Synthesis and Magnetic Properties of Gd Doped EuS Nanocrystals with Enhanced Curie Temperatures

Abstract

EuS nanocrystals (NCs) were doped with Gd resulting in an enhancement of their magnetic properties. New EuS and GdS single source precursors (SSPs) were synthesized, characterized, and employed to synthesize Eu_1-xGd_xS NCs by decomposition in oleylamine and trioctylphosphine at 290 °C. The doped NCs were characterized using X-ray diffraction, transmission electron miscroscopy and scanning transmission electron microscopy, which supports the uniform distribution of Gd dopants through electron energy loss spectroscopy (EELS) mapping. X-ray absorption spectroscopy (XAS) revealed the dopant ions in Eu_1-xGd_xS NCs to be predominantly Gd³⁺. NCs with a variety of doping ratios of Gd (0 ≤ x < 1) were systematically studied using vibrating sample magnetometry and the observed magnetic properties were correlated with the Gd doping levels (x) as quantified with ICP-AES. Enhancement of the Curie temperature (T_C) was observed for samples with low Gd concentrations (x ≤ 10 %) with a maximum T_C of 29.4 K observed for NCs containing 5.3 % Gd. Overall, the observed T_C, Weiss temperature (θ), and hysteretic behavior correspond directly to the doping level in Eu_1-xGd_xS NCs and the trends qualitatively follow those previously reported for bulk and thin film samples.

Introduction

Magnetic semiconductors, materials whose magnetic and electronic properties are coupled, hold promise for applications in spintronics and magneto-optics.^1–6 In order to harness these physical capabilities, there has been significant study into many compositions and morphologies of such materials, most recently focusing on nanostructures.^6–12 An extensively studied subset of model magnetic semiconductors are the europium chalcogenides (EuQ, where Q = O, S, Se, Te), whose magnetic behavior ranges from ferromagnetic EuS and EuO to antiferromagnetic EuTe.^6,13–15 As a concentrated ferromagnetic semiconductor (band gap 1.6 eV) in which the aligned magnetic spins are intrinsic to the lattice, EuS is an appealing material.^16–19 Furthermore, EuS has a high spin polarization (86 %), a large band splitting (0.36 eV when T < T_C), and large effective magnetic moment (7.9 μ_B).^13,14 Despite these promising characteristics, its low Curie temperature (T_C, the temperature at which it transitions from paramagnetic to ferromagnetic behavior) of 16.6 K makes it impractical for device applications without enhancement.

As a well-understood intrinsic ferromagnetic semiconductor, EuS is a model material for exploring modulation of magnetism in nanocrystals (NCs). Magnetic semiconductors can benefit from being synthesized and studied on the nanoscale. The high surface to volume ratio of NCs means that the surface activity has an increased contribution to the overall behavior of the material as compared to the bulk, so modification to the surface can allow for modulation of electronic and magnetic properties.^9,20 When the crystals’ size is small enough that the anisotropy energy of the NC is comparable to the thermal energy, they begin to exhibit superparamagnetic behavior.^21–27 Solution synthesis of NCs enables large quantities of material to be prepared and allows for improved processing and assembly, for instance, EuS NCs can be employed as starting materials for the synthesis of magnetic composites. ^28–34

Utilization of magnetic semiconductors such as EuS for practical applications depends on the ability to improve and/or tailor their electronic and magnetic properties. In the case of EuS NCs, a major obstacle is the low T_C, therefore developing a method for enhancing the T_C is crucial. Doping with Gd to form Eu_1-xGd_xS in bulk and thin film samples has been shown to enhance the T_C to ~50 K. The similar lattice parameters of the same rock salt structure types (5.968 Å for EuS as compared to 5.5655 Å for GdS) and similar ionic radii (1.15 Å for Eu²⁺ as compared to 1.12 Å for Gd²⁺ and 0.938 Å for Gd³⁺) allow for exchange of Eu ions with Gd ions to create doped crystals for all ratios of Eu and Gd.^35,36 EuS is an ideal Heisenberg ferromagnet where each Eu²⁺ ion carries the full magnetic moment of all unpaired spins of half-filled 4f shell.^13,14 GdS itself is an antiferromagnetic metal with a Néel temperature (T_N) of 59 K.³⁷ Doping EuS with trivalent Gd ions, which also have the 4f ⁷ spin ground state, introduces additional electrons without significantly impacting the spin structure. These additional electrons first compensate for the Eu³⁺ acceptor states known to be common defects in EuS, then a small number of additional electrons significantly increases both T_C and θ (the Weiss constant) at low Gd doping level (x < 0.2). Such enhancement peaks for 0.01 ≤ x ≤ 0.03 in the bulk, because this enhancement of magnetic coupling is due to an oscillatory RKKY-type indirect exchange mechanism: when the doping level is further increased, magnetic benefits rapidly decrease, resulting in loss of enhancement and ultimately the emergence of antiferromagnetic behavior, like that seen in GdS.^14,38–40

