The role of Al in rapidly solidified FePrBAl permanent magnets

~

ELSEVIER

Journal of Magnetism and Magnetic Materials 137 (1994) 18-24

,~

journalof magnetism and magnetic malerials

The role of A1 in rapidly solidified FePrBA1 permanent magnets M. Seeger, H. Kronmiiller * Max-Planck-Institut fiir Metallforschung, Institut fiir Physik, Heisenbergstr. 1, D-70569 Stuttgart, Germany Received 20 January 1994; in revised form 23 March 1994

Abstract

Magnetic and microstructural investigations were performed on melt-spun permanent magnets of compositions Fevs_xPr15BvAlx to check the influence of the addition of A1 on melt-spun FePrB-based permanent magnets. The microstructural parameters a K and N ~ t t , which describe the effects of the non-ideal microstructure, were determined from the temperature dependence of the critical field. Both parameters show significant increases with increasing A1 concentration, i.e. the improvement of the grain surfaces is compensated by increased local demagnetization fields. Long annealing treatments lead to a deterioration of the hard magnetic properties; this is correlated with an increase in the Neff parameter and a nearly constant a K value. These magnetic results were compared with SEM investigations carried out on cross-sections of the melt-spun ribbon flakes.

1. Introduction Since the discovery of the tetragonal Fe14Nd2 B phase [1,2] which is suitable for the production of p e r m a n e n t magnets because of its large values of the saturation polarization Js and crystal anisotropy, many additional elements have been investigated in this class of p e r m a n e n t magnets in order to improve the hard magnetic properties with respect to the Curie t e m p e r a t u r e To coercive field /x0H c a n d / o r remanenee JR" A summary of these investigations of dopants is given in the review article of Herbst [3]. In general, these elements change either the intrinsic material parameters (Tc, Js, anisotropy constants K1, K 2 . . . ) or the microstructure (grain size, shape of the

* Corresponding author.

grains, quality of the grains, decoupling ability, etc.). Many investigations have been done with the element AI in sintered magnets and in single crystals [4-14]. In these investigations a partial solution of A1 in the hard magnetic phase and therefore variations of the intrinsic parameters could be observed. For instance, Hock [9] found a reduction in Tc by 1 1 K in the composition (Fe0.gsA10.02)14NdzB. Furthermore, new phases in the boundary region of sintered magnets were detected (e.g. the 3 phase with composition Fe70_xNd30Alx, 7 < x < 2 5 [14]). These new phases are responsible for a significant improvement in the hard magnetic properties, especially the coercivity, of such Al-doped sintered magnets [7,14]. The aim of this work was to show whether or not the addition of AI has a similar beneficial influence on the hard magnetic properties of melt-spun 1 4 : 2 : 1 magnets. Our investigations were performed on rapidly solidified magnets of

0304-8853/94/$07.00 © 1994 Elsevier Science B.V. All rights reserved SSDI 0 3 0 4 - 8 8 5 3 ( 9 4 ) 0 0 3 2 8 - 0

M. Seeger, H. Kronmiiller/Journal of Magnetism and Magnetic Materials 137 (1994) 18-24

nominal compositions Fe78_xPr15B7Alx with x = 0, 4 and 6. To obtain quantitative information about the microstructure of a nucleation hardened permanent magnet we analyze the temperature dependence of the coercive field using a modified form of Brown's equation for the nucleation field of a homogeneously magnetized particle [15-18]: /z0Hc

= aK/LoH~

in -- N e f f J s .

(1)

The microstructural parameter a/~ describes the influence of the reduced crystal anisotropy at the surface of the grains due to the disturbed crystal lattice, and the so-called minimum nucleation field / . t , 0 H ~ ain denotes the value for the nucleation field of the most unfavorably aligned grains belonging to a misalignment angle of about 45° according to the Stoner-Wohlfarth model [19]. The effective demagnetization factor Neff arises predominantly at the sharp edges of the grains. Neff values of sintered and melt-spun 14 : 2 : 1 based magnets are usually in the range 1.0-2.0. These values also were confirmed by theoretical calculations applying the finite-element method using a model Qf grains with square cross-section [20]. According to Eq. (1), a plot of iXoI-Ic/J s min versus/x0H N /Js yields the aK and Neff values.

19

curve in the range 160 K < T < 540 K were measured. Instead of the coercive field the so-called critical field/x0Hcrit is analyzed, which is defined by the maximum of the susceptibility. The magnetic field was applied parallel to the ribbon direction, therefore the macroscopic demagnetization factor was N < 10 -2. The external field was corrected by the demagnetization field. To correlate the magnetic results with the microstructure, cross-sections of cut ribbon flakes were analyzed with the SEM (scanning electron microscopy) technique.

