The results will be comparatively presented on samples obtained using the two deposition methods, in terms of structure and morphology, as well as of the magnetic properties.
3.1. Morphology and Structure of Fe-B Films
The comparative SEM measurements on similar films from the two batches of samples (obtained by TVA and MS, respectively) are presented in
Figure 2. Films from the two batches, with 20 at. % and 50 at. % of B, were chosen to be presented due to the clear differences between each other.
Figure 2 shows the SEM images of Fe-B films at a magnification of 10 kX. The morphology is lacking significant roughness on the surface of the films in both deposition batches. However, a clear difference can be seen between the films obtained by TVA and MS. According to
Figure 2, the level of mesoscopic uniformity and roughness of the films obtained by TVA is clearly superior to the ones obtained by MS. Lower differences (films of very good uniformity) were observed for other B contents.
XRD patterns on two samples from each batch are presented for comparison in
Figure 3.
First, it is observed that for low concentrations of B (5%), a well crystallized film is obtained by TVA (black line), whereas an amorphous-like or nanocrystalline film (red line) is obtained by the more energetic MS method. For high concentrations of B, the films obtained by both methods are amorphous. However, XRD is not able to determine an accurate estimation of the relative content of crystalline and amorphous phase, as expected at low to intermediate content of B. In this respect, CEM spectroscopy results, as presented subsequently, are more concluding.
The XRR measurements for both films with 20 at. % boron nominal composition obtained by TVA and MS, respectively, are shown in
Figure 4. Much better formed oscillations are evidenced in the XRR pattern of the film obtained by TVA, as direct proof of a much better uniformity of the film at nanoscale compared to the case of the similar film obtained by MS. However, from the involved beats, a different thickness is evidenced for the two films: only approximately 71(1) nm in case of the film obtained by TVA and 103(2) nm for the film obtained by MS.
EDX measurements were made on the two films with 50 at. % of B obtained by TVA and MS, respectively. The difficulty of an accurate determination of light elements (e.g., Be, B) by this technique is well known. However, the EDX spectra of the two samples with the highest nominal concentrations of B (50 at. %) prepared by the two methods give evidence for a lower amount of residual oxygen and to a higher boron content in the sample prepared by TVA as compared to the one prepared by MS, and consequently, to a much more proper ratio between Fe and B in the first sample. Nevertheless, this is rather a qualitative result and not an accurate one, especially related to the B amount. A much more proper method for the elemental investigation of such films is RBS. The elemental content was also checked by NRMS, which is even more sensitive to the light elements when compared to RBS, especially when the substrate has a high Z number.
An exemplification of the RBS data for samples of different B content obtained by MS are presented in
Figure 5a, whereas NRBS spectra collected on samples obtained by TVA are shown in
Figure 5b. The RBS spectra were simulated using SIMNRA software package [
23]. Areal density or thin film units (10
15 atoms/cm
2) are the natural units for ion beam analysis (IBA) since the energy loss is measured in eV/(atoms/cm
2), and one monolayer is of the order of 10
15 atoms/cm
2 [
24,
25]. The following tables present the RBS results regarding the stoichiometry of the samples.
On all the samples, the presence of some impurities that could be due to some surface contamination of the samples is noticeable, e.g., it is well-known that also during ion beam analysis there is a carbon build-up on the sample surface. This is a mixture of C, H and O with thicknesses in the order of 50 TFU (1 TFU = 10
15 atoms/cm
2). Other contaminants may be present on the sample surfaces (in a very low concentration) due to the handling or from some other reasons. In addition, a small and negligible concentration of W of 0.03% was observed in all TVA samples, probably resulting from the tungsten filament used as a cathode in TVA deposition. However, the most significant observation is related to an order of magnitude higher content of oxygen (tenths of at. %) in samples obtained by MS compared to samples obtained by TVA (order of at. %) showing definitively the much better quality of samples obtained by the second method. Concerning the Fe-B films obtained by MS, the Mössbauer results show the presence of an Fe oxide phase superposed over Fe-B intermetallic phases, and therefore, the average composition of the Fe-B phase cannot be subtracted directly from
Table 2 by a simple renormalization between Fe and B. We noted that the
57Fe Mössbauer results presented in the next section definitively prove the lack of Fe oxide phases and the formation of only Fe-B intermetallic phases in the Fe-B samples obtained by TVA. In this case, the few at. % of oxygen in the samples should be seen as inclusions and not making chemical bounds. As a consequence, the oxygen content will influence only the relative content of Fe and B in the Fe-B phases formed in the films, with direct influence on their magnetic properties. The real elemental contents in the Fe-B thin films obtained by TVA (in at. %), to be considered in the following for the interpretation of Mössbauer and magnetic results are presented in
Table 3.
