Influence of W Addition on Phase Constitution, Microstructure and Magnetic Properties of the Nanocrystalline Pr₉Fe₆₅W_xB_26-x (Where: x = 2, 4, 6, 8) Alloy Ribbons

Abstract

The aim of the present work was to investigate an influence of W addition on the phase constitution, microstructure and magnetic properties of the Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) alloy ribbons. Ribbons were obtained using the melt-spinning technique under low pressure of Ar. The as-cast samples were fully amorphous and revealed soft magnetic properties. These facts were confirmed by X-ray diffractometry, Mössbauer spectroscopy and magnetic measurements. Differential scanning calorimetry and differential thermal analysis allowed us to determine the thermal stability parameters of the amorphous phase. The Kissinger plots were constructed in order to calculate the activation energies for crystallization. Heat treatment carried out at various temperatures caused changes in the phase constitution and magnetic properties of the alloys. The phase analysis has shown the presence of the hard magnetic Pr₂Fe₁₄B and paramagnetic Pr_1+xFe₄B₄ phases. Additionally, for the x = 2 and x = 6 alloys, a crystallization of soft magnetic Fe₂B and α-Fe phases was observed. The Mössbauer spectroscopy allowed us to determine the volume fractions of constituent phases formed during annealing. The microstructure of annealed ribbons was observed using transmission electron microscopy.