Although methods for preparing bulk Eu_1-xGd_xS have been developed including co-melting and cooling of EuS and GdS ingots, electron-beam deposition, and flash evaporation,^39,41 Gd doping of EuS NCs has not been previously reported. Pure EuS NCs can be synthesized using a variety of methods including growth from liquid ammonia,^42,43 white light⁴⁴ and microwave irradiation,⁴⁵ zeolite templating,⁴⁶ and decomposition of single source precursors.^47,48 The traditional colloidal syntheses using separate precursors that are more prevalent for common NC growth have not been shown to be effective for synthesizing EuS NCs. Synthesis through decomposition of single source precursors (SSPs) of europium complexes with dithiocarbamate (dtc) ligands have yielded good quality, monodisperse, single crystalline EuS NCs with a stable coating of organic capping ligands.^{26,34,47–54}

In this work, in order to improve the EuS NC synthesis and enable effective doping, we first expanded the repertoire of SSPs by synthesizing and characterizing three new Eu dtc SSPs and two new Gd dtc SSPs using the ligand library shown in Scheme 1. We tested each precursor using a set of standard reaction conditions and focused on one EuS SSP, Eu(Pytc)₃Bipy (1) and its structurally similar Gd SSP, Gd(Pytc)₃Bipy (4) due to somewhat improved morphology and crystallinity. We then developed the methods for reproducibly synthesizing Eu_1-xGd_xS NCs with a range of doping ratios (x) using the simultaneous decomposition of mixtures of Eu and Gd SSPs 1 and 4. Finally, systematic structural and magnetic characterization of this series of doped NCs will be presented to show the effect of Gd incorporation on the enhancement of Curie temperatures and magnetic properties of EuS NCs.

Scheme 1.

Chemical structures and abbreviations for the ligands in Eu(Dtc)₃L and Gd(Dtc)₃L. The precursors as seen in Table 1 are as follows: Eu(Pytc)₃Bipy (1), Eu(Pipdc)₃Bipy (2), Eu(Mpipdc)₃Phen (3), Gd(Pytc)₃Bipy (4), Gd(Ddtc)₃Bipy (5)

Experimental Section

Pipecolyldithiocarbamic acid pipecolinium salt was purchased from TCI America. All other chemicals were purchased from Sigma-Aldrich Chemical Company. They were used without further purification unless otherwise noted. For syntheses of Eu(Pytc)₃Bipy, Gd(Pytc)₃Bipy, and Gd(Ddtc)₃Bipy, the dithiocarbamate salts were recrystallized from hot ethanol prior to use.

Synthesis of NH₄·Pipdc

As piperidine dithiocarbamate is not available commercially, it was prepared following a standard procedure. To an ice bath containing a flask with a vigorously stirring mixture of mixture of 0.5 mol 35 % NH₄OH (1.97 mL, 0.88 g/mL) in 15 mL of 1:1 (v/v) ethanol:H₂O was added 0.05 mol (4.94 mL) of piperidine. Carbon disulfide 0.05 mol (3.02 mL) was slowly added dropwise resulting in the precipitation of the dithiocarbamate salt. After the resultant yellow mixture was stirred for 30 minutes, white crystals were collected by vacuum filtration, washed with cold ethanol, and dried under vacuum.

Synthesis of Eu(Pytc)₃Bipy (1), Eu(Pipdc)₃Bipy (2), and Eu(Mpipdc)₃Phen (3)

Europium pyrrolidine dithiocarbamate, Eu(Pytc)₃Bipy (1); piperidine dithiocarbamate bipyridine, Eu(Pipdc)₃Bipy (2); and europium pipecolyl dithiocarbamate phenanthroline, Eu(Mpipdc)₃Phen (3) were prepared based on previously published procedures.^47–49 A clear solution of 1 mmol (0.366 g) of EuCl₃·6H₂O in 10 mL boiling distilled water was added dropwise to a clear solution of 1 mmol (0.156 g) of 2,2′-bipyridine or 1 mmol (0.180 g) of 1,10-phenanthroline in 10 mL boiling distilled water. A clear solution of 3 mmol of the desired dithiocarbamate salt in 30 mL dH₂O was prepared (0.492 g of NH₄·Pytc for 1; 0.523 g of NH₄·Pipdc for 2; or 0.823 g of Mpipdc for 3). The dithiocarbamate solutions were the added dropwise to the vigorously stirred solution of europium salt and 2,2′-bipyridine or 1,10′-phenanthroline, rapidly forming crystals of the respective SSP (orange for 1, red-orange for 2, and red-orange for 3). After stirring for 20 minutes and then cooling in an ice bath, the crystals were collected by vacuum filtration, rinsed with cold dH₂O and stored in a desiccator in the dark. The yield was ~64 % for 1, ~30 % for 2, and < 20 % for 3. IR for 1 ν_C-N = 1473 cm⁻¹, ν_C-S = 1012 cm⁻¹; 2 ν_C-N = 1477 cm⁻¹, ν_C-S = 1012 cm⁻¹; 3 ν_C-N = 1468 cm⁻¹, ν_C-S = 1002 cm⁻¹.

Synthesis of Gd(Pytc)₃Bipy (4) and Gd(Ddtc)₃Bipy (5)

The procedure for preparing the Gd SSPs was slightly modified from that for Eu SSPs. To a 10 mL solution of 1 mmol (0.327 g) of Gd(NO₃)₃·6H₂O was added a solution of 2,2′-bipyridine (1 mmol, 0.156 g) in 10 mL boiling dH₂O. The dithiocarbamate salts used were NH₄·Pytc and Na·Ddtc (3 mmol, 0.492 g and 0.514 g respectively). Upon addition of the dithiocarbamate salts to the lanthanide and bipyridine mixture, off-white crystals were rapidly formed. After stirring for 20 minutes and cooling in an ice bath, the crystals were collected by vacuum filtration, rinsed with cold dH₂O and stored in a desiccator in the dark. The yield was ~68 for 4 and ~46 % for 5. IR for 4 ν_C-N = 1473 cm⁻¹, ν_C-S = 1012 cm⁻¹, 5 ν_C-N = 1482 cm⁻¹, ν_C-S = 1006 cm⁻¹.

