn NPs are produced according to the reaction scheme in Figure c, and the resulting product is composed of two phases: a solid and liquid phase. The liquid phase is further purified while the precipitate is washed with methanol and characterized by X-ray diffraction. Although the precipitate is not used further, identifying the products in the precipitate provides insight into Mn2+ incorporation into the solution containing poly(allylamine)-coated SiMn NPs. The precipitate shows the presence of crystalline Si along with different manganese silicides and other phases (Supporting Information). The identified phases for the precipitate in all reactions are Si with increasing amounts of Mn4Si7. This seems reasonable since it can be observed from the Mn–Si phase diagram that manganese silicides are formed with an excess of either silicon or MnSi.[49]Table provides the lattice parameters for the precipitated Si obtained in these reactions and shows that while the lattice parameter increases for preparations A and B (1 and 3% Mn), it decreases for preparation C, the 5% initial composition of Mn2+. This result suggests there is a threshold of how much Mn2+ can be incorporated into Si by this method. In addition to the lattice parameters, Table provides the composition of the Mn2+ in the solution for the SiMn NPs. As the concentration of Mn(acac)2 is increased from 1 to 5% in the Na4Si4 precursor, the highest concentration obtained for Mn in the SiMn NPs in solution was found to be 0.4 ± 0.3%. This is a 4 times greater Mn content compared to the nanoparticle product from Mn powder which yielded a maximum of 0.1 mol %.[33] The Mn complex allows for an increase in the Mn concentration in the NPs, which did not occur when using Mn powder. Mn(acac)2 has the advantage of being more precisely weighed because of its molecular weight and better distributed in the Si powder via the ball milling than Mn metal powder which typically has a thin oxide coating. Since the sample is heated under flowing argon, the organic acetylacetonate is easily decomposed and removed from the sample. The lattice parameters obtained for the solid precipitate are consistent with the chemical analysis of the SiMn NP solution, suggesting that the highest Mn2+ content is for the materials preparation B with 3% Mn(acac)2.

Table 2

Resulting Mn:Si (mol %) from the Nanoparticles in Solution from ICP-MS, Lattice Parameters of Si Precipitate from Powder X-ray Diffraction, and Quantum Yield (QY) of the Solution

		lattice parameters
Mn:Si (mol %)	preparation	a (Å)	V (Å3)	QY (%)
0.14 ± 0.04	C	5.4293(1)	160.04	13.2 ± 1.5
0.20 ± 0.1	A	5.4317(1)	160.25	11.1 ± 2.0
0.44 ± 0.3	B	5.4355(1)	160.59	13.1 ± 1.2

The morphology and size of SiMn NPs from Mn(acac)2 (preparation B) were obtained by both low- and high-resolution TEM and EDS (Figure ). In Figure a, the inset indicates a histogram from the high resolution data with the average diameter of 4.2 ± 1.0 nm, consistent with previously reported average sizes that are obtained from this reaction.[33,50] The lattice fringes of 1.87 Å were consistent with the (220) silicon planes (Figure b). Lattice fringes measured from several HRTEM images verified that the NPs are diamond structured Si (Supporting Information, Figure ). Selected area electron diffraction (SAED) is also consistent with crystalline Si (Figure c). Energy dispersive X-ray spectroscopy (EDS) (Figure d) provides further evidence confirming the presence of Si. However, Mn was not detected presumably due to the low concentrations within the NPs.

Figure 2

(a) TEM of SiMn nanoparticles preparation B with a histogram inset indicating an average size of 4.2 ± 1.0 nm. (b) High-resolution TEM showing the (220) lattice plane for Si. (c) SAED showing crystalline planes corresponding to Si. (d) EDS showing presence of Si.

Figure 4

SiMn nanoparticles preparation B (0.4 ± 0.3 Mn:Si mol %), and concentrations are indicated for Mn2+. (a) T1 and T2 weighted MRI of phantoms and (b) emission intensity measured at 350 nm.