1. Introduction

For over half a century, much attention has been paid to the development of various types of high-performance permanent magnets based on rare-earth (RE) transition-metal (TM) intermetallic compounds. Their exceptional magnetic properties can be attributed to the favorable combination of properties such as high magnetocrystalline anisotropy, which is provided by the RE elements and high magnetic moments of TM elements also present in the unit cells of these compounds [1]. The most important advance in RE hard magnetic materials was made in 1984 when a new Nd₂Fe₁₄B ternary compound was announced by two independent groups—Sagawa et al. [2] at Sumitomo Special Metals Co., Ltd. (Osaka, Japan) and Croat et al. [3] at General Motors Co. (Warren, MI, USA). The first group mentioned above developed the anisotropic magnets based on aligned and sintered 10 µm diameter particles of the Nd₂Fe₁₄B phase. The magnetic parameters measured for this alloy were the coercivity _JH_c = 1000 kA/m, the polarization remanence J_r = 1.2 T and the maximum magnetic energy product (BH)_max = 288 kJ/m³. The second group developed the melt-spun ribbons that contained ultrafine randomly oriented grains of the Nd₂Fe₁₄B phase of the average diameters less than 100 nm. The typical magnetic properties of these magnets were _JH_c = 1120 kA/m, J_r = 0.8 T and (BH)_max = 122 kJ/m³. Further studies on the Nd-Fe-B type magnets were oriented towards improving their properties and finding more favorable manufacturing techniques. The composition of the alloy was modified many times over the years to give a whole family of RE-(Fe, TM)-B magnets [4,5,6]. In particular, Pr-Fe-B type magnets are of considerable interest, due to the fact that the Pr₂Fe₁₄B phase is responsible for the hard magnetic properties. The Pr₂Fe₁₄B phase has similar magnetic properties and the same crystalline structure as the Nd₂Fe₁₄B [7]. Furthermore, the saturation polarization J_s and Curie temperature T_C are also similar [8], but for the Pr₂Fe₁₄B phase, a higher magnetocrystalline anisotropy field was measured [9]. Furthermore, due to the fact that the Pr₂Fe₁₄B phase does not undergo a spin reorientation down to 25 K [10], the Pr-based alloys sustain good magnetic properties at low temperatures. Up to now, numerous pieces of work have been carried out in order to investigate the effect of micro-alloying on the microstructure and magnetic properties in the Pr-Fe-B magnets [11,12,13,14,15,16,17,18]. Some studies have been published on the influence of W addition on the microstructure and magnetic properties of the RE-Fe-B alloys [19,20,21,22,23,24,25,26,27]. It was shown in Reference [26] that small additions of W to the alloy composition can significantly improve the glass forming abilities of the alloy. Furthermore, its admixture to the Nd-Fe-B alloy resulted in some changes in the magnetic properties [27,28]. The studies carried out on Nd-Fe-B magnets produced by the mechanical alloying have revealed that a gradual increase of W in the alloy composition causes an increase of the coercivity _JH_c in a wide range of W contents. It was also proven that this effect is accompanied by a gradual decrease of the remanence J_r and maximum magnetic energy product (BH)_max [27,28]. It should be noticed that obtaining the optimal magnetic parameters depends also on the production technology. An adequate and cheap processing technique is an important factor in the final market price of the magnets. The application of the rapid solidification techniques combined with heat treatment is a response to these needs. It allows us to obtain appropriate nanocrystalline structure and hard magnetic properties at shorter processing time, which can significantly reduce the production costs. Moreover, these costs can be further cut down by lowering the RE content in the alloy composition. Previous studies on Pr-(Fe,Co)-B-type alloys with high boron contents and additions of Ti revealed a high coercivity in the as-cast state [29]. It was shown for the Fe_91-xZr₅B_xNb₄ (x = 0 to 36) alloys, that increase of B content above 20 at. %, result in significant improvement of the glass forming abilities (GFA). Good GFA was observed up to 30 at. % of B, while a further increase of B resulted in a sharp drop of GFA [30]. A beneficial impact of B content on the GFA was also utilized in the RE-Fe-B-type magnets. In the case of the Fe₆₇Co_9.5Nd₃Dy_0.5B₂₀ alloy it was possible to cast up to 0.5 mm diameter rods [31,32], while for Fe₆₁Co_13.5Zr₁Pr_4.5-xDy_xB₂₀ (x = 0, 1) alloys [33], fully amorphous 1 mm diameter rods and up to 3 mm outer diameter tubes were suction-cast. In general, the RE-Fe-B magnets can be divided into two groups: the Fe-reach [34] and B-reach alloys [35]. The main advantages of the B-reach magnets are the low contents of expensive RE elements, their low melting points and good GFA. Furthermore, the devitrification of the B-reach RE-Fe-B alloys results in the formation of nanocomposite magnets consisting of Fe₃B, Nd₂Fe₁₄B, and α-Fe phases [36]. For these reasons, the B-reach magnets can be an interesting alternative to the commonly used Fe-reach RE-Fe-B magnets. It is well established that Nb addition to the amorphous alloys results in retarding the diffusion of iron atoms [37]. This prevents the fast growth of α-Fe during casting. Furthermore, it was shown, for the rapidly solidified (Fe_86-xNb_xB₁₄)_0.88Tb_0.12 (2 ≤ x ≤ 8) alloys [38], that addition of Nb and appropriate rapid solidification are responsible for the magnetic hardening effect. This is caused by the formation of dendrites of the Tb₂Fe₁₄B hard magnetic phase, with Nb-rich interdendritic regions. On the other hand, the addition of Nb resulted in good glass forming abilities. Subsequent annealing allowed us to tailor the magnetic properties of the bulk glassy samples [39].

In the present work, the influence of the W addition on phase constitution, microstructure and magnetic properties of the rapidly solidified Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) alloy ribbons, was discussed. Our goal was to determine the influence of W addition on the glass forming abilities of the Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) alloys. Furthermore, for the devitrified samples, the influence of W content on the phase constitution and magnetic properties will be investigated.