Characterization of the SSPs

The precursors were characterized by FTIR, single-crystal X-ray crystallography, and thermogravimetric analysis. SSPs dissolved in chloroform were dropcast onto ZnSe windows and IR spectra were taken using a Nicolet 740 FTIR from 800 and 4000 cm⁻¹. Alternatively, dry, powdered samples were observed using an attenuated total reflectance (ATR) attachment on a Nicolet 740 FTIR. Thermogravimetric analysis was performed on a Netzsch STA 409CD thermogravametric analyzer in a Pt pan from 100 – 650 °C under flowing N₂ gas. Single crystals for X-ray crystallographic analysis were grown by slowly diffusing isopropanol or hexanes into CHCl₃ solutions of SSPs. The data collection parameters and refinement results are presented in Table 1. Full details of crystal structure determination are included in the Supporting Information.

Table 1.

Summary of crystal data structure refinement for Eu(Dtc)₃L and Gd(Dtc)₃bipy.

	(1) Eu(Pytc)3Bipy	(2) Eu(Pipdc)3Bipy	(3) Eu(Mpipdc)3Phen	(4) Gd(Pytc)3Bipy	(5) Gd(Ddtc)3Bipy
Empirical Formula	EuC25H32N5S6·3CHCl3	EuC28H38N5S6·CHCl3	EuC33H47N5S6·3CHCl3	GdC25H32N5S6·3CHCl3	GdC25H38N5S6
Formula Weight	1104.98	908.32	1213.16	1110.27	758.21
Crystals system	Orthorhombic	Monoclinic	Triclinic	Orthorhombic	Monoclinic
Space group	Pbca	P21/c	P1̄	Pbca	P21/c
a (Å)	18.2696(16)	17.4375(5)	10.9505(3)	18.2665(13)	17.3870(4)
b (Å)	17.3678(15)	15.3706(4)	13.0006(4)	17.3854(12)	10.5216(2)
c (Å)	26.635(2)	13.7483(3)	18.4281(5)	26.6725(19)	17.2950(4)
α (°)	90	90	99.158(2)	90	90
β (°)	90	92.703(2)	95.478(2)	90	96.6080(10)
γ (°)	90	90	103.997(2)	90	90
Volume (Å3)	8451.4(13)	3680.79(16)	2488.44(12)	8470.4(10)	3142.91(12)
Z	8	4	2	8	4
Wavelength (Å)	0.71073	1.54178	1.54178	0.71073	1.54178
Density (calc) (mg/m3)	1.737	1.639	1.619	1.741	1.602
Abs Coefficient (mm-1)	2.379	17.609	16.083	2.459	17.558
Crystal size (mm3)	0.32 × 0.23 × 0.16	0.47 × 0.40 × 0.08	0.30 × 0.12 × 0.08	0.50 × 0.50 × 0.20	0.32 × 0.30 × 0.28
θ range (deg)	2.23 – 28.70	2.54 – 71.22	2.45 – 71.64	1.79 – 30.00	4.92 – 71.83
Reflections Collected	116495	39811	41777	151308	49621
Independent Reflections	10905(0.0760)	6860(0.0621)	9452(0.0469)	12320(0.0333)	5949(0.0335)
Data/Restraints/Parameters	10905/0/442	6860/0/397	9152/1270	12320/0/442	5949/0/340
GOF on F2	.962	1.058	1.016	0.947	1.076
R1/wR2 [I > 2σ]	0.0363/0.0903	0.0511/0.1232	0.0544/0.1419	0.0264/0.0642	0.0227/0.0587

Synthesis of Eu_1-xGd_xS NCs

All SSPs were made fresh within two weeks before NC synthesis. The procedure for synthesizing the doped Eu_1-xGd_xS NCs was modified from published procedures^47–51 for synthesizing EuS NCs but using SSPs 1 and 4. The desired molar ratio of Eu to Gd SSPs was determined such that the molar sum of both SSPs was 0.1125 mmol. For example, to synthesize EuS NCs (x = 0), 0.1125 mmol (0.0840 g) of Eu(Pytc)₃Bipy (1) was required; to synthesize Eu_0.5Gd_0.5S (x = 0.5), 0.563 mmol (0.0420 g) of 1 and 0.563 mmol (0.0423 g) 4 were required. The precursors were suspended in 2 mL of trioctylphosphine (TOP) and degassed under vacuum for 30 minutes. In a three necked flask outfitted with a septum and a thermocouple, 3 mL of oleylamine was kept at 120 °C under vacuum for 30 minutes, then heated to 290 °C under N₂, after which the precursor suspension was rapidly injected into the solution under vigorous stirring. After 30 minutes of stirring at 290 °C, 15 mL of anhydrous ethanol was injected and the solution was centrifuged for 30 minutes at 3500 rpm giving a dark purple precipitate and a pale yellow supernatant which was discarded. The precipitate was resuspended in 2 mL of ethyl acetate, precipitated with 40 mL of ethanol and centrifuged for 30 minutes at 3500 rpm. The solid was dried under a stream of nitrogen and stored under inert atmosphere in a glove box. The products varied in color from dark purple for pure EuS NCs to brown for pure GdS NCs.

