1. Introduction
Magnetic nanoparticles represent one of the most studied classes of nanoparticles because of their size-depended magnetism, especially in the nanoscale regime, and their potential applications in many fields [
1,
2,
3,
4,
5,
6]. Consequently, a lot of research effort focuses on the development of synthetic methodologies to precisely control their size, shape, composition, crystal structure and surface chemical modification [
6,
7,
8,
9].
Chemically ordered, bimetallic CoPt nanoparticles are extensively studied due to their technological interest in a variety of fields such as permanent magnets, high density magnetic storage media, catalysis, and biomedicine [
10]. The CoPt phase diagram includes several chemically ordered phases as Co
3Pt, CoPt and CoPt
3 [
11,
12]. The equiatomic CoPt, when chemically ordered, forms the L1
0 high magnetic anisotropy (K
u = 4.9 MJ/m
3) tetragonal phase with alternating atomic layers of Co and Pt along the c axis. This phase is reported to exist in the range of 40-60 at. % Platinum composition. The off-equiatomic stoichiometry compositions can be only partially ordered and consequently have somewhat lower anisotropies. On the Pt-rich and the Co-rich sides the CoPt
3 and Co
3Pt are formed respectively, which are both chemical ordered cubic L1
2 phases and have moderate magnetic anisotropy [
13,
14].
The high anisotropy values of these CoPt phases [
15,
16] allow the stabilization of their magnetization against thermal fluctuations and demagnetizing effects at very low dimensions [
17,
18,
19,
20,
21], properties which are fundamental for application in permanent magnets and recording media [
22,
23,
24,
25]. Additionally, their chemical stability in alkaline and acidic environments makes them excellent candidates for low platinum content electrocatalysts [
26,
27,
28,
29,
30,
31,
32,
33,
34].
One of the major commercialization issues still remains the high precious metal content, up to 50 at. %, equivalent to more than 77 wt. %, which is necessary, as previously mentioned, in order to form the chemically ordered L1
0 structure. Therefore, there is still a great challenge on the synthesis of CoPt bimetallic alloys, with as low as possible Pt content, maintaining high magnetocrystalline anisotropy, and consequently high coercivity. Previous studies have presented a plethora of chemical routes for the synthesis of ultrafine, monodispersed, bimetallic CoPt nanoparticles [
35,
36,
37,
38,
39,
40,
41]. The as-made nanoparticles, in the majority of the studies poses the disordered face centered cubic (fcc) crystal structure and shows superparamagnetic behavior. Our group recently published the effect of bismuth addition on the equiatomic CoPt L1
0 ordering, in which the particles show partial ordering without any post annealing exhibiting 1.7 kOe room temperature coercivity [
42]. In general, thermal treatment at elevated temperatures is required to obtain the fully ordered, magnetically hard, L1
0 crystalline phase. The bulk order-disorder transition of CoPt occurs at 825 °C [
43], while in the nanosize regime, ordering can take place at temperatures above 650 °C [
44,
45]. Regarding the synthesis of the L1
2 Co
3Pt phase, there are a limited number of works in the literature, and the majority of them are related to synthesis in thin films form [
9,
46,
47,
48,
49,
50,
51]. Here we report the colloidal synthesis of platinum lean CoPt bimetallic nanoalloys following a modified polyol methodology. The materials reveal high uniformity, narrow size distribution and exhibit enhanced room temperature coercivity, up to 14.5 kOe, value which is among the highest in the literature concerning CoPt based nanoalloys of any composition. The enhanced magnetic properties result from the coupling between the magnetically hard L1
0, and semi-hard Co
3Pt phases, leading to the formation of an exchange-spring nanocomposite magnetic material.
3. Results and Discussion
The size and the morphology of the platinum lean CoPt nanoalloys in both as-made and extensively annealed (700 °C, 7 h) form were studied by TEM microscopy (
Figure 1). The as-made nanoparticles are uniform, and quite monodispersed, while after annealing at 700 °C for 7 h under H
2/Ar flow, the particles became slightly irregular without significant sintering and agglomeration. The as-made particles are well spherically shaped with 7.1 nm mean diameter and narrow size distribution, while the annealed material loses the spherical morphology, and has an average size of 15.7 nm (size distribution histogram is presented in the
supplementary material section, Figure S1).
