Superspinglass state in functionalized zeolite 5A-maghemite nanoparticles

Magnetic nanoparticles (NPs) are prominent materials, which can be used in many areas of current applied science.^1–6 In biosensing, they are called nanozymes, where their catalytic properties enhance the electrochemical response of the sensor, for example, in the detection of viruses such as SARS-CoV-2 or Ebola.^1–3 Their conjugation with zeolites, forming composite solid matrices with magnetic NPs (nanohybrids), is also promising materials in the removal of heavy metals and nanoagriculture for improving crops’ and plants’ quality.^4–6 Therefore, magnetic interactions and the magnetic state of an ensemble of magnetic NPs are subjects that still need deeper investigations for a qualified application of the developed nanohybrids in different areas.

Specifically, by regulating magnetic NPs’ synthesis, it is possible to handle their surface and magnetic properties, thereby allowing predictable tuning of their biocompatibility, toxicity, and adsorption properties.^7,8 To synthesize pure magnetite (Fe₃O₄) and maghemite (γ-Fe₂O₃) NPs, several physical and chemical methods have been proposed.^8,9 Among the physical methods, ultrahigh energy milling and laser pyrolysis should be pointed out. On the other hand, chemical methods, such as the coprecipitation method and thermal decomposition, are comparatively more advantageous because they produce NPs with high purity (99.9%), monodispersity, and shape-controlled morphology.⁹ Furthermore, green biosynthesis is an emerging promising method to obtain magnetic NPs.^10,11 The magnetic NPs synthesized by this method are often functionalized with the reductant agent, such as polyphenols.¹⁰ However, size-controlled effects and reduction in saturation magnetization values, caused by these reductants, have not been studied in detail, and also, there is the question of scalability for industrial levels that is still in development, requesting additional systematic studies.^10,11 Moreover, chemical procedures of coating the NPs with organic agents or by dispersing them in mesoporous matrices lead to new and compelling magnetic effects, for example, exchange bias or superspinglass (SSG) phase transitions.^12–14 

A problem reported with the coprecipitation method is the fast oxidation that often occurs in Fe₃O₄ synthesis even in functionalized samples,¹⁵ precluding a reliable conclusive assignment of the oxide as Fe₃O₄ or γ-Fe₂O₃. Because of their similar x-ray diffraction (XRD) patterns, similar infra red (IR) vibration modes, and similar hyperfine parameters when experiments are done at room temperature (RT),^15–18 the two iron-oxide phases cannot easily be differentiated using standard experimental procedures. Consequently, this issue leads to frequent discussions concerning the proper assignment of a specimen assumed as Fe₃O₄ without deeper magnetic analysis. In this respect, Mössbauer spectroscopy can give essential clues via the distinct characteristic nuclear hyperfine parameters such as magnetic hyperfine field (B_hf) and isomer shift [center shift (S) of spectra] of bulk γ-Fe₂O₃ and Fe₃O₄. Yet, the situation for functionalized particles with ultrasmall sizes or USPIOSs (Ultrasmall Superparamagnetic Iron Oxides) is less simple. Their Mössbauer spectral shapes are usually complex due to the magnetic relaxation effects, precluding an identification of the typical magnetic hyperfine spectra of nano-Fe₃O₄ and nano-γ-Fe₂O₃ at RT.^15–18 Thus, it makes necessary Mössbauer measurements at very low temperatures to yield a reliable assignment for the desired application and sometimes in-field Mössbauer experiments as well.

To the best of our knowledge, the interactions and the functionalization mechanism between the zeolite 5A cage and magnetic γ-Fe₂O₃ NPs have not been studied until now. In addition, the magnetic state of the ensemble of γ-Fe₂O₃ NPs functionalized with zeolite 5A in a magnetically blocked regime is still an open issue. In a previous study,⁴ we have reported, in detail, the structural, morphological, and Pb(II) adsorption properties of NPZEO systems. In this paper, we are reporting, for the first time, the vibrational properties, which give insights into the functionalization mechanism of the zeolite 5A and γ-Fe₂O₃ NPs through the hydrated surface chemical groups of zeolites. We also studied the development of static and dynamic magnetic effects of these magnetic NPs by following the temperature dependence of Mössbauer relaxation spectra and AC magnetic susceptibility under different frequencies. From details of the Mössbauer spectral shape together with AC susceptibility data, we have concluded that the NPs are magnetically interacting and that a SSG state is established in the three samples, i.e., we have shown that the SSG state found in our γ-Fe₂O₃ NPs is a consequence of the functionalization with the zeolite type 5A since pure γ-Fe₂O₃ NPs synthetized by co-precipitation did not present such magnetic properties.^7,14–16

