Spinel ferrite nanoparticles with general formula MFe2
(where M is a bi-valent transition metal ion such as Mn2+
, etc.) have been the object of intense investigation due to their interesting optical, magnetic and catalytic properties for their potential application in storage devices [1
], photo-catalysis [2
], magnetic fluids [3
], sensors [4
] and biomedicine [5
]. In particular, the introduction of copper in the general spinel ferrite crystal structure gives rise to a soft magnet, which has been deemed an interesting candidate for multiple applications in many diverse fields, ranging from sensors to catalysis [6
]. This has led to a large number of studies in recent years, focusing on attaining an always increasing degree of control over the shape, size, distribution and phase of the copper ferrite nanoparticles. In fact, copper ferrite represents a special case among ferrites, since it displays two distinct crystalline phases, namely a high temperature disordered cubic Fd-3m
) and an ordered tetragonal I4I/amd
) that can be obtained during slow cooling in air [9
]. The tetragonal phase arises from the collective Jahn–Teller distortion along one of the axes of the octahedral sites, which is typical for Cu2+
ions (d9) as a consequence of the removal of the eg
orbital degeneracy [10
]. For this reason, many studies have been focused on synthesizing copper ferrite nanoparticles with a wide number of methods including thermal decomposition [12
], hydrothermal [13
], solvothermal [14
], co-precipitation [15
], electrospinning [16
] and sol-gel [17
], also taking into account that the desired crystalline phase might be obtained by selecting an appropriate thermal treatment. Moreover, in a recent report it has been shown that, independently of the final heat treatment, the cubic phase and tetragonal phase can be obtained by employing sol-gel and co-precipitation methods, respectively [18
]. All the aforementioned synthetic methods yield nanoparticles with dimensions of several tens of nanometers and a poor degree of size homogeneity, mainly due to the aggregation and sintering of the nanoparticles during the thermal treatments needed to obtain the desired crystalline phase. In turn, this also affects the magnetic properties and makes it difficult to achieve a precise control over the different parameters at play in giving rise to the final magnetic behavior [19
A fruitful strategy to produce crystalline magnetic nanoparticles with a controlled dimension at the nanometer range and to control their spatial distribution and aggregation relies on embedding them into a suitable support or matrix capable of keeping them durably dispersed, thus limiting interparticle magnetic interactions. [20
] As a consequence, the magnetic response of the resulting magnetic nanocomposites is due to the degree of control over the monodispersity (in size, shape, and phase) of the nanoparticles population, and to the loading and dispersion of the magnetic phase within the matrix. In particular, magnetic interactions and hardening (or softening) of the magnetic behavior of the nanocomposite is directly influenced by the loading and dispersion of the magnetic phase within the matrix. The capability of obtaining small-sized nanoparticles is of interest because it guarantees that the whole magnetic phase will comprise single domain nanoparticles, i.e., nanoparticles that act as one single magnetic domain under an applied magnetic field, not dividing their atomic moments along different orientations and thus maximizing their overall magnetization.
In this framework, our group has successfully applied a sol-gel route to produce Mn, Zn, Co, Ni, and Cu ferrite nanoparticles which have been crystallized within a porous silica matrix [20
]. The sol-gel technique offers the possibility to obtain different forms of porous materials, such as xerogels and aerogels, by adopting specific procedures to obtain the dry gel. It was found that the crystallization of the ferrite nanoparticles within a porous silica matrix represents a valid method to obtain nanocomposites with a homogeneous distribution of the nanoparticles within the matrix, and to limit size growth. In particular, xerogel and aerogel nanocomposites with 10 wt% loading of copper ferrite enabled us to gain insights on the structural properties of CuFe2
nanocrystals, such as the cation distribution within the crystal structure, using X-ray absorption spectroscopy [26
]. Interestingly, this study showed that the Jahn–Teller distortion is present, which is consistent with the formation of the tetragonal phase. However, the distortion only interests the copper environment while the iron environment is the same of the disordered cubic phase. Here, we study the magnetic properties of CuFe2
nanocomposites obtained by sol-gel synthesis in the form of aerogels and xerogels by either supercritical or conventional drying, respectively. The nanocomposites were prepared with ferrite loading ranging between 5 wt% and 15 wt% and submitted to thermal treatments at increasing temperature up to 900 °C. This choice has allowed us to investigate two aspects that come into play in giving rise to the magnetic response of the nanocomposites, namely the presence of small magnetic nanoparticles below the single domain size threshold, and their distribution according to different loadings within matrices with two kinds of porosity.