Figure 2(a), illustrates the infrared (IR) spectrum of CoPt nanoparticles capped with oleic acid-oleyl amine. The spectrum collected in CCl
4 colloidal solution in a liquid cell and shows strong bands at 2854 and 2928 cm
-1, assignable to the symmetric and asymmetric CH
2 stretches of the hydrocarbon moiety respectively; a shoulder at 2960 cm
-1, due to the asymmetric stretch of the terminal CH
3 group and a weak yet definite band at 3006 cm
-1, which is attributed to the olefinic CH stretch from both oleic acid and oleyl amine molecules. The broad bands around 3400 cm
-1 are attributed to the NH
2 group, confirming the presence of oleyl amine on the nanoparticle surface. Additionally, two other absorptions at 1569 cm
-1 and 1412 cm
-1 are characteristic of the presence of carboxylate (-COO-) groups. Moreover, the frequency separation between the bands at 1569 cm
-1 and 1412 cm
-1 indicates that the oleate moieties bind the surface atoms in a chelating or bridging mode of coordination. Finally, the band at 1464 cm
-1 is associated with the CH
2 deformation (ν
CH2). These data clearly demonstrate the anchoring of both capping agents to the nanoparticles surface. The as-made oleic acid-oleyl amine capped CoPt nanoparticles are easily dispersible in non-polar organic solvents such as hexane, toluene and chloroform with concentration up to several dozen of mg/mL, and are stable for weeks without precipitation. Thermogravimetric analysis was used to determine the mass of the organic matter absorbed on the CoPt nanoparticles (
Figure 2(b)). The experiments were carried out under nitrogen atmosphere with 60 mL/min flow and 5 °C/min heating rate. The organic molecules desorption begins around 140-150 °C and the total weight loss is 28 %. At temperatures about 360 – 450 °C the curve approaches an intermediate plateau. This behavior suggests, in agreement with the literature [
52], that the weight loss from 140 to 360 °C is due to the desorption of the amine ligand, and the weight loss above 360 °C is mainly due to the desorption/decomposition of carboxylic residue. The overall desorption procedure completed at approximately 500 °C. The reaction yield was estimated gravimetrically. The dry CoPt powder from a single batch was weighed 254 mg. Considering that 28 wt. % belongs to the organic capping molecules, the net mass of CoPt nanoparticles is 182.9 mg, resulting in a reaction yield of up to 93.6 %. Assuming that the Pt
2+ reduced quantitative, due to the positive reduction potential (+1.18 V) in contrast with the Co
2+ (-0.282 V), the estimated nanoalloys atomic composition is Co
74.4Pt
25.6, which is very close to the nominal composition, (Co
77Pt
23), according to the reaction precursors molecular quantities.
Figure 3(I), shows the powder XRD patterns of the as-made, as well as, the annealed at 700 °C, bimetallic CoPt nanoalloys. The pattern of the as-made material, (
Figure 3I(a)), shows clearly that the dominant phase is the chemically disordered fcc. As for the reflection at 2-theta = 31.7° degrees, which remains unmatched and indicated with an asterisk, can be attributed to cobalt oxide phase, probably the Co
3O
4 [
53]. Furthermore, the broad peaks correspond to a Scherrer structural coherence size of 6 nm in diameter, which is slightly smaller than the particle size which was estimated by the TEM studies (7.1 nm). It is worth to mention that the large unit cell, a=3.88 Å, shows a Pt rich phase [
8,
54]. This behavior has been previously reported from our group [
51], for similar materials, and due to strong compositional gradient from the nanoparticles core to the shell, indicating the presence of a Pt-rich composition at the center of the particle which progressively decrease, and leads to a Co-rich shell and can be explained by the reaction conditions. PEG’s, without alkaline environment, as well as, oleyl amine, are mild reducing agents and not able to reduce Co
2+ to the metallic state [
55]. The presence of platinum ions in the reaction, which due to their electropositivity, reduced very easily, without the need of strong reducing agents, leads to the initial formation of tiny Pt seeds, possessing very negative redox potential [
56], which consequently can reduce much more easier the Co
2+ to the metallic state, leading to the formation of particles in which their core is Pt rich [
54]. We have also noticed that the chemical ordering requires a short-length scale diffusion process which means that this compositional gradient can be present even if chemical ordering to various phases PtCo-Pt
0.4Co
0.6-PtCo
3 (
from center to shell) has been reached. After thermal treatment at 700 °C for 4 hours, the X-ray diffraction pattern in