The synthesis of the nanocomposites (nanohybrids) has been described in a recent study.⁴ Briefly, three samples were synthesized by the coprecipitation method. The balanced chemical reactions are expressed as follows:

Equations (1) and (2) were carried out in the presence of 0.5 g of zeolite type 5A (it will be used hereafter as zeolite 5A) for the NPZEO1 and NPZEO2 samples with a molar ratio of $Fe2+Fe3+∼$ 0.5. For the NPZEO1 synthesis, the zeolite 5A was dispersed in ultrapure water for 30 min. Thereafter, amounts of 6 g of FeCl₃ (37 mmol) and 5.1 g of FeSO₄ · 7H₂O (18.5 mmol) were added to the dispersion and the alkaline medium pH (10) was adjusted by using NaOH (1.5M), and for the NPZEO2 synthesis, a NH₄OH (30%) solution was dropped to the iron salt dispersion as a precipitator. The chemical reactions were kept at 80 °C for 6 h. Then, the reaction was left to cool and the solid separation was done with a magnetic decantation assistant. For the NPZEO3 sample, 2.5 g of zeolite 5A was dispersed in ultrapure water for 30 min, and then the γ-Fe₂O₃ NPs (1.6 g), synthesized by coprecipitation, were added to the dispersion at RT for 24 h (pH = 7). After that, the dispersion was filtered with a membrane (2 µm) to remove the excess of NPs (not precipitated or/and interacted) over the zeolite 5A matrix. Finally, the three samples were dried in an oven for 36 h under normal conditions.

XRD data and Transmission Electron Microscopy (TEM) images (previously published in Ref. 4) were performed, respectively, in a diffractometer Bruker D8, operating with CuK_α radiation, and a JEOL-JEM2100 microscope. Fourier Transform Infrared (FTIR) spectra were taken using a Varian EXCALIBUR SERIES 3100-UMA 600 equipment, using a transmission mode with 4 cm⁻¹ of resolution.

⁵⁷Fe Mössbauer absorption spectra have been collected in transmission geometry using a standard spectrometer with sinusoidal velocity sweep. The powder absorbers were enclosed into nylon containers. Absorber thicknesses were chosen equivalent to ca. 0.1 mg ⁵⁷Fe per cm². Absorber temperatures were varied between 20 and 300 K using a variable temperature He-flow cryostat (Cryovac). As a 14.4 keV γ-radiation source, we used 20 mCi ⁵⁷Co in a Rh matrix that was kept at RT during all experiments.

AC susceptibility measurements were performed in a Physical Property Measurement System (PPMS) equipment from Quantum Design. A range of frequencies of 100 Hz–10 kHz with an oscillating probe field of 10 Oe in a temperature region of 2–300 K were used (in the case of the NPZEO1 sample, experiments with a superposition of a dc field of 300 Oe were also done).

As previously demonstrated by XRD experiments,⁴ only Bragg diffraction peaks for the γ-Fe₂O₃ and zeolite 5A crystalline phases were observed. The mean crystallite sizes of the γ-Fe₂O₃ phase, found by Rietveld refinement, were 6.6(2), 10.5(2), and 15.0(3) nm for the NPZEO1, NPZEO2, and NPZEO3 samples, respectively, while the TEM analysis has revealed sizes of 8.9(4) and 8.7(4) for the NPZEO1 and NPZEO2 samples and 9.4(5) nm for the NPZEO3 sample.⁴ The difference in the crystallite and particle size can be explained on the basis of the polydispersity index (PDI) values. For our TEM analysis, we got values higher than 0.4,¹⁹ which is representative of a broad particle size distribution, as typically expected of magnetic NPs synthesized by coprecipitation. Then, despite we have high values of crystallite sizes, we must consider that the real particle sizes are ranging from 3 to 20 nm, as suggested by TEM data.