2. Materials and Methods

The Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) alloys were synthesized by arc-melting from pure elements (Pr ingot of the purity 99.9%; Fe granular of the purity 99.98%; W foil of the purity 99.9% were purchased at Sigma-Aldrich (St. Louis, MO, USA)) with the addition of the Fe-B pre-alloy (of the purity 98%, bought at Alfa-Aesar) under the Ar atmosphere. The alloys were subsequently melt-spun under an Ar atmosphere to obtain ribbon samples. In this process, the molten alloys were injected on the surface of the copper roll by the overpressure of the argon. The linear velocity of the copper roll surface was 25 m/s. The thickness of the ribbon depends on the composition of the alloys. The average thicknesses were 15 μm, 20 μm, 18 μm and 20 μm, respectively for the Pr₉Fe₆₅W₂B₂₄, Pr₉Fe₆₅W₄B₂₂, Pr₉Fe₆₅W₆B₂₀ and Pr₉Fe₆₅W₈B₁₈ alloys. In order to obtain a nanocrystalline microstructure and induce some changes in the magnetic parameters, the samples were subjected to heat-treatment. For this purpose, the ribbons were sealed-off in quartz tubes under a low pressure of Ar (0.5 bar) and subjected to annealing for 5 min at temperatures ranging from 943 K to 1023 K and then rapidly quenched in icy water. These annealing conditions have been proven beneficial for this alloy [40]. Thermal stability parameters were determined using differential scanning calorimetry (DSC) (DSC 404, Netzsch, Selb, Germany) and differential thermal analysis (DTA) (LabSys DTA/DSC, Setaram, Caluire, France) techniques. To establish the activation energies for crystallization, the DSC curves were measured with different heating rates (1 K/min ≤ α ≤ 10 K/min) for each alloy composition. The phase constitution of specimens was studied using X-ray diffractometry (XRD). The XRD patterns were measured using Bruker D8 Advance diffractometer (Bruker AXS GmbH, Karlsruhe, Germany) with CuK_α radiation, LynxEye semiconductor detector, (linear focus of 25 mm, primary beam divergence slit of 0.6 mm) and Soller slits on the primary and diffracted beam. The measurements were performed in the Bragg-Brentano configuration with the Ni-K_β filter located on the detector side. A continuous scan mode in the 2θ range between 30 and 100 deg with the step size of 0.02 deg and a step time of 5 s were used. The phase analysis was performed using EVA software (Bruker AXS GmbH, Karlsruhe, Germany). The Mössbauer spectra were measured at room temperature using a Polon 2330 Mössbauer spectrometer (Polon, Poland) working in transmission geometry, utilizing the ⁵⁷Co:Rh source. The Mössbauer spectra were fitted with Win Normos for Igor 3.0 package. In order to obtain a specimen representative for the entire volume of the material, the samples were crushed to powder. Measurements of magnetic hysteresis loops were performed using a LakeShore 7307 Vibrating Sample Magnetometer (VSM, Lakeshore Cryotronics, Westerville, OH, USA), operating in external magnetic fields up to 2 T at room temperature. The observation of the microstructure was carried out using transmission electron microscopy (TEM, JEOL JEM 1200 EX, JEOL Inc., Peabody, MA, USA). The diagrams of the crystallite size distributions were constructed base on the analysis of the TEM images, taken from different areas of the samples. To reduce the measurement uncertainties, a large series of crystallite size measurements were carried out (~750 for each sample).

3. Results and Discussion

The XRD patterns measured for the rapidly solidified Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) alloy ribbons in the as-cast state are shown in Figure 1. These samples had an amorphous structure, which was indicated by the presence of wide bumps on XRD scans in the range of 2θ between 35 and 50 deg, characteristic for the amorphous phase.

Analysis of the Mössbauer spectra measured for the as-cast Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) (Figure 2) ribbons confirmed their amorphous structure. In all cases, large broadenings of the Mössbauer lines were measured. In the fitting procedure, the hyperfine field distributions were calculated and experimental spectra were fitted by broad lines corresponding to these distributions. The bimodal character of the distributions suggests the existence of nonequivalent surroundings of Fe atoms in the amorphous phase. Furthermore, there is no significant shift in the maxima of the hyperfine field distributions with the increase of the tungsten content. This indicates similar magnetic properties of the amorphous phase for all alloy compositions.