Figure shows the FT-IR attenuated total reflectance (ATR) absorption spectrum of poly(allylamine)-coated SiMn NPs produced from the Mn(acac)2 reaction. An N–H stretch is observed at 3393 cm–1 with CH2 stretches at 3018 and 2929 cm–1. An NH2 deformation can be seen at 1660, 1640, and 1531 cm–1. The C–N stretch is at 998 cm–1, and the CH2 rocking is at 934 cm–1. This spectrum is similar to what is found in the literature for poly(allylamine).[51]

Figure 3

FT-IR ATR spectrum of poly(allylamine)-coated SiMn NPs.

The pH and temperature stability of the SiMn NPs was tested. A photoluminescence study was used to determine possible decomposition based on a decrease in emission after treatment. The particles were allowed to sit for periods of time at increasing pH, and the emission was recorded. A quantum yield of 11–15% was observed (Supporting Information). In a second experiment, the samples were heated at 37 °C for periods of time and the quantum yield was maintained at 11–15%, suggesting high stability of the NPs. However, higher temperatures (90 °C) for 4 h resulted in a significant decrease in quantum yield to about 5%.

Figure shows the T1 and T2 weighted MRI of phantoms of preparation B SiMn NPs and the optical emission of the same samples excited at λex = 350 nm. The maximum emission is red-shifted and reduced as the concentration of the sample increased, indicating aggregation and quenching of the intensity. This result is similar to that observed for iron-doped Si, SiFe, NPs.[41] The intensities of the MRI phantoms demonstrate the brightening (top row) of T1 weighted images and the darkening (bottom row) of T2 weighted images vs concentration. Thus, SiMn NPs are both T1 and T2 relaxation agents and can be used in T1 or T2 weighted applications to deliver drugs coordinated in the polymer or conjugated through the amines for several different types of imaging. These data indicate that at least some of the Mn2+ ions are near the surface of the NP in order to relax water protons via inner- and outer-sphere mechanisms for T1 weighted contrast. This is also consistent with EPR results (see below).

Table provides the Mn content in mol % (Mn:Si) and the measured relaxivities with and without dextran coating in mM per Mn2+. The preparation label is the same as for Table ; however, since we want to correlate the results with the amount of Mn2+ in the colloid, the order changes. Both Mn concentration and the dextran coating vary the relaxivities r1 and r2. The fits to the relaxation data vs concentration that yield the values in Table are shown in the Supporting Information.

Table 3

Resulting Manganese Content in mol % (Mn:Si) and Measured Relaxivities per [Mn2+] with and without Dextran Coating of the SiMn NPs

		SiMn NPs		dextran-coated SiMn NPs
Mn content (mol %)	preparation	r1 (mM–1 s–1)	r2 (mM–1 s–1)	r1 (mM–1 s–1)	r2 (mM–1 s–1)
0.14 ± 0.04	C	18.2 ± 3.6	211.0 ± 18.2	7.8 ± 0.3	822.4 ± 77.9
0.2 ± 0.1	A	0.40 ± 0.01	9.5 ± 1.0
0.4 ± 0.3	B	11.1 ± 1.3	32.7 ± 4.2	27.1 ± 1.6	1078.5 ± 200.0

Previously reported samples of SiMn NPs equivalent to preparation A, but prepared using Mn metal powder to form Na4Si4Mn, were dextran sulfate coated and had an r2/r1 of 3.49 (with an r1 relaxivity of 25.50 ± 1.44 mM–1 s–1 and an r2 relaxivity of 89.01 ± 3.26 mM–1 s–1 with respect to [Mn2+] (37 °C, 1.4 T)), indicating their use as a positive contrast agent.[33] Preparations B and C of SiMn NPs prepared with Mn(acac)2 were chosen to be coated with dextran since the poly(allylamine)-coated particles showed promising relaxivities. An r1 relaxivity similar to the previously published results, 27.1 ± 2.8 mM–1 s–1, was obtained but showed a larger r2 relaxivity of 1078.5 ± 1.9 mM–1 s–1. Likewise, preparation C showed similar results with a large r2 of 822.4 ± 77.9 mM–1 s–1.

The measured relaxivities are consistent with X-band electron paramagnetic resonance (EPR) results. EPR spectra of the NPs from the three preparations are shown in Figure .