Figure 3I(b) reveals mixtures of structures consisting of the following phases: A tetragonal phase with a=3.79 Å and c=3.68 Å clearly indicated by the presence of: (001)∽24°, (110)∽33.2°, superlattices, and additionally the presence of (111)∽41.7°, (200)∽47.9° and (002) at 49.3°. The lattice constants of this tetragonal phase are those expected for the stoichiometric fully ordered L1
0 CoPt alloy [
53,
57,
58]. On the other hand, the weaker diffraction at 40.5° can be assigned to the (111) of cubic structured CoPt binary alloy. In the case of annealing for longer time, 700 °C for 7h, the shifting and splitting of the main diffractions of the as-made nanoparticles, shows that part of the alloys transformed to the ordered L1
0 and cubic L1
2 Co
3Pt phases [
59,
60]. More detailed examination of the powder X-ray diffraction pattern in
Figure 3I(c), shows obvious existence of the (001) and (110) superlattice peaks which clearly indicate the transformation to the chemically ordered tetragonal L1
0 phase [
54,
57,
58], while some extra peaks indicate the presence of another phase with weak tetragonality: a=3.73Å and c=3.71Å indicated by the presence of: (111)∽41.88° and (200)∽48.8°, (002)∽49.7°. Since the order parameter S scales with the tetragonicity as S
2∽ (1- c/a) the c/a=0.996 compared to the 0.973 of the fully ordered phase gives S=0.37. This could be due to the off-stoichiometric composition of this phase as the lattice parameters are close to what is expected for Pt
0.4Co
0.6. The S of this cobalt-rich phase shows that could reach an anisotropy of 1.8 MJ/m
3. Finally, the stronger peaks can be assigned to the cubic L1
2 Co
3Pt phase with a=3.66Å: (111) ∽42.6°, (200) ∽49.5°, (220) ∽72.9° and (311) ∽88.4°, and is in agreement with the estimated by HR-TEM images, d-space value (
Figure 1(f), 0.211 nm), and assigned with the (111) plane of the Co
3Pt phase, which is also very close to the results reported in the literature [
16]. It should be noted that several peaks of the L1
2 phase are located very close to those of L1
0 making difficult their detection since they are masked each other. According to the phase diagram of bulk CoPt, Co-rich L1
2 phase reaches its maximum ordering temperature at around 900 °C, for the optimum stoichiometry and decreases to 700 °C for Co
83Pt
17 [
50,
57,
61,
62]. Consequently, increasing annealing time at 700 °C, is expected to increase the proportion of Co-rich L1
2 phase. Regarding the crystallite size, obviously, the reflections peaks become sharper and more intensive as the annealing temperature increases. This indicates that the particles size increases with longer annealing times. In particular, the size of the particles derived by the Scherrer equation after annealing at 700 °C for 4 h and 7 h was estimated to 8.8 nm and 17.1 nm, respectively. These values are in good agreement with the sizes estimated from the TEM images.
Room temperature magnetic hysteresis loops of CoPt nanoalloys, in the as-made and annealed forms, have been performed in order to study their magnetic properties, and presented in the
Figure 4. The as-made CoPt nanoparticles exhibit 59.3 emu/g saturation magnetization, and a moderate coercivity (H
c) value, around 1 kOe (
Figure 4), which is significantly higher compared to the so-called “soft” magnetic materials where the coercivity is in the range of a few Oe. It is obvious that the as-made CoPt shows hysteretic ferromagnetic behavior which cannot be explained by the structural defects or imperfections (
pinning centers). Shape anisotropy should also be excluded since the particles are almost perfectly spherical and nearly monodispersed as it turns out from the TEM images. Meanwhile, XRD measurements showed that the as-made nanoalloys consist dominantly of a disordered fcc phase with probably the presence of small amount of cobalt oxide. A possible interpretation should be connected with the presence of compositional gradient in the nanoalloys which is also in agreement with the previously discussion in the XRD section, as well as, the reaction mechanism where the positive (+1.18 volts) Pt
2+ reduction potential, ensures the initial formation of tiny Pt seeds before the reduction of Co
2+ ions. Under the regime of such growth mechanism, it is expected that the nanoalloys core would be platinum-rich in contrast with the shell [
44]. In the case of annealed nanoalloys it is obvious from the hysteresis loops in
Figure 4 that the magnetization value does not correspond to the saturation value, as the 2 T applied field is not sufficient to saturate the magnetization. Therefore, the magnetization, at 2 T magnetic field, is 46.8, 43.6, 36.8, and 33.7 emu/g for the samples after annealing at 700 °C for 30 min, 2, 4, and 7 h respectively. The saturation values were also obtained by fitting the high-field data with the M=M
s(1-a/H) law (
Figure S2). Extrapolation gives M
s=36.4 emu/g, and therefore M
R/M
S= 26.25/36.4=0.72.