The DSC and DTA scans measured for the amorphous ribbon samples of all investigated compositions are presented in Figure 3 and Figure 4, respectively. These measurements allowed us to determine the glass transition temperature T_g, onset crystallization temperature T_x, melting temperature T_m, supercooled liquid region ΔT_x = T_x − T_g, reduced glass transition temperature T_rg = T_g/T_m as well the glass forming ability parameter γ = T_x/(T_g + T_m) (

Table 1. Thermal stability parameters determined for amorphous ribbons of Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) alloys: T_g—glass transition temperature, ${\mathrm{T}}_{{\mathrm{x}}_1}$ and ${\mathrm{T}}_{{\mathrm{x}}_2}$—first and second onset crystallization temperature, ${\mathrm{T}}_{{\mathrm{m}}_1}$ and ${\mathrm{T}}_{{\mathrm{m}}_2}$—first and second melting temperatures, ΔT_x = T_x − T_g —supercooled liquid region, T_rg = T_g/T_m—reduced glass transition temperature, γ = T_x/(T_g+T_m)—glass forming ability parameter.

x (at. %)	Tg (K)	${\mathbf{T}}_{{\mathbf{x}}_1}\left(\mathbf{K}\right)$	${\mathbf{T}}_{{\mathbf{x}}_2}\left(\mathbf{K}\right)$	Tm (K)	ΔTx (K)	Trg	γ
2	920	940	-	1371	20	0.67	0.41
4	905	915	-	1371	10	0.66	0.40
6	914	920	-	1369	6	0.67	0.40
8	934	945	1020	1370	11	0.68	0.41

) [41,42].

The changes of T_g and T_x, as well as ΔT_x, were observed with the change of W content. However, other parameters (T_m, T_rg and γ) did not significantly change. The values of ΔT_x suggest that the Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) alloys should not exhibit good glass forming abilities. On the other hand, high values of T_rg (~0.6) and γ (~0.3) indicate good glass forming abilities and a possibility of manufacturing amorphous samples with significant geometric dimensions [43,44,45]. However, our previous studies [22] of the Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) alloys in the form of 1 mm rods and 0.5 mm thick plates produced using the suction-casting technique have shown partial crystallization of the as-cast specimens.

The activation energies E_a for crystallization for each alloy compositions were calculated using the Kissinger analysis [46] base on the plots shown in Figure 5. The values of E_a were collected in

Table 2. The activation energies E_a for crystallization for the Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) alloy ribbons.

x (at. %)	Ea (kJ/mol)
2	544 ± 11
4	330 ± 7
6	644 ± 13
8	320 ± 7

. Error analysis was carried out using standard methods, taking into account apparatus input, apparatus calibration quality and subjective assessment of uncertainty associated with the determination of characteristic temperatures based on DSC curves. The results indicate that for Pr₉Fe₆₅W₄B₂₂ and Pr₉Fe₆₅W₈B₁₈ alloys, the crystallization processes feature lower values of E_a compared to those for the Pr₉Fe₆₅W₂B₂₄ and Pr₉Fe₆₅W₆B₂₀ alloys. These differences can be attributed to the crystallization of additional soft magnetic Fe₂B or α-Fe phases during heat treatment of alloys containing 2 and 6 at. % of W, respectively (see later).