On the other hand, it is important to highlight that the sizes for the zeolite crystallites are similar (∼100 nm), suggesting that the chemical reaction does not significantly affect the zeolite composition and chemical reactivity of their total specific surface areas, an important result considering that these nanohybrids can be used in the magnetic remediation process.

Figure 1(a) shows the IR bands for the pure zeolite 5A sample. In the spectral region of 3600–2500 cm⁻¹, the zeolitic Z-OH or Si(OH)Al IR bands are found at 3429 cm⁻¹;²⁴ the other two bands are assigned to Si–OH and –OH hydroxyl groups.²⁰ At 1668 cm⁻¹, the bending vibration related to the physisorption of H₂O groups is observed.²¹ 

In the low IR region (between 1500 and 400 cm⁻¹), the characteristic bands of the zeolite 5A are given at²² (i) ∼1001 cm⁻¹: asymmetric stretching vibrations of the bridge bonds—ν_as Si–O(Si) and ν_as Si–O(Al), (ii) 665 cm⁻¹: symmetric stretching vibrations of the bridge bonds—ν_s Si–O–Al, (iii) 554 cm⁻¹: (complex band) symmetric stretching vibrations of the bridge bonds—ν_s Si–O–Si and bending vibrations—δ O–Si–O, and (iv) 464 cm⁻¹: bending vibrations—δ O–Si–O, occurring in “antiphase.” For the NPZEO1, NPZEO2, and NPZEO3 samples, the IR spectra, shown in Figs. 1(b)–1(d), display the presence of zeolite 5A in all materials, confirming that it has not dissociated in the coprecipitation chemical reaction during the NPs’ synthesis. However, the low zeolitic IR strong bands are screened by the Fe–O bonds. The band at ∼580 cm⁻¹ is typically obtained for the tetrahedral sites of the γ-Fe₂O₃ phase.^23,24 These results are in agreement with previous Selected Area Electron Diffraction (SAED) patterns and Rietveld analysis, where interplanar atomic distances of d₍₆₂₂₎ = 3.7 Å and d₍₂₂₂₎ = 7.1 Å were calculated.⁴ Briefly, the vibrational properties are important to elucidate if the γ-Fe₂O₃ NPs are anchored/functionalized to the zeolite cage via aluminosilicate functional groups since silica layers are well-known to be conjugated and may modify the magnetic NPs’ properties.^25,26 So, in a first scenario, it is possible that, in the NPZEO1 and NPZEO2 samples, there is a coordination between Fe–OH and Si–OH groups that favors the functionalization process, as proposed by Singh et al.²⁷ The NPs are not located onto zeolite pores because the zeolite 5A has an average micro-pore width of 0.5 nm⁴ and it is smaller than the mean particle size obtained by a coprecipitation method. A pictorial representation for the synthesis procedure, functionalization mechanism, and TEM images for real morphologies configurations are given in Figs. 2(a)–2(d). It is also possible to observe a decrease in the absorption intensity of the IR spectra due to the higher content of iron-oxide in the NPZEO1 and NPZEO2 samples.

It is important to mention that the NPZEO1 sample has a secondary goethite phase with an ∼15 nm size, as estimated previously by the size strain-method.⁴ This phase is likely formed by the following mechanism: A possible variation of the theoretical molar ratio of Fe²⁺/Fe³⁺ ∼ 0.5 (for the spherical/pure γ-Fe₂O₃ phase) because of the presence of the zeolite, which holds hydroxyl and water groups, retains some iron ions and hence forms Fe(OH)₂ precipitates that also favor the goethite seed growth.²⁸ Here, we must point out that the zeolite 5A surface often retains a huge amount of water, as proved by atomistic simulations and previous studies.^20,21,29 From previous studies, we have observed that solid matrices affect significantly the coprecipitation method, giving rise to the formation of secondary-phases or even to functionalized surfaces with hydrated surface environments.^4,23 Moreover, the influence of the zeolite on the alkaline reactant properties used for the synthesis cannot be discarded, in this case, the NaOH and NH₄OH compounds. Previous studies have suggested that the OH⁻ radicals play a significant role in the sample reactivity, purity, size, and NPs’ morphology.^30–32 The presence of unbalanced OH⁻ may also be due to the zeolite cage, as explained above. This is confirmed by the high IR intensity of the Z-OH or Si(OH)Al, as seen in Fig. 1(d). These hydroxyl IR bands are screened/superposed by the goethite IR bands in the case of the NPZEO1 sample.³³ 