Characterization of Eu_1-xGd_xS NCs

The structure and composition of Eu_1-xGd_xS NCs were characterized by transmission electron microscopy (TEM), scanning transmission electron microscopy (STEM), powder X-ray diffraction (PXRD), and ion-coupled plasma atomic emission spectroscopy (ICP-AES). Eu_1-xGd_xS NCs suspended in chloroform were dropcast onto TEM grids with lacey carbon films for STEM and TEM analysis. HRTEM imaging was performed on a Phillips CMT200 TEM with an accelerating voltage of 200 kV. STEM results were obtained on a 200 kV FEI Titan STEM with CEOS hexapole aberration corrector. Noise in the EELS spectrum image was reduced using principle component analysis (PCA), retaining the first 6 principal components.⁵⁵ Samples for PXRD were prepared by dropcasting Eu_1-xGd_xS NCs from a chloroform suspension onto Si substrates. PXRD was collected on a Siemens STOE diffractometer using Cu Kα radiation.

Knowing the exact ratio of Gd to Eu in each Eu_1-xGd_xS NC sample is crucial to this study. To determine these ratios, ICP-AES was performed using a Perkin Elmer Optima 2000 DV Optical Emission Spectrometer. ICP-AES samples were prepared by thoroughly drying the Eu_1-xGd_xS NCs to a powder, then digesting the NCs to ions with concentrated (68 %) HNO₃ and diluting to 5 % HNO₃. Calibration standards were made by diluting 1000 ppm standards of both Eu and Gd (purchased from GFS chemicals) over the range of 0.5 ppm – 3.2 ppm for each. Calibration curves for high and low concentrations were made for three emission wavelengths because instrument response was not linear over the broad range of concentrations required to accurately determine small x values. Three emission wavelengths for each Eu (381.967, 393.048, and 412.97 nm) and Gd (336.223, 342.247, and 376.839 nm) were measured. Concentrations were calculated from the emission intensities for the three Gd and three Eu wavelengths for each sample were converted to concentrations using the appropriate calibration curves and averaged.

X-ray Absorption Spectroscopy

Synchrotron X-ray Absorption Spectroscopy (XAS) measurements were performed at the Advanced Photon Source at Argonne National Laboratory using beamline 4-IDC. The NCs were dropcast onto silicon substrates and were placed in a split-coil superconducting solenoid in a liquid He cryostat. X-rays absorption scans were performed at 8 K and absorption was measured in electron yield mode. Theoretical spectra for Gd²⁺ and Gd³⁺ valence at the Gd M_5,4 X-ray edges were generated via atomistic multiplet calculations using code by van Veenendaal, in which the Hamiltonian includes the Coulomb interactions in the 4 f shell and those between the 4 f shell and the core hole and the spin-orbit coupling. Parameters were obtained in the Hartree-Fock limit using Cowan’s code and the values for the Coulomb interaction were scaled down to 80 % to account for screening effects, producing good agreement between theory and experiment.^56,57 These theoretical XAS spectra for Gd²⁺ and Gd³⁺ were interpolated and fit to the experimental XAS data.

Magnetic Measurements

Magnetic characterization was performed on a Quantum Design Physical Property Measurement System (PPMS – 9T) equipped with a Vibrating Sample Magnetometer (VSM). Eu_1-xGd_xS NC samples were stored in the glove box after synthesis and were placed in a Teflon bucket attached to a carbon fiber rod for measurement. Zero field-cooled and field-cooled scans of magnetization as a function of temperature were performed from 5 K to 300 K with an external field of 100 Oe. Hysteresis curves were collected from − 9 T to 9 T at low (5 – 6 K) and high (100 K and/or 300 K) temperatures. Magnetic isotherms, where magnetization is measured as a function of field for a given temperature, were measured for a range of temperatures surrounding the T_C of EuS (16.7 K) for use in Arrott plot calculations of T_C. Diamagnetic contributions were negligible.

Results and Discussion

Synthesis and Characterization of Precursors

Thermally decomposing Eu SSPs with dithiocarbamate (dtc) ligands in hot organic solvents is the most commonly employed and arguably the most effective method for synthesizing EuS NCs.^44,47–51 Such SSPs are neutral coordination complexes of an Eu³⁺ ion with three dithiocarbamate (dtc) ligands, and either a phenanthroline or bipyridine ligand. When heated to ca. 300 °C in organic solvents, they decompose via an internal redox reaction, resulting in EuS NCs. Since bonds within the SSP must break during the reactions, the structure and properties of these precursors could affect their performance. We designed new precursors attempting to refine/modify the decomposition behavior of these SSPs and therefore better control the synthesis of EuS NCs. A more refined synthesis of EuS NCs provides the foundation for designing an effective method for synthesizing Eu_1-xGd_xS NCs. Bulk crystals and thin films of Gd doped EuS have been prepared through co-melting, electron-beam deposition, and flash evaporation (methods incompatible with all existing EuS NC synthetic schemes) and have been shown to have enhanced T_C up to 50 K.^39,41

First, we made Eu SSPs using dtc ligands with bulkier hydrocarbon rings, such as Pytc, Pipdc, and Mpipdc (the chemical structures and abbreviations for these precursors are found in Scheme 1). Furthermore, following the most successful Eu SSPs, we designed analogous Gd dtc SSPs using Ddtc and Pytc and bipyridine ligands and synthesized them in similar aqueous reactions. The theory was that the similarity of these structures would allow for similar decomposition behavior and reactivity to facilitate the growth of mixed NCs. Prior to attempting the doping synthesis, we characterized these new SSPs (1–5) and determined the suitability of each for reproducibly growing monodisperse, crystalline NCs.