The XRD scans measured for Pr₉Fe₆₅W_xB_26-x (where x = 2, 4, 6, 8) alloy ribbons subjected to annealing are shown in Figure 6. Short-time annealing of the specimens resulted in significant changes in their phase constitution. The crystalline phases detected for all annealed samples are the hard magnetic Pr₂Fe₁₄B and paramagnetic Pr_1+xFe₄B₄. Furthermore, the phase analysis for the Pr₉Fe₆₅W₂B₂₄ alloy ribbons has shown the presence of the reflexes corresponding to the soft magnetic Fe₂B phase (Figure 6a), while for the Pr₉Fe₆₅W₆B₂₀ alloy, the α-Fe phase was detected (Figure 6d). Heat treatment at 943 K and higher temperatures resulted in nucleation and growth of the crystalline phases. The increase of annealing temperature caused changes in the volume fractions of the constituent phases, which was reflected in their magnetic properties. Moreover, for the Pr₉Fe₆₅W₂B₂₄ alloy, the widening of the diffraction peaks is related to low volume fractions and smaller grain sizes of the crystalline phases.

Based on the XRD studies of annealed Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) ribbons, one can deduce that the wide crystallization peaks on the DSC and DTA curves corresponding to ${\mathrm{T}}_{{\mathrm{x}}_1}$ (Figure 3 and Figure 4) indicate the formation of the Pr₂Fe₁₄B and Pr_1+xFe₄B₄ phases. It is expected that the ${\mathrm{T}}_{{\mathrm{x}}_2}$ may be related to the decomposition of the hard magnetic Pr₂Fe₁₄B phase. This might be related to some oxidation processes leading to a reduction of the volume fraction of the hard magnetic phase. This might be a reason for the decrease of _JH_c in ribbon annealed at 1023 K (see Table 4).

The transmission Mössbauer spectra measured for the Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) ribbons annealed at 1003 K for 5 min are shown in Figure 7. The contribution from the hard magnetic Pr₂Fe₁₄B phase was represented by six sextets each of them corresponding to a distinct Fe site in this phase. They correspond to six nonequivalent positions of Fe atoms in its elementary cell. Base on the Wyckoff notation they are labeled as 16k₁, 16k₂, 8j₁, 8j₂, 4e and 4c, for which the relative intensity ratios of 4:4:2:2:1:1 were chosen. The standard fitting procedure for the Pr₂Fe₁₄B phase was based on assumptions introduced by Pinkerton et al. [47]. In order to match theoretical spectra to the experimental data, additional spectral components were added. Particularly, the doublet line corresponding to the paramagnetic Pr_1+xFe₄B₄ crystalline phase was appended to the fitting model [48]. In the case of the Pr₉Fe₆₅W₂B₂₄ alloy (Figure 7a), the presence of the soft magnetic Fe₂B phase was also considered [49]. The Fe₂B phase was represented by a single sextet. The quantitative analysis of the spectrum has shown that the Pr₉Fe₆₅W₂B₂₄ ribbon sample contains low volume fractions of the hard magnetic Pr₂Fe₁₄B (9 vol. %) and paramagnetic Pr_1+xFe₄B₄ (9 vol. %) phases. These calculations have also shown that the largest volume fraction corresponds to the soft magnetic Fe₂B phase (82 vol. %) thus explaining relatively weak magnetic properties of the annealed Pr₉Fe₆₅W₂B₂₄ ribbons. The shape of the Mössbauer spectrum measured for the Pr₉Fe₆₅W₄B₂₂ alloy ribbon suggests the presence of a major fraction of crystalline phases (Figure 7b). In order to obtain a good fit an additional broad sextet line, defined by the hyperfine field distribution, was included. As there are limitations regarding the number of sub-spectra that can be independently fitted by the Mössbauer spectra analysis software, the parameters for the low-intensity sub-spectra corresponding to 4e and 4c positions of Fe atoms were arbitrary set. One could expect that the short-time heat treatment (for 5 min) led to the presence of a remnant amorphous phase. However, the broad sextet line more likely can be attributed to the highly disordered hard magnetic Pr₂Fe₁₄B phase, formed during short-time annealing and then rapid quenching in icy water. A narrow hyperfine field distribution (instead of a broad one, which is typical for the amorphous phase) was calculated. It can be an argument in favor of the formation of a highly disordered crystalline Pr₂Fe₁₄B phase. The Pr₉Fe₆₅W₄B₂₂ ribbon sample consists of 58 vol. % of the paramagnetic Pr_1+xFe₄B₄ phase, 36 vol. % of the hard magnetic Pr₂Fe₁₄B phase and 6 vol. % of the disordered phase. The Mössbauer spectrum of the Pr₉Fe₆₅W₆B₂₀ alloy ribbon (Figure 7c) has indicated the presence of the components corresponding to the hard magnetic Pr₂Fe₁₄B (13 vol. %), paramagnetic Pr_1+xFe₄B₄ (53 vol. %), highly disordered phase (23 vol. %) and a low volume fraction of soft magnetic α-Fe phase [50] (11 vol. %). The best fitting of the experimental data for the annealed ribbon of the Pr₉Fe₆₅W₈B₁₈ alloy (Figure 7d) was attained by taking into account three components—the paramagnetic Pr_1+xFe₄B₄, hard magnetic Pr₂Fe₁₄B and highly disordered phases. The corresponding volume fractions of the constituent phases were 37 vol. % of the paramagnetic Pr_1+xFe₄B₄ phase, 48 vol. % of the hard magnetic Pr₂Fe₁₄B phase and 15 vol. % of the highly disordered phase. A large volume fraction of the paramagnetic Pr_1+xFe₄B₄phase reduced the saturation magnetization J_s. The presence of large volume fractions of hard magnetic and crystallographically highly disordered (also hard magnetic) phases were responsible for the ferromagnetic behavior of the investigated alloys. However, the volume fractions of the constituent phases changed significantly with the W addition, which was reflected in the magnetic properties of the investigated alloys.