Selected Mössbauer spectra at 20 and 300 K are displayed in Figs. 3(a)–3(f) and for other temperatures in Figs. S1–S3 for the NPZEO1, NPZEO2, and NPZEO3 samples. From a first simple visual inspection, all samples reveal relatively well resolved though broadened magnetic patterns for the Mössbauer spectra recorded at 20 K with the outer sextet structures being typical for trivalent iron oxides.²⁴ As will be shown below, the main contributions of the spectra are due to the γ-Fe₂O₃ phase (A and B sites of the spinel structure), but the broad absorption lines are indicating a not completely magnetically blocked regime. The 300 K Mössbauer spectra show clearly that the entire ensemble of Fe-oxide NPs is not yet in a superparamagnetic regime since no complete collapse of the magnetic hyperfine splitting is observed. For the NPZEO1 and NPZEO2 samples, the first signs for a magnetically unblocked regime appear around 50 K and above, where a doublet structure emerges in the central part of the Mössbauer spectra (see data in Figs. S1 and S2). The situation is slightly different for the NPZEO3 sample (see Fig. S3). The spectra at low temperatures can be associated even clearer with γ-Fe₂O₃ (see the typical asymmetry of negative vs positive spectral parts). The onset of relaxation effects occurs at temperatures around 150 K, and at 300 K, the spectral contribution by fast fluctuating hyperfine fields is less than for the NPZEO1 and NPZEO2 samples. Thus, while for the NPZEO1 and NPZEO2 samples there is a clear separation of the more static and the more dynamic parts of the spectra at all temperatures, the distribution of relaxation frequencies for the NPZEO3 sample is apparently more continuous, as can be seen in the central part of the spectra (see all data shown in Figs. S1–S3).

For a more quantitative interpretation of the spectra, one has to observe a number of complications. There are (i) contributions from iron in various positions (core and surface of NPs, different crystalline phases) having different hyperfine parameters, leading to the inhomogeneous broadening of the spectra, as experimentally observed, and (ii) the Néel-type of magnetic fluctuations of magnetic moments of NPs, causing the relaxation patterns, will not be homogeneous, but will present a distribution of fluctuation frequencies due to distributed particle sizes and anisotropies. For treating the magnetic hyperfine field fluctuations in small particle systems, various models have been introduced in the scientific literature, e.g., superferromagnetism/superspinglass SFM/SSG models³⁴ and multilevel relaxation (MLR) models.³⁵ While MLR does not include interparticle magnetic interactions, this is the case for the SFM/SSG approach. Therefore, the complex spectral patterns of our samples rather preclude the application of these kinds of analyses. We have decided to use a purely phenomenological approach (though also hampered by most of the above quoted problems) for separating the faster relaxing inner parts from the more slowly relaxing better resolved outer parts of the spectra. In this way, a qualitative discussion of the fast relaxing and a more quantitative one of the slowly relaxing spectral contributions is expected.

For the data analysis, we used the MossWinn 4.0i software.³⁶ The partial spectra, with well resolved outer lines, were fitted to two sextet patterns with Gaussian broadenings σ of magnetic hyperfine fields B_hf (the parameters are listed in Tables I, II, and S1). For all three samples, the found center shift S and magnetic hyperfine fields B_hf for the two sites are close to those expected for the stoichiometric and crystalline γ-Fe₂O₃ phase.³⁷ The absorption area ratios of tetrahedral (A) and octahedral (B) sites of the spinel structure of the γ-Fe₂O₃ phase were kept fixed to the ideal values of 3:5. We associated these spectral contributions with the particle cores. For the fits, we needed in addition, at least, two relaxation sextet patterns (mrelax), especially for reproducing the strongly broadened inner spectral parts with a center shift and magnetic hyperfine field typical for ferric iron in the octahedral oxygen environment, with vanishing nuclear electric quadrupole interaction (these spectra are further labeled “octahedral contribution”). We attributed these spectral contributions to non-stoichiometric shell and surface sites.