To perform this study, CuFe2O4-SiO2 aerogel and xerogel nanocomposites with ferrite loading of 5 wt%, 10 wt% and 15 wt% were synthetized and investigated from a structural and magnetometric point of view, to assess the effect of different matrices on the CuFe2O4 nanoparticles and on the net magnetic features of the nanocomposites.
3. Results and Discussion
In Figure 1
, the powder XRD patterns of the aerogel samples are reported, whereas the XRD patterns of the corresponding xerogel samples are reported in Figure 2
. In Figure 1
a–c, the XRD patterns compare the evolution of the copper ferrite crystalline phase within the aerogel matrix, as a function of the calcination temperature for any given loading, whereas in Figure 1
d, the XRD patterns of the three different loadings are compared after the same thermal treatment at 900 °C. In the case of the 5 wt% nanocomposites (Figure 1
a), the XRD pattern is dominated by the presence of the amorphous SiO2
phase, highlighted by the typical halo centered at ~22° recurring in all the XRD patterns. Only in the case of the sample treated at 900 °C, some very broad reflections appear, the most intense one centered at ~35.5°, which can be assigned to the formation of either the tetragonal or the cubic crystalline phase of the CuFe2
In the case of the aerogels with a 10 wt% and 15 wt% of dispersed phase (Figure 1
b,c), some peaks are also detectable in the samples treated at 450 °C and 750 °C, providing some insights on the formation of CuFe2
nanoparticles within the aerogels. In particular, peaks that can be ascribed to CuO [29
] are visible in the samples treated at 450 °C, especially in the aerogel with 15 wt% of dispersed phase. As previously found in the sol-gel synthesis of other ferrites dispersed in silica aerogels [30
], iron is very likely present in the form of ferrihydrite, whose peaks are hidden within the amorphous silica background because of the poor crystallinity of this phase. When the same samples with a 10 wt% and 15 wt% dispersed phase are submitted to 750 °C some additional peaks appear, which further evolve with increasing the temperature of the thermal treatment. Peaks due to either tetragonal or cubic CuFe2
increase progressively while the peaks due to CuO tend to progressively disappear. Moreover, a peak centered around 33° appears with thermal treatment at 750 °C and then tends to disappear with thermal treatment at 900 °C, indicating the formation of an intermediate phase. Based on the peak position and the composition of the sample, it seems very likely that this peak is due to some hematite forming as intermediate phase from ferrihydrite [31
]. The XRD patterns of the samples treated at 750 °C for 1 h and 6 h seem very similar, which proves that thermal treatments longer than 1 h do not induce further crystallization and/or evolution of the initial and intermediate phases. The differences in terms of loading of the crystalline phase are highlighted in Figure 1
d, where the intensity of the reflections corresponding to the CuFe2
increases with loading, as was expected.
It should be noted that in the case of the aerogels with a 10 wt% and 15 wt% dispersed phase, two small peaks due to some unreacted α-Fe2O3 and CuO are still detectable together with the peaks of CuFe2O4, which appears to be the predominant phase.