The calculated hyperfine parameters (B_hf—hyperfine field, IS—isomer shift, QS—quadrupole splitting) corresponding to the constituent phases of the Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) ribbons annealed at 1003 K for 5 min are collected in

Table 3. The hyperfine parameters of the Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) alloy ribbons annealed at 1003 K (B_hf—hyperfine field, IS—isomer shift, QS—quadrupole splitting, V—fractions of constituent phase).

x	Identified Phases		Bhf (T)	IS (mm/s)	QS (mm/s)	V (vol. %)
2	Pr2Fe14B	16k1	28.7	−0.19	0.58	9
		16k2	30.6	0.07	0.18
		8j1	19.6	−0.07	0.29
		8j2	34.8	−0.17	0.85
		4e	25.6	−0.64	−0.67
		4c	27.9	0.80	−0.21
	Pr1+xFe4B4		-	−0.01	1.08	9
	Fe2B		23.3	−0.00	0.03	82
4	Pr2Fe14B	16k1	28.1	−0.25	0.13	36
		16k2	30.6	0.13	0.99
		8j1	27.7	−0.71	0.28
		8j2	26.0	0.53	0.01
		4e	25.6	0.01	−0.67
		4c	28.1	−0.15	−0.21
	Pr1+xFe4B4		-	−0.07	0.53	58
	disordered Pr2Fe14B		-	-	-	6
6	Pr2Fe14B	16k1	27.2	−0.14	0.55	13
		16k2	27.3	−0.14	0.23
		8j1	22.7	−0.71	−1.05
		8j2	32.9	0.40	−0.09
		4e	25.6	0.01	−0.67
		4c	29.6	−0.15	−0.21
	Pr1+xFe4B4		-	−0.07	0.58	53
	α-Fe		33.1	−0.15	0.12	11
	disordered Pr2Fe14B		-	-	-	23
8	Pr2Fe14B	16k1	27.7	0.50	0.71	48
		16k2	28.0	−0.14	0.17
		8j1	29.6	−0.70	−0.40
		8j2	31.6	−0.54	0.69
		4e	25.6	−0.63	−0.67
		4c	29.6	−2.09	2.22
	Pr1+xFe4B4		-	−0.08	0.55	37
	disordered Pr2Fe14B		-	-	-	15

. Maximum uncertainties for the calculated hyperfine parameters are ΔB_hf = ± 0.1 T for hyperfine field, ΔQS = ± 0.05 mm/s for the isomer shift, ΔIS = ± 0.002 mm/s for quadrupole splitting.