For the NPZEO1 sample, the goethite, admixture in the sample, was taken into account by a static sextet with the typical quadrupole interaction of about −0.25 mm/s at low temperatures. Above 150 K, its Gaussian broadening increases, indicating the onset of fast relaxation, which is then realized at 250 and 300 K. All fitted spectra of the three samples are presented in Figs. S1–S3.

For the relaxation patterns, we used the classical Blume–Tjon 2-level model, yet the peculiar spectral shape made necessary for allowing unequal jump rates from and to differently populated spin levels, respectively, which is an indicative for interparticle magnetic interactions.³⁸ A typical signature for the unequal populations is the pronounced symmetric “doublet” structure visible in the center of the spectra. It is due to the still unbroadened inner lines of a relaxing sextet and not by quadrupole interaction in a superparamagnetic regime (which should reveal a singlet and not a doublet pattern for the γ-Fe₂O₃ phase). Although there is some arbitrariness concerning the involved number of subspectra, we can use this approach for a qualitative interpretation of the relaxation with a rough estimation of fluctuation frequencies instead of a physically questionable distribution of static fields. The number of free parameters was kept to a minimum for reproducing the temperature dependent spectra.

Between 50 K up to about 200–250 K, the static or only slowly relaxing stoichiometric γ-Fe₂O₃ contributions of the NPZEO1 and NPZEO2 samples only reveal weak variations in the spectral area [see Figs. 4(a) and 4(b)]. With a goethite contribution in the NPZEO1 sample of about 15%–20% of area (Rietveld refinement of XRD data gave a phase percentage of 24% of goethite⁴), the stoichiometric amount is about 25%–30% of the oxide contribution. The remaining 50% are represented by two relaxation patterns with Fe³⁺ in octahedral symmetry (the octahedral contribution). A slight variation in the hyperfine parameters is found for T > 200 K that can be related to a particle size distribution of antiferromagnetic goethite NPs, which are experiencing overbarrier fluctuations at these temperatures. For the NPZEO2 sample, the stoichiometric γ-Fe₂O₃ contribution has about 30%–40% and the octahedral contribution has 70%–60% of the total spectral area. We have to stress that a clear separation of the individual relaxation patterns of the octahedral contribution with reliable parameters is impossible. It is, however, clearly visible from the shapes that up to about 100 K the relaxation processes must be of underbarrier-type (rates ranging between MHz and 100 MHz), gradually changing to fast overbarrier-type (rates GHz and above), as can be traced from the development of the superparamagnetic singlet line contribution at higher temperatures.

As can be seen from Fig. 4(c), the fits for the NPZEO3 sample give a continuous decrease in the static/slow relaxing contribution of the stoichiometric γ-Fe₂O₃ component and a corresponding increase in the faster relaxing “octahedral” contributions. This is clearly demonstrating that for NPZEO3, the separation of the spectral contributions by stoichiometric γ-Fe₂O₃ and octahedral sites is not possible and, in this case, the “octahedral” contribution cannot be understood literally. The reason is an apparent wider distribution of relaxation frequencies affecting also the stoichiometric γ-Fe₂O₃ cores when compared with those for the other samples.

From our fits, it is possible to derive the development of the spectral areas [Figs. 5(a) and 5(b)] of the fast-relaxing overbarrier contributions with temperature. For the NPZEO1 and NPZEO2 samples, even at 300 K, only about half of the samples are in a superparamagnetic regime. According to the conventional estimate for blocking temperatures as the temperatures where paramagnetic and magnetically split spectral patterns have equal spectral weights, the blocking temperatures of these two samples are around 300 K or slightly higher.