It is worth noticing that the real compositions are almost similar to the nominal ones in cases with low B content (i.e., less than 10 at. %) as well as in cases with higher B content (i.e., higher than 40 at. %), whereas in cases with intermediate B content, the real compositions are higher than the nominal ones.
3.2. Mössbauer Spectroscopy
The Mössbauer spectra of the Fe-B thin films obtained by TVA are presented in
Figure 6. The emission spectrum of sample with 5 at. % B nominal composition consists in a sextet with narrow lines and hyperfine parameters close to metallic Fe (body centered cubic structure –bcc). Hence, the specific isomer shift (IS) is 0.01 mm/s and the hyperfine magnetic field (B
hf) is 33.1 T (0.00 mm/s and 33.15 T, respectively, for metallic Fe). The corresponding quadrupole splitting (QS) is 0.00 mm/s. Two superposed magnetic components, i.e., an external narrow and an inner broad magnetic sextet, respectively, are evidenced in the spectrum of the Fe-B film with 10 at. % B nominal composition. The first magnetic pattern with IS = 0.01(1) mm/s and B
hf = 33.1 T and with a relative spectral contribution of 29% corresponds again to a metallic Fe-like phase. The broad sextet was fitted via a hyperfine magnetic field distribution (presented on the right side of the spectrum), corresponding to a continuous distribution of local atomic configurations around the central Fe, specific to an amorphous Fe-B intermetallic phase.
The specific average hyperfine parameters of this Fe-B intermetallic phase are: <IS> = 0.14 mm/s, <Bhf > = 24 T and QS almost 0.0 mm/s. An average of the hyperfine parameters over the two main local configurations corresponding to the two spectral components gives: 29.4 T for the average hyperfine magnetic field and 0.04 mm/s for the average Isomer Shift. Further on, at higher B contents, the Mössbauer spectra consist only in broad magnetic sextets, specific to amorphous Fe based phases [
26,
27,
28,
29], which start to collapse at higher B or RE content due to enhanced magnetic relaxation. We noted the uniform distribution of B
hf with nominal B content between 20 and 40 at. % and the tendency of forming two average local configurations for B content of 50 at. % simultaneously with enhanced magnetic relaxation phenomena. On the other hand, the intensity ratio between the second and the third absorption line, R
23, in either the elemental sextets under the hyperfine magnetic field distributions at higher B concentrations as well as in the narrow sextet of samples of low B concentrations is 4.0, giving a direct proof for an in-plane orientation of the Fe spins [
26] in the Fe-B films obtained by TVA.
The evolution of the average hyperfine magnetic field and average isomer shift versus the real B content in such intermetallic compounds in case of reduced magnetic relaxation (i.e., for B nominal content lower than 40 at. %) is given in
Figure 7. An almost linear decrease in the average hyperfine magnetic field and an almost linear increase in the isomer shift is observed. Similar tendencies were also previously reported for Fe-B compounds [
12].
The Mössbauer spectra of Fe-B thin films obtained by MS are shown in
Figure 8. Two distinct features from the RT CEM spectra of films obtained by TVA are to be mentioned in this case: (i) the presence of a relative intense central component, and (ii) the R
23 ratio between the second and the third emission line of the external sextet in the spectra of low B content is 0, giving a direct indication for a perpendicular to the film plane orientation for the Fe spins in the intermetallic compounds. In this context, the following fitting procedures were considered for the films obtained by MS. The central components were always fitted by a narrow hyperfine field distribution extending from 0 up to maximum 20 T, whereas depending on the nominal B content, either an external relative broad sextet (at low B concentration) or an external hyperfine field distribution spreading out from 10 to 35 T (at high B content) were considered. The corresponding hyperfine magnetic field distributions are shown on the right side of each spectrum.