The selected hysteresis loops J(H) measured for the Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) alloy ribbons are shown in Figure 8. The hysteresis loops were measured as a dependence of magnetic polarization J vs. external magnetic field H, therefore, the determined coercive fields correspond to the magnetic field at which J reaches zero. J(H) measured for as-cast samples are characteristic of the soft magnetic materials. Annealing of ribbons containing 2 at. % of W did not cause significant changes in their magnetic properties. Low coercivity _JH_c indicate a low volume fraction of a hard magnetic Pr₂Fe₁₄B phase, which was previously confirmed by the Mössbauer spectra analysis. A change of W content resulted in an evolution of microstructure and phase constitution of annealed specimens, thus leading to significant changes in their magnetic properties. Magnetic measurements carried out on ribbons containing 4, 6 and 8 at. % of W have shown their hard magnetic properties. Especially, an increase of the coercivity _JH_c and maximum magnetic energy product (BH)_max with the increase of the annealing temperature was observed. However, for the x = 6 and x = 8 alloys annealed at 943 K, the wasp-waisted shapes of the hysteresis loops were measured. This suggests the formation of the hard magnetic Pr₂Fe₁₄B phase within the soft magnetic amorphous matrix. For the alloys containing 4 and 8 at. % of W annealed at higher temperatures, the presence of large volume fractions of the hard magnetic Pr₂Fe₁₄B phase resulted in wide hysteresis loops. Based on our Mössbauer studies (Table 3) one can assume that the improvement of _JH_c is caused by the larger volume fraction of the hard magnetic phases in the vicinity of PrFe₄B₄ type borides in the x = 4 alloy, while the presence of α-Fe and Fe₂B for the annealed samples of the x = 2 and x = 6 alloys, respectively, resulted in a higher J_r value. Higher (BH)_max for the x = 8 alloy might be somewhat related to a more homogeneous microstructure of the annealed specimens (see Figure 10). It was shown that the increase of W contents at the expense of Fe resulted in a decrease of the saturation polarization J_s of the as-cast specimens. The magnetic parameters measured for the Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) alloy ribbons are collected in

Table 4. Magnetic parameters: coercivity _JH_c, remanence polarization J_r, maximum magnetic energy product (BH)_max and saturation polarization J_s for annealed ribbon samples.

x (at. %)	T (K)	JHc (kA/m)	Jr (T)	(BH)max (kJ/m3)	Js (T)
2	943	3	0.05	-	1.12
	963	3	0.07	-	1.12
	983	21	0.17	-	1.13
	1003	60	0.35	3	1.04
	1023	77	0.33	4	1.04
4	943	1013	0.29	14	0.42
	963	1053	0.30	16	0.45
	983	975	0.29	14	0.48
	1003	1148	0.30	13	0.49
	1023	972	0.30	13	0.45
6	943	14	0.01	-	0.83
	963	204	0.26	6	0.58
	983	390	0.33	1	0.61
	1003	319	0.32	11	0.62
	1023	359	0.32	10	0.66
8	943	136	0.25	2	0.77
	963	910	0.40	22	0.63
	983	960	0.39	23	0.62
	1003	983	0.37	20	0.60
	1023	882	0.34	15	0.63

. Based on technical specifications of the LakeShore VSM 7307 magnetometer, uncertainties for the measurement of the magnetic field were taken as 1% of the readings from the apparatus. The maximum coercivity of _JH_c = (1148 ± 12) kA/m was measured for the Pr₉Fe₆₅W₄B₂₂ alloy ribbon annealed at 1003 K, while the largest (BH)_max = (23 ± 3) kJ/m³ was determined for the Pr₉Fe₆₅W₈B₁₈ alloy ribbon annealed at 983 K.