The new precursors were characterized using single-crystal X-ray crystallography. The structure refinement data for precursors 1–5 can be found in Table 1 and the structures are displayed in Figure 1. As seen in Table 2, despite the varied composition of the ligands, the most important bond lengths between metal and coordinating ligand atoms, and their bond angles are quite similar. Additionally, these bond lengths and angles are similar to those published for a variety of EuS dithiocarbamate NC precursors.⁴⁷

Figure 1.

X-ray crystal structures for precursors (1)–(5).

Table 2.

Selected bond lengths and angles in precursors (1)–(5)

Bond length (Å)	(1)	(2)	(3)	(4)	(5)
M(1)-N(1)	2.560(3)	2.595(5)	2.569(4)	2.5493(19)	2.573(2)
M(1)-N(2)	2.587(3)	2.642(5)	2.577(4)	2.5770(18)	2.5914(19)
M(1)-S(1)	2.8891(9)	2.8416(16)	2.8038(14)	2.8857(6)	2.7941(6)
M(1)-S(2)	2.8879(9)	2.8438(15)	2.8314(15)	2.8553(6)	2.8891(5)
M(1)-S(3)	2.8879(9)	2.8584(14)	2.984(6)	2.8360(6)	2.8918(6)
M(1)-S(4)	2.8611(9)	2.8154(15)	2.865(3)	2.8526(6)	2.8274(6)
M(1)-S(5)	2.8443(10)	2.8578(14)	2.8671(12)	2.8269(6)	2.8267(5)
M(1)-S(6)	2.8577(10)	2.8522(15)	2.8588(13)	2.8865(6)	2.8374(6)
Bond angle (deg)	(1)	(2)	(3)	(4)	(5)
S(1)-M(1)-S(2)	62.68(3)	62.75(4)	63.35(4)	62.469(16)	62.851(16)
S(3)-M(1)-S(4)	62.34(3)	62.69(4)	60.39(13)	63.177(17)	62.167(16)
S(5)-M(1)-S(6)	62.96(3)	62.19(4)	61.84(3)	62.843(15)	63.075(15)
N(1)-M(1)-N(2)	63.85(9)	62.85(16)	64.30(13)	64.19(6)	63.34(6)
Averages	(1)	(2)	(3)	(4)	(5)
M – N (Å)	2.573(8)	2.619(0)	2.573(4)	2.5631(7)	2.582(3)
M – S (Å)	2.8639(3)	2.8448(8)	2.868(5)	2.857(3)	2.8444(7)
C – S (Å)	1.719(1)	1.725(1)	1.727(1)	1.720(9)	1.725(0)
S – M - S (deg)	62.66(3)	62.54(7)	61.86(3)	62.829(8)	62.69(8)

FTIR shifts for the C–S and C–N stretches show remarkable similarity between precursors. TGA (Figure 2) showed relatively similar decomposition temperature for all the precursors (over the range of 200 – 350 °C) however the slope of the decomposition curve was much steeper for some precursors than for others. It should be noted that TGA is solid-state thermal decomposition in a nitrogen atmosphere, while NCs synthesis is solution thermolysis. Therefore, even though the behavior observed in TGA is a useful indicator, it might not fully predict the behavior expected during NC synthesis.

Figure 2.

TGA of single source precursors (1)–(5) and Eu(Ddtc)₃Bipy

Optimization of Precursors in the Synthesis of Pure EuS Nanocrystals

Use of SSPs allows for the synthesis of NCs from reactive or insoluble precursors (such as the Eu²⁺ salts) while providing all elements found in the final NCs in a fixed ratio. In the case of Eu dtc SSPs the ratio is 6 S²⁻ ions for every Eu³⁺ ion. The highest quality EuS NCs in crystallinity, morphology, and dispersity have been observed from decomposition of these SSPs in high boiling point solvents such as trioctylphosphine, oleylamine, oleic acid, and octadecene. Despite the general success, synthesis of EuS NCs using SSPs still suffers from a lack of reproducibility (specifically SSP batch variation) and purity (NCs samples can contain high percentages of Eu³⁺ ions, likely due to amorphous coatings, despite appearing to be 100 % EuS by X-ray diffraction techniques).⁵⁸

A primary motivation for exploring new SSPs was the thought that varying the structure of the ligands on dithiocarbamate precursors could affect the decomposition behavior, i.e. potentially changing the temperature of decomposition or the identity/quality/structure of the decomposition products. However, only minor variance was observed. One potential reason for this similarity in decomposition is the structural similarity of all SSPs as revealed by X-ray crystallography (Table 2). Solution-phase thermolysis was performed using all new Eu SSPs 1–3 plus Eu(Ddtc)₃Bipy to ensure that they would form the desired EuS NCs (TEM images of representative NCs can be found in Figure S1). After extensive comparison of the results, 1 was shown, by a slight margin, to most reproducibly yield the most monodisperse EuS NCs. 1 was used as the Eu SSP for all syntheses of doped NCs.