We noted that <IS> values belonging to the central component accounted by the distribution at low fields (e.g., with <B
hf> from 6 to 11 T, depending on the sample composition) range between 0.4 and 0.5 mm/s, which are specific values for Fe
3+ ions, whereas <IS> values belonging to the broad external sextet accounted by the distribution at high fields (e.g., with <B
hf> from 20 to 29 T, depending on the sample composition) range between 0.1 and 0.2 mm/s, as specific to Fe-B intermetallic phases with a B content ranging from approximately 25 at. % to 45 at. % (see
Figure 7). As a consequence, and also based on the RBS results presented in
Table 2, the central component was assigned to very disordered Fe
2O
3 clusters under a strong magnetic relaxation regime. The distributed sextet of higher average hyperfine magnetic fields to an amorphous Fe-B phase of higher amount of B and the most external sextet is evidenced mainly for films of low B content, to a disordered nanosized metallic Fe phase with inclusions of B.
In cases with samples with a nominal content of B of 5 at. %, the external sextet is characterized by Bhf = 32.3(2) T and IS = 0.01(1) mm/s, suggesting the formation of an intermetallic Fe-B phase with 7(2) at. % of B with the Fe spins oriented perpendicularly to the film plane. Its relative spectral contribution is 50(1) %, meaning that almost half of the Fe atoms enter in this phase. The rest of the Fe atoms/ions give rise to the Fe oxide clusters (50(1) % spectral contribution).
A quite similar spectral decomposition results also for the film with a nominal content of 10 at. % of B. A higher IS value for the intermetallic component (IS = 0.04(1) mm/s) suggests a higher B content in the Fe-B phase (approximately 10(2) at. % of B) of this sample. The relative spectral area of the Fe oxide phase is approximately 53(1) %, meaning that approximately 27 Fe
3+ ions and 40 O
−2 ions (from 100 atoms) participate in the Fe oxide phase and approximately 24 Fe atoms and 3 B atoms participate in the Fe-B phase, in reasonable agreement with RBS data in
Table 2. The same perpendicular to plane orientation of the Fe spins in the Fe-B phase is evidenced by Mössbauer spectroscopy (R
23 = 0) for this sample as well.
In cases with samples with nominal content of B of 20 at. %, except the external sextet characterized bt B
hf = 32.9(1) T and IS = 0.01(1) mm/s and associated to a local configuration close to nanosized metallic Fe (i.e., of very low B content), there is a second component fitted by a distribution of hyperfine magnetic fields with <B
hf > =29.2(2) T and <IS > =0.12(2) mm/s, easily associated via
Figure 7 to a Fe-B phase with about 30 at. % of B (i.e., Fe
70B
30). The relative spectral area of the last component is 44(1) % meaning, according to
Table 2, 23 atoms of Fe and 11 atoms of B (from 100 atoms) in this Fe-B phase. The most inner spectral component, fitted this time via a distribution of hyperfine magnetic fields with <B
hf > =7.2(2) T and <IS > =0.45(2) mm/s, is associated with the Fe
2O
3 phase. By the relative spectral area of this component of 26(1) %, a number of 14 ions of Fe
3+ and 21 atoms of O
2- from 100 atoms contribute to the Fe oxide phase, again in reasonable agreement with the RBS data in
Table 2. We mention the perpendicular to the film plane orientation of the Fe spins in the metallic phase poor in B (similar to the one of spectra corresponding to films with less than 10 at. % of B) and the in plane orientation of the Fe spins in the amorphous intermetallic phase of higher B content (Fe
70B
30).
Only two spectral components were considered for samples of higher B content. In cases with samples with nominal content of B of 30 at. %, the outer magnetic component, fitted via a distribution of hyperfine magnetic fields with <B
hf> = 20.5(2) T and <IS> = 0.15(2) mm/s, was assigned to an amorphous intermetallic Fe-B phase of type Fe
60B
40 (see
Figure 7), whereas the inner magnetic component was fitted by a distribution of hyperfine magnetic fields characterized by <B
hf> = 10.5(2) T and <IS> = 0.53(2) mm/s to the Fe
2O
3 phase. The relative spectral areas of 70(2) % and 30(2) %, respectively, infer 33 atoms of Fe and 22 atoms of B in the Fe-B phase and 15 ions of Fe
3+ and 22 ions of O
2- in the Fe oxide phase from 100 atoms, again in reasonable agreement with
Table 2. A similar reasoning is applied for the film with nominal composition of 40 at. % of B. Accordingly, the Mössbauer results support the formation of an amorphous intermetallic phase of type Fe
52B
48 (< B
hf> = 19.5(2) T and <IS > =0.19(2) mm/s) with relative spectral area of 54(2) % and an Fe oxide phase (<B
hf> = 6.5(2) T and <IS> = 0.51(2) mm/s) with a relative spectral area of 46(2) %. The corresponding atomic contribution at 100 atoms in the film is 23 Fe and 21 B in the Fe-B phase and 18 Fe and 28 O in the Fe oxide phase, in reasonable agreement with RBS results. The Fe spins of the intermetallic phases reached in B are in the film plane (R
23 = 4).
