Low temperature texture development in Nd 2 Fe 14 B / α-Fe nanocomposite magnets via equal channel angular pressing

Low temperature texture development in Nd2Fe14B/α-Fe nanocomposite magnets via equal channel angular pressing L. Besley,1,a J. S. Garitaonandia,2 A. Molotnikov,1 H. Kishimoto,3 A. Kato,3 C. Davies,4 and K. Suzuki1 1Department of Materials Science and Engineering, Monash University, 3800 Melbourne, Australia 2Department of Applied Physics II Faculty of Science and Technology University of the Basque Country UPV/EHU, Sarriena, 48940 Leioa, Bilbao, Basque Country (Spain) 3Toyota Motor Corporation, Mishuku, Susono, Shizuoka 410-1193, Japan 4Department of Mechanical and Aerospace Engineering, Monash University, 3800 Melbourne, Australia


I. INTRODUCTION
2][3] In order to realise the full potential of nanocomposites, texture must be developed without losing exchange coupling.While texture can be developed in nanocomposites via thermomechanical processing such as die upsetting and the incorporation of low melting point eutectic Nd-rich additives (such as Nd-Cu and Nd-Ga) that help improve the workability and facilitate texture development in these materials, [4][5][6][7][8] the temperature and timescale required for such processing causes grain coarsening, resulting in a final microstructure with grain sizes on the order of 200-1000nm, 8 well above the predicted exchange length for Nd 2 Fe 14 B. 1 Other works show that exchange coupling may be observed experimentally at grain sizes slightly above this limit, but still on the order of 30-50nm. 9,10In order to maintain a Electronic mail: Luke.Besley@monash.edu2158-3226/2018/8(5)/056219/7

II. METHODS
Elements with >99.99% purity were arc melted with composition Nd 90 Cu 10 .Ingots were then melt spun with a wheel speed of 40 m/s and pulverised into particles of < 250µm under Ar.The Nd 90 Cu 10 was then added to a commercial hypo-stoichiometric Nd 7.91 Pr 2.68 Fe 84 B 5.41 alloy 13 (herein referred to as MQP15-7), at a ratio of x [NdCu] =4, 7, and 10 wt%, and the two powders were subsequently mixed under Ar.Mixed powders were then pre-compacted under Ar at T = 550 • C under a pressure of σ ≈ 125 MPa and then annealed at T = 550 • C under a vacuum of 10 3 Pa over a range of times (t).After annealing, the samples were then processed via ECAP at T = 520 • C (unless otherwise stated).Further details of the ECAP process and subsequent sample characterisation are described in previous work. 12Scanning Electron Microscopy images were taken with a JEOL 7001-F.Estimations of grain size were calculated from X-ray diffraction patterns using the integral breadth method. 14 57Fe Mössbauer spectra were acquired at room temperature using a constant acceleration transmission spectrometer with a 57 CoRh source.

A. The effect of annealing treatment
Fig. 1 shows the change in M-H curves with varying annealing time, while Table I shows changes in |BH| Max , ∆M r and grain size.Initially, without pre-annealing (i.e.t=0h), ∆M r and |BH| Max are quite low.With the a pre-annealing treatment, |BH| Max and M r increase.As t further increases, H c begins to significantly begins to significantly improve as well.However, beyond t=25h, H c , |BH| Max and M r begin to I also shows the change in grain size with varying annealing time.Even at the longest annealing time, the grain size tends to remain well below 100nm.Fig. 2 shows pole figures characterising the distribution of (001) planes after ECAP throughout the plane normal to the exit channel with varying annealing times (PD = pressing direction, TD = transverse direction).As t increases, the maximum m.r.d. 15 observed increases to a maximum at t=25h.The most prominent texture is observable in c), where the (001) basal plane aligns parallel to the shear direction produced in ECAP, distributed approximately 40 • away from the exit direction, normal to the easy axis (c-axis).This behaviour of (001) planes in Nd 2 Fe 14 B aligning parallel to the planes of shear in plastic deformation is consistent with previous studies involving ECAP of Nd 2 Fe 14 B 11 and die upsetting. 16Consistent with the trends observed in Fig. 1 and Table I, the maximum m.r.d.increases as t increases until t=25h where the strongest texture is observed and decreases afterwards.Fig. 3 shows the fracture surface observed via SEM after ECAP of a sample that had been annealed for t=25h with x [NdCu] =7wt%.Although a range of larger (i.e.>100nm) grains can be observed, many smaller grains are clearly visible, consistent with the mean grain size estimated by XRD in Table I.The variation in Table I suggests that first, the texture is weak after ECAP until a sufficient pre-annealing treatment is introduced, after which texture strength increases appreciably.In previous trials involving the processing of hyper-stoichiometric Nd 2 Fe 14 B with a well-dispersed rare earth-rich grain boundary phase, the timescale of processing via ECAP is adequate to develop a strong texture. 11,12However, since ECAP generally occurs on a much shorter timescale and lower temperature than die-upsetting (on the order of ∼100s), the processing of hypostoichiometric Nd 2 Fe 14 B/α-Fe with rare earth-rich additives must then require additional mechanisms such as annealing to aid the diffusion of the Nd-rich phase into the grain boundary.Without such additional mechanisms, the degree of Nd 90 Cu 10 diffusion after ECAP is severely reduced compared to that of die upsetting.Based on the diffusion coefficients 17 of 147 Nd, t=25h corresponds to a diffusion length of approximately 300µm, which just exceeds the average size of precursor powders (<250µm) in these experiments.After t=25h, further increases in pre-annealing time result in a decrease in texture strength and |BH| Max , supporting the notion that t=25h is sufficient for adequate Nd 90 Cu 10 diffusion.