For the synthesis of Eu_1-xGd_xS NCs, we desired an Eu SSP and a Gd SSPs which would decompose simultaneously to result in homogenous distribution of Eu and Gd within each NC. The corresponding Gd precursor 4 was chosen for being most likely to share reactivity and decomposition behavior with 1 due to their structural similarities. The assumption of simultaneous solution decomposition is partially supported by the similar thermal decomposition behaviors seen in Figure 2, where 1 and 4 decomposes at 190 – 315 °C and 210 – 305 °C, respectively. For all doped Eu_1-xGd_xS NCs whose magnetic properties were analyzed in this paper, this pair of SSPs was used in their syntheses.

Doped (Mixed) Nanocrystal Synthesis

While there are previous reports of the synthesis of pure EuS NCs^47–49 there have been no reports of doped EuS NCs syntheses. The first step in the synthesis of EuS NCs is to make a suspension of the Eu SSP in TOP. Similarly, the first step in the synthesis of Eu_{1- x}Gd_xS NCs was to make a suspension of 1 and 4 in a desired ratio in TOP. Due to their similar decomposition behavior, when mixed suspensions of Eu and Gd SSPs are injected into a hot mixture of TOP and oleylamine at 290 °C, they decompose simultaneously resulting in the nucleation of a single alloyed phase and subsequent growth of doped NCs. It should be noted that successful uniform doping in NCs is known to be challenging due to both thermodynamic and kinetic factors,⁵⁹ therefore simultaneous decomposition does not necessarily guarantee uniform distribution of Gd ions. However, as discussed later, extensive characterization of Eu_1-xGd_xS prepared via this method provided strong evidence for the fact that they have been homogenously doped.

All of the Eu_1-xGd_xS samples we prepared and investigated in detail are listed in Table 3. The majority of Eu_1-xGd_xS samples we have prepared (samples A–K) have low concentrations of Gd (≤ 10 %) because these are ratios predicted by theory and shown in the bulk to yield the highest enhancements in T_C. The synthesis was also attempted where the SSPs were injected separately 30 minutes apart with no appreciable difference in physical properties or morphology of the resultant crystals as in the case for shaded samples E, G, I, M, O, P in Table 3.

Table 3.

Summary of doped samples of Eu_1-xGd_xS NCs

Sample	% Gd in SSP mixture	% Gd in NCs by ICP-AES	TC (K)	θ (K)
A*	0	-	16.8	15.8
B	0	-	19.5	16.7
C*	0	-	24.1	15.2
D	1	0.3	27.0	19.1
E	5	2.6	28.9	78.6
F	5	2.9	23.3	78.0
G	5	3.7	26.1	81.5
H	10	3.9	22.6	20.9
I	5	4.3	26.6	79.7
J	5	5.3	29.4	83.2
K	10	9.0	20.2	77.6
L	20	17.7	10.9	32.2
M+	50	41.6	10.9	1.4
N+	60	46.7	-	−21.7
O+	50	50.4	4.5	0.6
P+	50	50.4	10.2	8.4
Q+	80	72.1	-	−54.4

Shaded rows signify samples synthesized using multiple injections of precursors.

^*Synthesized from Eu(Ddtc)₃Phen.

⁺The T_C values are approximate for samples M–Q due to deviation from parameters required for Arrott plot assumptions to be valid.

Nanocrystal Structure/Composition Analysis

TEM examination (Figure 3) revealed semi-spherical NCs with dispersity and diameters (~6–12 nm, average 8 nm) comparable to those reported for pure EuS NCs and relatively consistent independent of the ratio of Eu:Gd precursors used (or the ratio of Eu:Gd found in the NCs). PXRD (Figure S2) gave peaks consistent with a rock-salt crystal structure and no other crystalline phase of rare earth compound, although from previous study we know there is likely an amorphous component.⁵⁸ GdS and EuS share the same rock-salt structure and nearly the same lattice parameters (EuS is 6.7 % larger).^35,36 Vegard’s Law predicts the PXRD peaks should shift with increasing Gd concentration due to differing lattice constants. The PXRD peak locations are within experimental error of those predicted (Figure S2). From the PXRD peak broadening, the average NC diameter can be estimated using the Scherrer formula to be about 6.7 nm (see Supporting Information), consistent with the average diameter estimated from TEM images.⁶⁰

Figure 3.

TEM images of doped Eu_1-xGd_xS NCs, a) x = 0.014, b) x = 0.026, c) x = 0.042, d) x = 0.065, e) x = 0.129, f) x = 0.066, g) HRTEM for a NC from f and h) FFT of g.

Understanding the changes to the magnetic properties requires the knowledge that Gd ions are integrated into the EuS NC lattice and with what exact amount. The ratio of Eu to Gd was confirmed by ICP-AES on digested NCs to be consistent with the ratio of Eu to Gd precursors employed for the synthesis, as tabulated in Table 3. The close correlation between the ratio of SSPs injected and the ratio of Eu to Gd found in the Eu_1-xGd_xS NC products demonstrates the targeted tunability of this method.

While conclusively proving true, homogeneous doping is very difficult for nanomaterial samples, from our various experimental observations we are confident in significant integration of Gd into the EuS NCs. Since ICP-AES unequivocally confirms that both Eu and Gd are contained in the samples, the fact that only one size and morphology of NCs was observed for each sample under TEM suggests that both ions are located within the same NCs.