For the NPZEO3 sample, the superparamagnetic singlet line is even not yet resolved at 300 K. This means that a “conventional” blocking temperature should also be above 300 K. A high lying blocking temperature can also be expected from the relatively weak temperature dependence of the relaxation rates that should reveal a strong variation near the blocking temperature. In fact, our AC susceptibility data also support our findings, as we will discuss in Sec. III D. These blocking temperatures are indeed high in view of very small particle sizes determined from the TEM and should be attributed to interparticle magnetic interactions via the zeolite 5A matrix (pure γ-Fe₂O₃ NPs, as synthesized by the chemical co-precipitation route, did not show this behavior^7,14–16).

As shown above, we could achieve a relatively clear separation of the stoichiometric γ-Fe₂O₃ core contribution with linewidths only moderately affected by inhomogeneous broadening. Recently, we have proved from in-field Mössbauer studies³⁹ that functionalized γ-Fe₂O₃ NPs, prepared by the coprecipitation method, tend to reveal stronger spin canting near the NPs’ surfaces for octahedral sites than tetrahedral ones. This has led us to the assumption that the extra octahedral contribution is due to Fe³⁺ atoms located at NPs’ surfaces, showing superparamagnetic underbarrier/overbarrier fluctuations. On the other hand, only a slight increase in the distribution of B_hf appears at and above 150 K. Therefore, it is possible to follow the development of coherent underbarrier fluctuations of these γ-Fe₂O₃ NPs. From the temperature dependence of the mean magnetic hyperfine fields of the γ-Fe₂O₃ NPs at low temperatures, we estimated the anisotropy constant of the particle cores using the method proposed by Mørup et al.^17,34

The hyperfine magnetic fields (B_hf) of site A (and consequently of the correlated B) show a linear dependence,

where the slope is expressed as

Turning Eq. (4) as a function of the mean TEM diameter, we have

where k_B = 1.38 × 10⁻²³ (J K⁻¹) is the Boltzmann constant, K_eff is the effective magnetic anisotropy constant, and V is the mean nanoparticles’ volume.

Figure 6 shows a reduction of ∼17% in the B_hf value at RT. In addition, the slope (m) and K_eff values, obtained from linear fitting, using Eqs. (3) and (5), are reported in Table S2 for the NPZEO1, NPZEO2, and NPZEO3 samples. The value of K_eff is one order higher than reported for the bulk material (4.3 × 10³ J m⁻³).²⁴ The change in the slopes and the obtained K_eff values (∼10⁴ J m⁻³) suggest the presence of strong interparticle magnetic interactions due to high magnetocrystalline anisotropy at the NPs’ surface.^13,25 These results are in agreement with those of the magnetic blocking temperature near RT and characteristic of broad intermediate relaxation components and the onset of spin-glass magnetic phase. These effects can be related with the maximum of the AC magnetic susceptibility curves, as we will discuss in Sec. III D.

Finally, to complement the in-detail description given above, it is worth noting that the mean center shifts have values of 0.50(2) mm/s (site B) and 0.28(2) mm/s (site A) for all samples and the hyperfine magnetic fields are characteristic of the γ-Fe₂O₃ phase.^16,21 Another feature of the presence of only γ-Fe₂O₃ is the apparent asymmetry between the left and right velocity sides of absorption.¹⁶ Nano-Fe₃O₄ does not exhibit this asymmetry, and extra resonant absorption lines are observed related to the Fe²⁺ ions, giving an asymmetry of the pattern in the opposite sense.¹⁸ In our case, this is not observed, confirming the presence of only γ-Fe₂O₃ in our samples (except the NPZEO1 sample that has also shown a fraction of goethite contribution). In addition, the mean center shift, S, of the Fe₃O₄ phase often has values S > 0.5 mm/s,^18,33 higher than those reported for our stoichiometric γ-Fe₂O₃ NPs in Tables I, II, and S1, really discarding the presence of this phase.