3.3. Magnetic Characterization
The large capabilities of the MOKE method to investigate the magnetic behavior in such thin films, including in plane anisotropies, are illustrated by the hysteresis loops presented in
Figure 9 and
Figure 10. The loops, consisting of the dependence of the Kerr angle on the applied magnetic field, were collected on the Fe-B thin films with thicknesses ranging from 70 to 100 nm, however, much thicker than the penetration depth of laser radiation (wavelength of 632 nm). Specific loops collected on films prepared by TVA, with different concentrations of B are shown in
Figure 9. Three orientations of the magnetic field relative to the (110) direction of the Si substrate, 0°, 45° and 90°, were considered for each sample in order to investigate the in-plane anisotropy, as previously reported in [
30,
31]. Briefly, the idea is that in the case of uniaxial anisotropy, the hysteresis loops evolve from a rectangular shape, specific to a field applied along the easy axis of magnetization to a linear shape, specific to a field applied along the hard axis of magnetization (which is perpendicular to the easy axis in this case). If the saturation field coincides with the switching field the magnetization reversal is achieved by a coherent rotation of the spins (e.g., for a magnetic monodomain), whereas in the case of a saturation field higher than the switching field, the magnetization reversal is via domain wall displacements. If the evolution of the loops is not between these two limits, then we deal with an angular distribution of the easy axis (the broader the distribution, the lower the variation of the loop shape, when the film is rotated, if the loops remain constant at different angular rotations of the samples (from 0 to 90° in the present experiments) a lack of in-plane anisotropy is demonstrated.
In this context, the following observation can be drawn from
Figure 9 for the films obtained by TVA: (i) the films with a B content lower or equal to 40 at. % are all ferromagnetic, (ii) the magnetic anisotropy is of uniaxial type and increases progressively with the B content starting from the case of isotropic behavior in those samples with very low B content, where mainly crystalline intermetallic phases are evidenced by Mössbauer spectroscopy (iii) magnetic domains are formed in the films and the magnetization reversal is through displacement of magnetic walls. (iv) The films are soft magnetic, with coercive fields ranging between 20 and 30 Oe, depending on the B content and reaching a maximum for the sample with 30 at. % of B which is completely amorphous and presents negligible magnetic relaxation.
The specific loops collected on films with equivalent nominal compositions of B, but prepared by MS, are shown in
Figure 10. The MOKE signal of the samples with the lowest B content (i.e., with nominal compositions of B lower than 10 at. %) is more than one order of magnitude lower than in the case of the films obtained by TVA. This observation, correlated also with the linear shape of negligible coercivity of the loop, gives support for the assignment of the MOKE signal to the Fe oxide phase rather than to the Fe-B intermetallic one. Nevertheless, this assignment is also supported by the Mössbauer results, proving the perpendicular to the film plane alignment of the Fe spins in the Fe-B phase of low B content with no influence on the in-plane magnetization reversal observed by the present longitudinal MOKE geometry. We noted that this intermetallic phase with a relative amount of 50% pointed by Mössbauer spectroscopy would be the only one susceptible to contribute to the magnetization reversal if the Fe spins had in-plane components. The situation is drastically changed in cases of the films with higher nominal B content (i.e., 30 or 40 at. %) where the MOKE signal can be assigned to the Fe-B amorphous intermetallic phase of the higher B content with the Fe spins in the sample plane. This time, the MOKE signal increased by more than one order of magnitude, however, remaining still lower than in the case of films obtained by TVA of equivalent nominal B content. This is due to a relatively lower content of the intermetallic phase in the film (more than 40% of Fe oxide are formed).
As a last observation, the loop of the sample with nominal content of B of 40 at. % is almost closed, with almost negligible coercivity, resembling rather a loop specific to a paramagnetic material, most probably due to the high content of B in the real Fe52B48 phase, where magnetic relaxation phenomena are substantially enhanced.