More importantly, STEM (Figure 4) was used to confirm that Eu and Gd were uniformly distributed between NCs and within the NCs. Each EELS map presented in Figure 4 is an image composed of 2500 pixels, where each pixel contains the analysis of one EELS spectrum. The average EELS spectrum of all 2500 spectra is shown in Figure 4b with the relevant Eu and Gd edges labeled. The EELS maps of the intensities of the Eu M₅ (Figure 4c) and Gd M₅ edges (Figure 4d) clearly show that Eu and Gd are contained within the same NC (i.e. the sample is not a mixture of EuS NCs and GdS NCs) and that the two elements are uniformly distributed within each NC. While not atomically resolved, this experiment is consistent with spatially uniform distribution of Gd and Eu ions in each NC.

Figure 4.

a) STEM image of Eu_1-xGd_xS NCs (x = 0.066) mapped using EELS (50 × 50 spectra); b) The average of all the EELS spectra present in the EELS map. The Eu M₅ and Gd M₅ edges used for examining elemental distribution are indicated. c) EELS map of Eu M₅ edge intensity and d) EELS map of Gd M₅ edge intensity.

Additional strong support for the integration of Gd and Eu within a single NC is the enhancement in magnetic properties observed (see below). If EuS and GdS were completely segregated, one would expect the magnetic measurements to be consistent with a simple addition of the magnetic behaviors of each separate component. For example, for a 10 % Gd sample, complete segregation would result in the magnetization curves composed of the sum of 90 % EuS ferromagnetic character and 10 % GdS antiferromagnetic character. The observed magnetic behavior cannot be explained by a mixture of separate EuS and GdS NCs.

The Electronic State of the Gd dopants

Synchrotron XAS data of the Gd M_5,4 edges (Figure 5) was collected for a Eu_1-xGd_xS NC sample (x = 0.129). Theoretical XAS spectra for Gd²⁺ and Gd³⁺ were generated, interpolated to the experimental energy values, and a linear combination of them fit to the experimental XAS data. The best fit of this data (R₂ = 0.89) yields a composition of 97 % Gd₃₊ and 3 % Gd₂₊. This shows the predominance of the dopant Gd₃₊ ions in the 4f ⁷ spin ground state, as expected.^14,38–40

Figure 5.

a) Experimental XAS spectrum of Gd M_5,4 edges for an ensemble of Eu_1-xGd_xS NCs (x = 0.066) and measured at 8 K and the weighted sum of theoretical Gd²⁺ and Gd³⁺ XAS multiplet spectra. b) Theoretical XAS multiplet spectra for Gd²⁺ and Gd³⁺ ions for comparison.

Characterization of Magnetic Properties

Systematic physical and magnetic characterization was performed on ensembles of each Eu_1-xGd_xS NC sample listed in Table 3. The dependence of dopant concentration in EuS on magnetic behavior has been studied extensively in the bulk. To determine the T_C of each Eu_1-xGd_xS NC sample, we employed Arrott plots, a method which uses magnetic saturations extrapolated from magnetic isotherms to determine the transition temperature.⁶¹ This technique has been used previously to determine the T_C of EuS NCs, and we followed the steps outlined by Stoll.⁵² Magnetic isotherms (M ² vs. H/M) were taken for temperatures near the expected T_C and the representative magnetic isotherms for NC sample E are shown in Figure 6a. The high-field regions of the isotherms can be fit to the linear equation M

Figure 6.

a) The Arrott plot for a representative doped Eu_1-xGd_xS NC sample (Sample E) with a T_C = 28.9 K; b) The M_s ² vs. T plots for several representative Eu_1-xGd_xS NC samples to determine the T_C.

$$M^2=A(H/M)+B(T_{\text{C}}-T)$$ (1)

where M is magnetization, H is applied field, T is temperature, and A= 1/bT and B = a/bT (a and b are constants). The spontaneous magnetization, M_s, is defined as the magnetization of a material at zero applied field. Therefore, the intercept of each isotherm with the M ² axis corresponds to its M_s ² value. Thus, considering only magnetization values at zero applied field, eq. (1) is reduced to

$${M_{\text{s}}}^2={BT}_{\text{C}}-BT$$ (2)

Each M_s² is plotted against the temperature of its corresponding isotherm and fitted with linear eq. (2), as shown for representative NC samples A, E, and F in Figure 6b. The intercept of the M_s² plot with the x-axis is the temperature where M_s ² = 0 and thus gives the T_C, below which spontaneous magnetization occurs.⁵² This approximation is only valid for ferromagnetic samples where the magnetic isotherms are taken near the T_C. These requirements are not met for samples O to Q.