AC susceptibility studies allow identifying very sensitive magnetic phases, especially antiferromagnetic ones.⁴⁰ Core–shell magnetic systems or hybrid magnetic matrices can present combined effects, such as exchange bias, exchange spring, spin canting suppression, and SSG behaviors.^18,41–43 Therefore, to characterize the samples correctly, this technique is sometimes mandatory. Magnetic nano-ensembles often show a dynamic relaxation effect when the samples are exposed to an AC field, and magnetic relaxation is dominated by the Néel–Arrhenius formula.⁴⁰ In this case, an increment in the maximum temperature (T_M) position of the real susceptibility (χ′) or imaginary (χ″) signals is expected when increasing the frequencies values, as reported for several magnetic nanosystems.^13,14,32,42 In the NPZEO1 and NPZEO2 samples [see Figs. 7(a) and 7(c)], the χ″(T) component exhibits a broad curve with not apparent frequency dependence at high temperatures [see Fig. 7(a)]. Two marked aspects can be regarded: (i) The lack of T_M dependence is due to larger particles with a broad distribution, which is consistent with the major prevalence of static magnetic components due to ordered magnetic nanocrystallites (in agreement with Mössbauer fitting) and (ii) the peak at T_f ∼ 71–75 K [inset of Fig. 7(a)] in the χ′(T, f) and χ″(T, f) curves may also suggest the presence of SSG phase in both samples. Indeed, the same tendency was observed by Maldonado et al. in “magnetite” NPs.⁴¹ Roca et al.⁴² have assumed this “T₁”-type transition to disorder Fe₃O₄ with B-site vacancy concentration, but as we have already discussed, we discarded this assumption since low temperature Mössbauer spectra only suggest the presence of the stoichiometric γ-Fe₂O₃ NPs. Besides, this T_f value is in good agreement with the blocking temperature found by our relaxing components in the two-level fitting model, i.e., at about 100 K for the NPZEO1 and NPZEO2 samples (magenta components). It is worth mentioning that due to particle size distribution, we also expect distribution of relaxation times, as we formerly discussed above.

In a previous study,⁴ we found goethite as the secondary iron-hydroxide phase in the NPZEO1 sample, according to Bragg diffraction peaks. Nevertheless, no exchange bias field (shifting of the M-H loop along the field axis) was observed in the M-H loops, avoiding either the idea of a possible core–shell-like structure or an interface between goethite- γ-Fe₂O₃ where the exchange bias effect could be switched in FC magnetization experiments. Moreover, goethite has T_N ∼ 400 K, a temperature that our PPMS equipment cannot reach, but a noticeable increment in the χ″(T) curve is appreciated, showing a peak at high temperatures [see Fig. 7(a)]. Balanda⁴⁰ reported some protocols to study AFM transitions in magnetic systems and they are often observed as fixed and sharp pronounced peaks. However, in a combined system of AFM/FM nanocomposites, we can switch a small probe DC field of 300 Oe [Fig. 7(b)], for example, and this AFM peak is no more seen due to the dynamic behavior of the ferrimagnetic NPs, where the blocking of which depends on the applied magnetic field and AC frequencies.

In the NPZEO2 sample, no secondary magnetic phase was observed. The NPZEO1 and NPZEO2 samples do not allow the determination of the typical empirical parameter ∅ defined in Eq. (6) since the T_M value cannot be determined with accuracy. However, the χ″(T) curve is more defined for the NPZEO3 sample [see Fig. 7(d)], which, therefore, permits to determine T_M by fitting the curves using a log-normal distribution curve. Then, to identify the presence of the SSG phase, we calculated the ∅ parameter using the following equation:

We found a value of ∅ ∼ 0.09, which confirms the presence of an interacting SSG phase.¹³ We also used the slowing down relation,⁴⁰ 