B. The effect of varying Nd 90 Cu 10 content
Given the optimal properties at t=25h, compositions with x [NdCu] =4, 7, and 10wt% additions were investigated with this fixed annealing time.
Fig. 4 shows XRD patterns for each composition after ECAP.When x [NdCu] =4wt%, both the (110) and (211) peaks belonging to the α-Fe phase become very large.When x [NdCu] =7wt%, the (110) and (211) peaks decrease significantly, but still are quite prominent.At x [NdCu] =10wt%, the (110) and (211) α-Fe peaks have almost disappeared.Fig. 5 shows demagnetisation curves of field-aligned powders measured both parallel (solid curves) and perpendicular (dotted curves) to the applied field for x [NdCu] =4wt%, 7wt%, 7wt% with T ECAP = 550 • C and 10wt%.At x [NdCu] =4wt%, H c , M r and |BH| Max have decreased significantly.Nonetheless, an appreciable ∆M r still remains.As x [NdCu] increases, M s systematically decreases and appreciable M r and |BH| Max are observed.The shape of the demagnetisation curve remains relatively smooth, suggesting a reasonable preservation of exchange coupling for samples c) and d). 18ig.6 shows (001) pole figures for these same compositions and processing conditions.The observable  trend is similar to Fig. 5.The texture is strongest when x [NdCu] =10wt% and decreases with decreasing x [NdCu] as the (001) plane alignment becomes increasingly isotropic.For x [NdCu] = 7 t%, the effect of increasing temperature during ECAP was also investigated, and can be observed in Fig. 5, as well as the difference between b) and c) in Fig. 6.There is a slight improvement in H c , |BH| Max , and M r , while M s shows a slight These trends are similar to the trend between x [NdCu] =7 and 10wt%, the latter of which displays the strongest texture and least α-Fe.This suggests that further increasing T ECAP could further enhance texture and |BH| Max .
Table II shows the change in composition of W [αFe] after ECAP with varying x [NdCu] , as calculated by peak-fitting of Mössbauer spectra.An example of one of the Mössbauer spectra used for calculation is shown in Fig 7 .When x [NdCu] =4wt%, W [αFe] is as large as ∼24wt%, and then reduces to 0 as x [NdCu] is increased.These results are consistent with the trend observable in Fig. 4. At x [NdCu] =4wt%, peaks belonging to the α-Fe phase are quite prominent, and gradually reduce, to the point of almost entirely disappearing as x [NdCu] is increased.
While an appreciable ∆M r is kept even at x [NdCu] =4wt%, H c and |BH| max have significantly decreased, and the texture has appreciably diminished.Given the presence of a rare earth-rich eutectic grain boundary phase is generally required for the formation of texture in Nd 2 Fe 14 B, 4,8 such a decrease in strength of texture is expected when the amount of grain boundary phase is reduced.Nonetheless, a slight texture still remains at x [NdCu] =4wt%, even when a significant amount of α-Fe is present.It is also observable in Fig. 5 that as W [αFe] increases, M s also increases.The shape of the demagnetisation curve remains relatively smooth in the second quadrant, which may indicate a preservation of exchange coupling, however when x [NdCu] =10 wt%, W [αFe] is very low and this shape may be explainable by a single hard phase Nd 2 Fe 14 B composition rather than exchange coupling between hard and soft phases.Additionally, at x [NdCu] =4wt% a very significant loss of coercivity is observed.Ideally, a well exchange-coupled nanocomposite should allow for much larger amounts 2,19 of soft phase before significant losses in coercivity are observed, but even at W [αFe] ≈ 24wt%, significant coercivity loss is observed in these experiments, suggesting that some exchange coupling may have been lost.

IV. CONCLUSION
Initial texture can be produced in hypo-stoichiometric Nd 2 Fe 14 B/α-Fe nanocomposites with Nd-Cu based additives at a suitably low temperature with both a grain size below 100nm and a significant amount of α-Fe remaining after thermomechanical processing.Given the short timescale of ECAP, a pre-annealing treatment aids diffusion of the Nd-Cu grain-boundary phase, further enhancing the texture.The degree of texture increases as the amount of Nd 90 Cu 10 added increases, however the amount of residual α-Fe also decreases to almost zero as more Nd 90 Cu 10 is added.Future work would involve further in-depth analysis of the nanostructure and further optimisation of Nd 90 Cu 10 diffusion to further enhance texture and increase |BH| Max while retaining α-Fe.Nonetheless, slight texture can be developed still with an appreciable amount of α-Fe remaining without incurring significant grain growth, all important necessary preconditions that must be maintained in view of the development of bulk-scale anisotropic exchange coupled Nd 2 Fe 14 B/α-Fe nanocomposites.
11,12r scale nanocomposites, bulk processing routes which do not incur grain growth must be investigated.Equal Channel Angular Pressing (ECAP) has been used successfully to develop texture in Nd 2 Fe 14 B-based materials without appreciable grain growth.11,12Inthis work, Nd 2 Fe 14 B/α-Fe nanocomposite materials are combined with Nd-Cu additives to develop suitable texture via ECAP while maintaining a grain size of approximately 50nm.

TABLE I .
Changes in grain size of Nd 2 Fe 14 B+10 wt% Nd 90 Cu 10 after ECAP. a a numbers in parentheses reflect absolute error.b (Am 2 /kg).c (kJ/m 3 ).

TABLE II .
The effect on composition (wt% α-Fe) after ECAP of varying wt% Nd 90 Cu 10 added, as determined by Mössbauer spectroscopy.