Figure 5

EPR spectra of the three different preparations A–C of Mn2+-doped Si NPs, SiMn.

These are classic looking Mn2+ EPR spectra with S = 5/2 and I = 5/2. The EPR spectra exhibit six large peaks that are the allowed (Δm = ±1) transitions, and the smaller doublets between are forbidden transitions (Δm = ±1 and Δm = ±1).[52] Several attempts at fitting the spectrum in EasySpin[53] were made.

Isotropic A and g tensors were used along with the zero field splitting parameters D and E (eq ). Strain in D and E was also incorporated. The data cannot be fit with one component. To reproduce the long tails on either side of the lines requires a large amount of D strain. But large amounts of D strain overly broaden the sharp features. Thus, two component fits were attempted. We label the two components as type 1 and type 2. The EPR parameters g, A, [DE], [D strain E strain], and line width were varied, albeit not simultaneously. The two-component fits were significantly better than fitting only one component in every case, and for preparation C, a third spin 1/2 component was required. Figure shows an example of the fit. In particular, the two-component fits reproduce the EPR spectrum significantly better in the region from 350 to 360 mT. The fits yielded a hyperfine coupling (A) value of ∼265 MHz for component type 1 for all three doping concentrations, which is similar to previously published results on Si-doped Mn2+ nanoparticles,[42] while for type 2, a value of ∼238 MHz was obtained. In the case of Mn2+ doping there was also a clear S = 1/2 signal at g ∼ 2 in an initial data set taken with a 1 G field modulation (data not shown). The 10 G modulation used for acquiring the data for fitting diminishes the spin 1/2 signal to the point where it is not evident. Thus, a third species of spin 1/2 was also included in the fitting. These data are consistent with previous EPR on similarly prepared Si:Mn NP also acquired with 10 G modulation.[42] We attribute this signal to silicon dangling bonds. The fitting of the X-band data is very slow, indicating shallow minima in the least-squares fitting.

Figure 6

EPR data for preparation A SiMn NPs and fits. The data are shown in black, one-component fit is in green, and two-component fit is in red. Expanded view shows the feature that the two-component fit reproduces that the one-component fit does not.

The EPR parameters of Mn2+ doped into Si take on a variety of values depending on sample preparation.[54−60,60−63] Both charge and spin state may be adjusted via doping, and each combination exhibits own hyperfine interaction. S = 5/2 is the natural state on Mn2+ in Si without doping, and S = 5/2 Mn2+ has been observed in Si in both interstitial and substitutional sites with A = −160 and −121 MHz, respectively.[58,63] Muller et al.[63] indicate that rapid quenching of transition metals in Si always produces interstitial type impurities. Substitutional Mn2+ is formed when vacancies annihilate with interstitial Mn2+.[57] Note that the values for A obtained in Table are about a factor of 2 larger than the studies cited above where Mn2+ is doped into single crystals but consistent with the previous report on SiMn NPs.[42]

Table 4

EPR and Relaxivity Parameters for the Three Preparations of SiMn NPs

	Mn content, mol % preparation
parameters	0.14 ± 0.04 C	0.20 ± 0.1 A	0.44 ± 0.3 B
type 1
A (MHz)	264	266	264
[DE] (MHz)	[459 126]	[374 151]	[453 121]
[DE] strain (MHz)	[231 56]	[361147]	[251 41]
weight	4.4	6.8	3.8
type 2
A (MHz)	237	238	237
[DE] (MHz)	[925 −296]	[502 −65]	[894 −31]
[DE] strain (MHz)	[12 57]	[67 7]	[12 75]
weight	3.8	0.8	1.8
type 2 to type 1 ratio	0.86	0.12	0.46
r1	18.2	0.4	11.1
r2	211	9.5	32.7
dextran-coated r1	7.8		27.1
dextran-coated r2	822.4		1078.5