The zv factor equal to 4.6(2) and τ₀ = 10⁻¹² s of characteristic time, as obtained from linear fitting of Fig. S4, using Eq. (7), also suggest the presence of SSG phase since the interval of the zv-parameter for this phase is 4 < zv < 12.⁴³ In contrast to the bare γ-Fe₂O₃ NPs, the NPZEO3 sample has a high T_f = 208(3) K, indicating that the functionalization of the γ-Fe₂O₃ NPs with zeolite 5A is responsible for changing severally the magnetic properties of the γ-Fe₂O₃ NPs. Also, we must recall that this sample was obtained by simple magnetic stirring between the zeolite 5A and the γ-Fe₂O₃ NPs. Therefore, once more it seems that the microporous zeolite 5A environment favors the SSG behavior in the case of bigger particles with broad particle size distribution. This result is in agreement with previous findings in the mesoporous NPSBA15 sample,¹³ where it was found T_f = 184 K. We finally point out that the SSG phase formation in the presence of zeolite 5A is attributed to a surface magnetic disorder caused by a reduction in the saturation magnetization value.¹³ In our case, octahedral Fe⁺³ atoms are strongly bound to the zeolite 5A chemical groups, as previously shown, and there is a high concentration of γ-Fe₂O₃ NPs in the zeolite matrix, favoring a dipolar-like magnetic interaction, which obviously will favor a SSG-like state. In addition, the SSG-like state is in accordance with the fact that the three samples have reduced values of magnetization, being 30 emu/gFe for the NPZEO1 and NPZEO3, whereas the NPZEO2 sample reports a value of 36 emu/gFe, as previously reported.⁴ 

γ-Fe₂O₃ NPs functionalized with zeolite type 5A were synthesized by the coprecipitation method. IR spectra have confirmed the presence of zeolite in all samples and have also suggested that the functionalization should occur through the hydrated surface environment in the zeolite cage. The synthesis was performed during and after the formation of the magnetic NPs, and according to the TEM analysis and PDI values, the NPs have presented a broad particle size distribution with size in the interval of 3–20 nm. A separation of static and relaxing superparamagnetic hyperfine patterns was achieved using a heuristic superposition of static and two-level relaxation patterns, allowing a qualitative description of the temperature dependent development of dynamic spectra. At 20 K, the hyperfine magnetic parameters have confirmed the presence of the pure trivalent γ-Fe₂O₃ phase in the samples. On the other hand, the static and dynamic magnetic parts of the γ-Fe₂O₃ NPs could also be differentiated by comparing Mössbauer data and AC susceptibility for the NPZEO1 and NPZEO2 samples. The temperature dependence (up to 150 K) of B_hf was studied by the modified linear Mørup’s relation for interacting magnetic particles that has allowed us to calculate K_eff of about 10⁴ J m⁻³. The values are, at least, one order of magnitude higher than the value found in bulk γ-Fe₂O₃, and hence, the high magnetocrystalline anisotropy has also confirmed the strong interparticle magnetic interactions present in our samples. The interparticle magnetic interactions are probably of dipolar type due to the fact that our samples have concentrated regions of NPs on the zeolite matrix, as suggested by TEM results. We also point out that the Fe⁺³ ions, in octahedral symmetry, are strongly bound to the zeolite 5A groups, and these complex interactions (magnetic among the NPs and bound with chemical groups of the zeolite 5A surface) have produced a SSG-like state. AC susceptibility studies have also shown a SSG formation in the three functionalized γ-Fe₂O₃ samples with T_f values (higher than in bare γ-Fe₂O₃ NPs) of 74 and 208 K, respectively. Therefore, the microporous zeolite 5A framework functionalizing the NPs favors the surface magnetic disorder and interparticle magnetic interactions giving rise to a SSG state. Considering that the chemical properties of the zeolites type 5A are maintained and also the magnetic properties of the γ-Fe₂O₃ NPs are still kept even at room temperature, let us suggest that these combined magnetic and adsorbent properties of our magnetic zeolite can be usefully employed in magnetic water remediation and environmental processes.

See the supplementary material for details on hyperfine parameters for the NPZEO3 sample, parameters obtained from the modified Mørup’s relation, Mössbauer spectra of the NPZEO1, NPZEO2, and NPZEO3 at selected temperatures, and slowing down plot of the NPZEO3 sample.

The authors thank the Universidad Nacional Mayor of San Marcos, IPKM-TUBS, and UFES for support. Edson Caetano Passamani also thanks FAPES and CNPq for supporting his research studies at UFES.

The data that support the findings of this study are available from the corresponding author upon reasonable request.