The XRD patterns corresponding to the xerogel samples (Figure 2
a–d) show a few differences in the evolution with the thermal treatment, with respect to the aerogel samples. Apart from the typical silica halo, no peaks are detectable in any of the samples treated at 450 °C, regardless of the composition, indicating that very poorly crystalline phases must be present at this stage. When the samples are treated at 750 °C, some peaks appear that become more evident as the loading increases. These peaks are ascribed to the formation of either tetragonal or cubic CuFe2
and further evolve with increasing the temperature of the thermal treatment. The main difference between the xerogel and aerogel samples treated at 900 °C is the width of the peaks, which is larger for the xerogel samples, indicating smaller crystallite sizes. Moreover, in the case of the xerogel samples, all the detectable peaks are due to CuFe2
, and no sign of CuO and/or hematite is visible. As mentioned above, the XRD patterns of the nanocomposites do not allow to distinguish between the tetragonal and the cubic crystalline phase of CuFe2
due to the nanocrystalline nature of the samples that generates broad reflections and the XRD patterns of these phases sharing most of the peaks. However, an X-ray absorption investigation on related samples suggested that, under the adopted conditions, CuFe2
nanoparticles crystallize in the tetragonal phase [26
]. The crystallite sizes for the samples thermally treated at 900 °C are: 3, 6 and 9 nm for the 5 wt%, 10 wt% and 15 wt% aerogel nanocomposites, respectively, and 3, 4 and 5 nm for the 5 wt%, 10 wt% and 15 wt% xerogel nanocomposites, respectively. Although the sizes increase as a function of the loading, the effect is slower in the xerogel samples versus the aerogel ones.
In Figure 3
and Figure 4
, bright-field (BF) and dark-field (DF) images for the aerogel and xerogel samples treated at 900 °C are shown. The images provide clear evidence that the nanoparticles are well dispersed in the SiO2
matrix, which is hindering their agglomeration and growth. Their presence can be more clearly observed in the DF images, where the nanoparticles appear as bright dots over a darker background, except for the samples with 5 wt% loading, due to the low amount of ferrite, combined with the very small size of crystallites (3 nm for both samples according to Scherrer calculations).
The images also show a finer and denser texture for the xerogels, as compared to the aerogels, where a more open texture is evident, as was expected [32
]. In fact, aerogels obtained via specific gel drying techniques exhibit an open porous network presenting a significant mesopores contribution. On the other hand, xerogels production implies partial collapse of the pores and generates a relatively denser matrix with smaller pores. The size of the nanocrystals increases with ferrite loading and is slightly larger in the aerogel as compared to the xerogel, in agreement with the average nanocrystal size as determined by XRD, likely due to the different matrix characteristics.
DC magnetometry was used to investigate the effect of the ferrite loadings on the overall magnetic properties of the aerogel and xerogel nanocomposites treated at 900 °C, as XRD analysis provides the clearest evidence that nanocrystalline CuFe2
is the most prevalent phase for the samples treated at this temperature. All the relevant parameters are reported in Table 1
and Table 2
for aerogels and xerogels, respectively.
The three aerogel samples are all magnetically unblocked at room temperature, as shown by ZFC-FC curves (Figure 5
a–c) with different TB
values that tend to increase together with the loading of magnetic phase and crystallite size. Similar trends, i.e., the variation of magnetic parameters with the loading of magnetic phase, can also be appreciated in the hysteresis loops recorded at 5 K (Figure 5
d–f). Both the magnetic hardness and the maximum measured magnetization, indicated by coercivity HC
, respectively, are affected by the quantity of magnetic phase present in the SiO2
matrix. These trends point at a progressive hardening with the increase of Cu ferrite, while the lowered magnetization is consistent with the presence of different crystal phases, as already indicated by the traces of α-Fe2
and CuO observed in the XRD patterns, that are expected to contribute negatively to the net magnetization due to their antiferromagnetic nature. Also, even considering the slight variation in terms of crystallite size between the two samples with highest loadings, the combination of lower magnetization and higher coercivity in the 15 wt% sample with respect to the 10 wt% suggests that a lower number of magnetic moments are available for the reorientation under an applied magnetic field (hence the lower MS
) and that the said reorientation requires higher applied magnetic fields (hence the higher HC
This finding is in good accordance with the presence of spontaneous exchange bias (HE
) in the hysteresis loops of the 10 wt% and 15 wt% aerogel samples, whose presence usually indicates structural disorder that depletes the number of available magnetic moments and affects negatively the net magnetization of the samples [33
On the other hand, the low coercivity, higher magnetization and null exchange bias (HE) observed for the 5 wt% sample can be explained by considering the 3 nm size of the crystallites, which puts them in the lower portion of the single domain regime. Such small-sized crystallites correspond to small single magnetic domains, which are easily aligned to applied magnetic fields and undergo superparamagnetic relaxation at very low temperatures, as shown by the low TB and TIRR values observed in the ZFC-FC curves in the present case. Also, no impurities were observed in the corresponding XRD pattern and in this range of sizes no clear distinctions between volume- and surface-based effects should be expected, so the absence of HE and the high values of magnetization can be expected.