The complete collection of Gd concentrations, Curie temperatures, and Weiss constants (see later) for doped Eu_1-xGd_xS NCs synthesized is summarized in Table 3. Figure 7 displays the T_C as a function of Gd concentration for all ferromagnetic samples (A – L). This clearly shows the most significant T_C enhancement occurs over the range 2.6 – 5.3 % Gd with the highest T_C at 29.4 K. This is slightly greater than the concentration regime of highest T_C enhancement reported for bulk Eu_1-xGd_xS (1 – 3 % Gd).³⁹ The overall behavior qualitatively follows the theoretical model of the RKKY coupling mechanism where localized spins are coupled through conduction electrons.^14,38–41 Just as seen experimentally, Gd³⁺ ions provide the same local moments as Eu²⁺ and add electron density to the conduction band, which in turn improves the magnetic coupling, therefore enhancing the T_C.⁴⁰ The added electron density also compensates for Eu³⁺ defects and S vacancies known to be present in EuS. When a small amount of Gd ions is added, the coupling between Eu²⁺ is increased. When too many electrons are added, the density is too large for only the lower energy spin state of the conduction band to be occupied and the benefit diminishes. With excess electrons, the exchange splitting of the conduction band is not sufficiently large to maintain complete spin polarization of the conduction electrons.^14,39 One possible explanation for the smaller enhancement of T_C as compared to those observed for doped bulk EuS is the size of the materials. Nanocrystals have a higher surface to volume ratio and this surface can act as defect sites trapping electron density. Additionally, reduction in overall magnetism has been reported for small, undoped EuS NCs because they are thought to be approaching their superparamagnetic limit.²⁶

Figure 7.

T_C of Eu_1-xGd_xS NCs vs. Gd fraction for 0≤x≤0.2. Circles represent samples prepared from single injection reactions, squares indicate multiple injection reactions of M(Pytc)₃Bipy (M = Eu, Gd), and triangles indicate syntheses using Eu(Ddtc)₃Phen SSP.

We also characterized the magnetization behavior of each Eu_1-xGd_xS NC sample. As shown in Figure 8, hysteresis loops (scans of magnetization as the function of external magnetic field) were taken, one below and one above the expected magnetic transition temperatures given that the T_C for pure EuS of 16.6 K and a T_N for pure GdS of 59 K.^36,62,63 For doped samples containing less than or equal to 9 % Gd (samples D–K) the T_C is enhanced as compared to pure EuS NCs (samples A–C). For those with greater than 9 % Gd, the T_C is depressed (samples L–Q). For samples A–K, below the T_C (determined using the Arrott method discussed above), a small but measurable hysteresis is present excluding superparamagnetic behavior but indicating ferromagnetism. As the concentration of Gd increases, the observed magnetism becomes softer. For samples L–S, below the T_C, the behavior is largely paramagnetic although it is not completely linear which suggests some saturation behavior. For all samples from A-Q, above their T_C, their behavior is paramagnetic.

Figure 8.

Representative hysteresis loops of Eu_1-xGd_xS NCs at temperatures above and below the magnetic transition temperatures. Sample F (Eu_0.970Gd_0.029S) shows ferromagnetic behavior at 6 K and paramagnetic behavior at 100 K. Sample N (Eu_0.533Gd_0.467S) shows antiferromagnetic behavior at 6 K and paramagnetic behavior at 100 K.

Finally we determined the Weiss constant for each NC sample using the Curie-Weiss law

$$1/\chi =(T-\theta )/\text{C}$$ (3)

where χ is the magnetic susceptibility, T is the temperature, C is the Curie constant, and θ is the Weiss constant. Temperature dependent magnetic susceptibility data were taken for each sample and inverse magnetic susceptibility was plotted as a function of temperature to give a linear fit of the paramagnetic region as shown in Figure 9a.⁶⁴ The Weiss constants for all Eu_1-xGd_xS NC samples are listed in Table 3 and are displayed in Figure 9b for samples A–Q. Positive Weiss constant values indicate that a material is ferromagnetic and negative values indicate antiferromagnetism. As seen in Figure 9b, for samples with low Gd concentrations, the Weiss constant is both positive and enhanced as compared to undoped EuS, consistent with bulk values⁴¹ where the literature values are represented by crosses and our experimental values are represented by circles. As the concentration of Gd increases, the Weiss constant decreases, eventually becoming negative indicating the transition to antiferromagnetism.^39,65–67 The trend and values for the Eu_1-xGd_xS NC samples are similar to those published for bulk samples.⁴¹

Figure 9.

a) Curie-Weiss determination of paramagnetic curie temperature θ (temperature dependence of inverse susceptibility); b) θ vs. the Gd composition (x) of Eu_1-xGd_xS NCs for 0≤x<0.8, where the circles represent experimental results determined in this study and the crosses represent literature reported values for bulk doped Eu_1-xGd_xS.⁴¹

Conclusion

In conclusion, enhancement of the Curie temperatures and other modifications of magnetic properties of EuS NCs have been shown through doping with Gd. New EuS and GdS single source precursors were synthesized and characterized, then explored to facilitate their use in the synthesis of mixed (doped) Eu_1-xGd_xS NCs with a variety of relative ratios of Gd to Eu. Those ratios were quantitatively measured using ICP-AES and the uniformity of the dopant distribution confirmed by EELS mapping with STEM. A systematic investigation of magnetic properties showed the doping concentration of Gd directly influences the T_C, θ, and hysteretic behavior of the Eu_1-xGd_xS NCs. The T_C is dependent on the Gd concentration with a peak T_C of 29.4 K for a sample containing 5.3 % Gd. The trends in magnetic behavior are qualitatively comparable to those observed for bulk Gd doped EuS samples. In order to fully understand the contributions from Gd and Eu to the observed magnetism, these Eu_1-xGd_xS NC samples will be further investigated using X-ray magnetic circular dichroism (XMCD) techniques⁵⁸ in the future.