The microstructure of samples subjected to annealing was observed using a transmission electron microscope. A heterogeneous microstructure was observed for the Pr₉Fe₆₅W₄B₂₂ sample (Figure 9). Here, next to the large grains (diameter of ~70 nm), some small crystallites having diameters lower than 20 nm, were measured. Moreover, the distribution of grain sizes obtained for the Pr₉Fe₆₅W₄B₂₂ alloy ribbon differs from those obtained for the Pr₉Fe₆₅W₈B₁₈ alloy. For the Pr₉Fe₆₅W₈B₁₈ alloy, the TEM studies have revealed the existence of agglomerates having the same crystallographic orientation that can also impact the magnetic properties of the sample (Figure 10). In the case of the Pr₉Fe₆₅W₈B₁₈ alloy ribbon annealed at 1003 K, the grain size distribution was relatively wide with a maximum at ~30 nm (Figure 10b). It has to be pointed out that except for the phase composition the microstructure can also play a significant role in shaping the magnetic properties. The electron diffraction patterns obtained for the investigated samples confirmed the presence of crystalline phases and indicated their nanocrystalline structure. The presence of a diffused electron diffraction pattern for the x = 8 alloy was caused by a large fraction of crystallites with the sizes lower than 10 nm. In the case of the x = 4 alloy, the presence of crystallites larger than 80 nm may give an additional point diffraction pattern superimposed on the diffused diffraction pattern.

4. Conclusions

In the present study, the phase constitution, microstructure and magnetic properties of the nanocrystalline Pr₉Fe₆₅W_xB_26-x (where: x = 2, 4, 6, 8) alloy ribbons in an as-cast state and subjected to annealing were investigated. XRD and Mössbauer spectroscopy have shown that the as-cast ribbons of all alloy compositions were fully amorphous. Moreover, the analysis of the Mössbauer spectra has shown the bimodal character of the hyperfine field distribution, suggesting the existence of nonequivalent surroundings of Fe atoms in the amorphous phase. The hysteresis loops measured for the as-cast samples were characteristic for soft magnetic materials. Heat treatment of the samples led to precipitation of various crystalline phases. The phase constitution changes with the composition of the alloys. However, for all annealed samples, the crystallization of the hard magnetic Pr₂Fe₁₄B and the paramagnetic Pr_1+xFe₄B₄ phases took place. Additionally, for Pr₉Fe₆₅W₂B₂₄ and Pr₉Fe₆₅W₆B₂₀ alloy ribbons, the soft magnetic Fe₂B and the α-Fe phases were observed. The crystallization of these additional crystalline phases during heat treatment can be related to higher values of the activation energy E_a for crystallization. An increase in tungsten content, as well as the rise of annealing temperature, resulted in the improvement of the magnetic properties. The maximum value of coercivity _JH_c = (1148 ± 12) kA/m was measured for the Pr₉Fe₆₅W₄B₂₂ alloy ribbon annealed at 1003 K, whereas the largest maximum magnetic energy product (BH)_max = (23 ± 3) kJ/m³ was determined for the Pr₉Fe₆₅W₈B₁₈ alloy ribbon annealed at 983 K. The TEM studies revealed a heterogeneous mixture consisting of large (of diameter of ~70 nm) and small (of the diameter < 20 nm) crystallites for the Pr₉Fe₆₅W₄B₂₂ alloy. For the Pr₉Fe₆₅W₈B₁₈ ribbon, the largest number of grain reached sizes of ~30 nm. TEM dark-field studies of the Pr₉Fe₆₅W₈B₁₈ alloy specimen revealed a presence of agglomerates having the same crystallographic orientation. This might be a reason for the highest (BH)_max measured for the ribbon annealed at 1003 K.