The magnetic features of the xerogel nanocomposites present some differences from their aerogel counterparts and should be compared in full, taking into account the differences in crystallite size occurring among xerogels as a function of the loading, and between aerogel and xerogel samples with the same loading, as well as the absence of the CuO and α-Fe2
impurities, whose presence was only evidenced in the aerogel samples (Figure 6
). At any given loading, blocking and irreversibility temperatures TB
and coercivity HC
have smaller values in the xerogels than in the aerogels. As for the aerogel samples, the lowest percentage of magnetic phase (5 wt%) corresponds to the smallest crystallites, with the xerogel sample having the same average size of the aerogel sample and forming small single magnetic domains that can be easily reoriented along with the external magnetic field. Additionally, spontaneous exchange bias HE
is negligible in all the xerogel samples, suggesting a low degree of structural disorder. Since structural disorder can affect the symmetric shape of the hysteresis loops and modify coercivity HC
and maximum magnetization values M5T
, the minimum HC
and maximum M5T
values observed in xerogel samples are all consistent with the fact that no impurities were observed by XRD in these samples.
The crystal size of the xerogels with 10 wt% and 15 wt% magnetic phase increases slightly with the loading, and it is safe to assume that the magnetic domains follow a similar trend. All the nanoparticles in xerogels also seem less affected by interparticle interactions with respect to the bigger-sized nanoparticles of the aerogels, as indicated by the lower blocking and irreversibility temperatures TB
observed in the ZFC curves and by their coercivity HC
. At any given loading, smaller values of all these parameters are observed in the xerogel samples compared to their aerogel counterparts, while their values grow with the loading. For non-interacting nanoparticles, variations in TB
values are expected, since bigger sizes require increasingly higher temperatures to overcome their blocked state and undergo superparamagnetic relaxation. However, the differences in size between the crystallites of aerogel and xerogel samples cannot solely account for the huge variation observed in the shape and parameters of the ZFC-FC curves. For these reasons, a large effect should be attributed to magnetic interactions affecting the aerogels since these generally work against the easy reorientation of the magnetic moments along an external magnetic field and manifest as a tendency towards magnetic hardening (i.e., higher coercivity) in the hysteresis loops and a broadening in the TB
peak of the ZFC curves. The former can be mostly attributed to intraparticle interaction-related structural disorder, while the latter is mostly driven by interparticle interactions [37
The presence of magnetic interactions in the aerogels can be further proved by comparing the magnetic response of nanoparticles of similar size (6 nm and 5 nm, respectively), present in samples A_10_900_1 and X_15_900_1. Even in this case, despite the lower loading of the aerogel sample A_10_900_1 and the higher defectivity of its crystallites, indicated by the non-null HE value, its TB, TIRR and HC values are more than four times higher than those of X_15_900_1, indicating that interparticle magnetic interactions modify the magnetic response of the aerogel samples.
Thermal treatment is known to trigger a progressive reordering of the Cu2+
ions in the octahedral and tetrahedral sites, which is proportional to the temperature and clearly modifies the magnetic characteristics of copper ferrite towards soft magnets already at 700 °C [38
]. In fact, the 900 °C-treated samples are all characterized by higher values of saturation magnetization and lower coercivities than those observed in nanoparticles of similar size that did not undergo post-synthetic thermal treatments [40
]. Overall, our DC magnetization studies confirm that the single-domain magnetic nanoparticles included in xerogel matrices are systematically smaller than in their aerogel counterparts and have a lower degree of superficial defectivity. Furthermore, nanoparticles in the xerogels better approximate the typical magnetic response of non-interacting nanoparticles, while magnetic interaction heavily affects the magnetic response of the nanoparticles included in aerogel matrices. In principle, although this difference might be the indication of a higher degree of dispersion of the magnetic nanoparticles in xerogel matrices, given the direct evidence provided by TEM images of the successful homogeneous distribution of the magnetic nanoparticles in both aerogel and xerogel matrices, it should be more likely ascribed to a matrix-based effect on the nanoparticles, which are also influenced by the presence of impurities in the aerogel samples that increase structural disorder in the nanoparticles.