Both interstitial and substitutional sites have tetrahedral symmetry, and thus D and E terms are not expected unless there is deviation from that symmetry. Incorporation of the zero-field splitting (ZFS) accounts for the crystal-field-induced lowering from cubic (or higher) symmetry. This term can be expanded (to first order) as a function of two empirical parameters, D (the axial contribution to the ZFS) and E (the rhombic contribution):[52]D and D strain (and thus E and E strain) were necessary for adequate fits and this is the first (that we know of) reported values of D and D strain in Mn-doped Si. We note that D and D strain have been used in fitting EPR spectra of Mn:CdSe quantum dots.[64] Previous studies of Mn doped into Si, which were performed on single crystalline Si, showed Mn doping levels several orders of magnitude lower than achieved here.[59] Thus, the Mn impurities were in sites of tetrahedral symmetry. Here, the small size (∼4 nm diameter) of the particles would argue that a significant fraction of the Mn atoms are not in a site of tetrahedral symmetry due to being near the surface. Moreover, the high Mn concentration indicates there are multiple Mn per NP (at 1% there are on average ∼9.5 Mn atoms per NP). Furthermore, the standard theory for EPR parameters on Mn in Si demonstrates that next-nearest neighbors are important to the understanding of EPR spectra.[54] Thus, the existence of nearby electron spins in the form of another manganese and loss of tetrahedral symmetry for sites near the edge provide sufficient justification for the incorporation of D and E and their respective strains into the fits. Results are given in Table . These numbers are approximate as the fit minima are shallow. A high field EPR experiment would provide more definitive information.[64]

An important result from the fitting is that we obtain the relative amounts of the two types of sites. We note that for plain NPs there is a mild correlation between the type 2 to type 1 ratio and the r1 relaxivity whereas there is a strong correlation between the type 2 to type 1 ratio and the r2 relaxivity data (Table ) but not measured Mn content. What is unclear from these data is the physical location of these two sites. There are several possible interpretations of this data. The first is that a “surface” type is increasing with the type 2:type 1 ratio, placing the magnetic moment closer to the water. We note that on a NP of 4.2 nm a shell of 0.4 nm (the Si next-nearest-neighbor distance) occupies approximately 1/4 of the volume of the NP. This suggests that the two sites could be from “surface” (in the outer shell) and “core”. The high relaxivity argues for the larger fraction of type 2 Mn2+ to be near the surface. Furthermore, the D values obtained for type 2 are a factor of 2 larger than that of type 1, which is what one would expect if the Mn atoms are near the surface due to lower symmetry.

The relaxivity does not correlate with resulting Mn content. This lack of correlation of relaxivity with measured Mn content would be consistent with the correlation with type 2/type 1 ratio if the different preparations produced different distributions of Mn2+ within the NPs. Differences in average size of the NPs may also play a role, leading to different distributions of Mn2+. Another possibility is that the spin 1/2 signal observed in preparation C is responsible for the large r2.

r1 and r2 for the dextran-coated SiMn NPs both have a negative correlation with type 2:type 1 ratio but correlate with measured Mn2+ concentration. Of note is the particularly large r2. The large r2 combined with the ability to be further conjugated with target specific molecules makes them attractive for future bioapplications.

Conclusion

A higher percentage of Mn incorporation into Si nanoparticles was achieved by using Mn(acac)2 to form the starting material, Na4Si4Mn. X-ray diffraction supports the claim that Mn is incorporated in the starting material, Na4Si4Mn, due to the increased lattice parameters. Incorporation of Mn into the SiMn NPs is assumed since the remaining precipitate consists of manganese containing phases. ICP analysis provides the amount of Mn and Si in the nanoparticle colloid. All samples maintained their photoluminescence and produced significant T1 contrast. The SiMn NPs also show stability for extended periods of time throughout broad pH ranges. EPR indicates a two-component system where the Mn2+ ions occupy two different sites within the NPs and that the sites are of lower symmetry compared to Mn2+ doped into crystalline bulk Si. Moreover, since r1 and r2 are large, at least one type of Mn2+ is probably close to the surface. Characterization of the poly(allylamine)-coated Si NPs showed promising relaxivities for bioapplications. A next logical step is a surface modification with target specific molecules to probe the application in biological systems.