On the other hand, the room temperature coercivity reached 1.3 4.2, 9.2 and 14.5 kOe, after annealing at 700 °C for 30 min, 2, 4 and 7 h, respectively. Considering the unsaturated magnetization, it is expected that these values will be even higher. By examining the shape of the M vs H curves, it is obvious that there are two different magnetic phases in the materials annealed for shorter times. This behavior becomes less prominent after 4 h annealing, and almost disappears after 7 h annealing. For instance, in the sample annealed for 2 hours, a significant change in the slope of M vs H curve is observed, as indicated by the two local maxima of the derivative (
Figure S3). In fact, the two maxima suggest two different magnetization switching mechanisms which correspond to two different magnetic phases with two different coercive fields. This finding is in accordance with the conclusions from the XRD analysis, where different structural phases were identified. Indeed, the present system of nanoalloys is a mixture of cubic phases with low magnetocrystalline anisotropy, and ordered L1
0 tetragonal CoPt phases exhibiting the highest magnetocrystalline anisotropy and coercivity [
63]. As the annealing time increases the degree of ordering and the proportion of the L1
0 phases increases which results in enhancement of coercivity value. It is evident from the hysteresis loop that, in addition to the increase in coercivity, the M vs H curves become smoother similar to single-phase magnetic hysteresis loops. This behavior it is well known and based on the interactions between magnetically hard and soft phases and described by the exchange-spring magnets theory [
64,
65]. It has been proven that the higher difference in the magneto-crystalline anisotropy between the two phases the lower should be the thickness (or the proportion) of the soft phase in order to have completely coupled magnetic composite without steps at the demagnetization curve [
66]. Therefore, the optimum conditions for effective coupling are more favorable under the presence of phases with intermediate hardness, as in graded stoichiometry nanoparticles. In nanostructured materials that need high temperature annealing is very difficult to precisely control the dimensions of the hard phase and the soft phase in particular [
65]. In our case the presence of the Co
3Pt phase, despite being a cubic phase, exhibits a moderate magnetocrystalline anisotropy (K
u = 107 erg/cm
3) [
15], it appears to be crucial for the formation of a fully exchange coupled spring magnet with a single magnetization switching field (H
c) as it’s progressive appearance with the annealing time eliminate completely the shoulder in the M vs H curve. The importance of the Co
3Pt phase formation, as well as the off-stoichiometric cobalt rich L1
0 phase, is also proved by the magnetic behavior of the nanoalloys when the annealing temperature slightly decreased. After annealing at 675 °C the magnetic hysteresis loops (
Figure S4), show that the presence of two different magnetic phases (a soft and a hard), which is much more intense compared to the sample annealed for the same time at 700 °C. The inefficient coupling, may be attributed to the lower L1
0 ordering, although we are convinced that it is also linked with the absence of the intermediate magnetocrystalline anisotropy Co-rich phase. Additionally, the remanence to saturation ratio (M
R/M
S) is enhanced well above the 0.5 value which is expected for an isotropic sample. This indicates the presence of strong interactions.
Furthermore, it is worth to mention that, by synthesizing equiatomic CoPt nanoalloys, i.e., following the same methodology, but using equimolar cobalt and platinum precursors, shows that after annealing at 700 °C, we obtain a single-phase ferromagnetic material with well crystallized face centered tetragonal phase, (L1
0), as presented in the
Figure S5, in the supplementary material section.