When comparing xerogel and aerogel samples, in addition to the differences in the TB
values from the ZFC-FC curves, a striking dissimilarity can also be observed with regard to the magnetic hardness in the hysteresis loops recorded at 5 K, i.e., a low temperature where all the samples are in a magnetically blocked state. Here, even taking into account the fact that coercivity increases with the size of the magnetic domain, aerogel samples A10_900_1 and A15_900_1 display coercivity values that are more than five times higher than those of xerogel samples X10_900_1 and X15_900_1 with the same loadings, while also showing non-negligible spontaneous exchange bias HE
, which is absent for the xerogel samples. Exchange bias manifests in hysteresis loops as a horizontal shift and has been observed in nano-sized compounds with a high degree of superficial structural disorder. Here, the structurally disordered surface of the nanoparticles becomes heavily prone to phenomena such as spin-canting and behaves differently from the structurally ordered core. Then, the system as a whole reacts to applied magnetic fields as a makeshift magnetic core/shell, whose components undergo magnetic coupling during hysteresis loops, giving rise to horizontal shifts that manifest as a difference in the moduli of the negative and positive measured coercive fields, |HC1
| and |HC2
|, respectively. When the horizontal shift is observed in zero field-cooled hysteresis loops (such as in the present case), it is usually indicated as spontaneous exchange bias (SEB) [33
Considering the small size of the magnetic nanoparticles of all the aerogel and xerogel samples, a single domain behavior can be expected in all the nanocomposites, but taking in great care the indications provided by DC magnetometry and XRD, some additional assumptions can be made on both classes of samples. XRD indicates that the nanoparticles of A10_900_1 and A15_900_1 are bigger than those of X10_900_1 and X15_900_1, and also shows the presence of a few low intensity peaks that are attributed to impurities of CuO and α-Fe2O3. While the formation of small impurities is not expected to revolutionize the overall magnetic response of A10_900_1 and A15_900_1, their presence paired with increased magnetic hardness, spontaneous exchange bias and higher unblocking temperatures suggest that the magnetic nanoparticles included in the aerogel matrices tend to be more structurally disordered and that the presence of impurities might be located at the nanoparticles’ surface, thus increasing the superficial disorder and consequently giving rise to a patched makeshift shell from the structural point of view. If one considers these “composed” nanoparticles, it is clear why they require higher temperatures to undergo superparamagnetic relaxation and can reach lower saturation magnetization: the structural disordered surface lowers the number of atomic moments that can be successfully reoriented under an external magnetic field, while opposing the thermal unblocking and the re-orientation of magnetic moments from the ordered “core” region. On the other hand, the lower unblocking temperatures and the narrow, symmetric hysteresis loops indicate that the xerogel samples contain smaller, well-formed crystal domains that do not show the disorder-related responses observed in aerogel samples and give rise to nanocomposites with a softer magnetic response and higher saturation magnetization.
Even considering the varied effect of the xerogel and aerogel matrices on the final magnetic features of the nanocomposites and the minor impurities observed in the aerogel samples, the controlled inclusion of copper ferrite nanoparticles in both xerogel and aerogel affects their magnetic features and distances them from the magnetic response of a typical population of similar sized unsupported nanoparticles [40
]. In a comparison between evenly sized unsupported nanoparticles and nanoparticles included in nanocomposites, the former are magnetically harder, with higher exchange bias and tendentially lower saturation magnetization values. This means that the inclusion of copper ferrite nanoparticles inside aerogel and xerogel matrices helps in controlling their size and maintaining it well within the single domain regime, lowers the interparticle interaction and consequently boosts the saturation magnetization, while restoring the soft magnetic features typically observed in copper ferrite.