3. Experimental results and discussion Fig. 1 shows the room-temperature hysteresis loops including the virgin curves of the series Fe78_xPr15B7Alx. Because of the higher sensitivity of the determination of the mass of a ribbon flake compared to its volume, Fig. 1 presents the magnetization in emu/g. The virgin magnetization curves show a two-step behavior that can be explained in the framework of the nucleation model by a variation of the grain size in the range of the critical radius of single-domain particles. The large grains which present multidomain behavior in the thermally demagnetized state are easily magnetized in moderate applied fields

2. Experimental background Details of the sample preparation and the parameters of the melt-spinning process are given elsewhere [18]. The wheel velocity was chosen to be 18-20 m/s. AI was added in form of an FeAI prealloy. The samples of the series Fe78_xPr15B7 AI x (0 < x < 6) were heat treated under high vacuum at a vessel temperature of T = 700°C for t a = 15 min to yield a homogeneous microstructure without amorphous soft magnetic regions. In a second series, magnets of the composition Fe74PrlsB7AI4 were annealed for different times in the range 15 min < t a _< 240 min at the same vessel temperature. The magnetic measurements were carried out with single ribbon flakes in a SQUID magnetometer, model MPMS. Roomtemperature hysteresis loops and the temperature-dependent behavior of the demagnetization

Fe78_xPr15B7AIx '

'

i

x ×

E

~

o

o

I

= =

i

0 4

I

I

7 15

I

-

0

#0 H

5

[T]

Fig. 1. Hysteresis loops at T = 300 K in the composition range Fe7s_xPr15B7Alx (0 < x < 6).

M. Seeger, H. Kronmiiller / Journal of Magnetism and Magnetic Materials 137 (1994) 18-24

20

(/z0H _< 0.5 T). The smaller single-domain grains rotate their magnetization in field values comparable to the coercive field of the sample. The sizes of the two steps provide measures of the proportions of multidomain and single-domain particles, respectively. In the Al-free sample the second magnetization step is slightly larger than in the M-containing samples corresponding to smaller grains. The room-temperature critical field values of the samples investigated show no variation (/z0Hcrit = 2.4 T), but the magnetization and therefore the remanence decreases significantly with increasing A1 content. The Al-ffee composition has a remanent magnetization of M r = 82 e m u / g and the values for the 4 and 6 at% Al-containing samples are 60 and 51 e m u / g , respectively. This means a nearly linear decrease with increasing x values. Fig. 2 shows the temperature dependence of the critical field. It can be seen that the three compositions are characterized by different behaviors. The Al-containing samples have a stronger temperature dependence, i.e. a more marked diminution of the critical field at elevated t e m p e r a t u r e s . T h e t e m p e r a t u r e coefficient (dHcrit/dT)/Hcrit determined at T = 300 K reflects this fact. The value for the composition Fe72PrlsBTAI 6 ((dncrit/dT)/ncrit = - 0 . 8 1 % / K ) is about 17% larger than that for the Al-free

Fe78_xPrlsB7AIx to

i

. . . .

i

. . . .

i

. . . .

~

i

x =0 x-4

.

.

.

I

.

.

.

.

.

[

.

.

.

I

.

.

.

.

I

I

i

--

Fe14Pr2B

!~,~.

--

Fe14Nd2 B

....

(geo.98Alo.02) 14Nd2B

o •

E z I

.

i

.\ \

o 100

200

300

400

500

T [K] Fig. 3. The minimum nucleation field/xoH~'i"(T) of Fe14Pr2B (data from Ref. [21]), Fel4Nd2B and (Feo.,~sAlo.o2)14Nd2B (data from Ref. [9]) calculated with Eq. (2). composition Fe78PrlsB 7 ((dHcrit/dT)/Hcrit = - 0 . 6 9 % / K ) . This property obstructs technical applications of these materials. Nevertheless, the curves cross at about 300 K, in accordance with the results in Fig. 1. Furthermore, these differences should be reflected in changes of the microstructural parameters o~K and N~ff. These parameters are determined from the slope and the ordinate intersection of the balance straight line min in a I,ZoHcrit/J s versus /zoH N /Js plot, i.e. in a plot of experimental versus theoretical data. Here tzoH~ i" denotes the minimum nucleation field of the hard magnetic Fe~4Pr2B grains. Under the assumption of small K 2 values, which is valid in Fel4Pr2 B, this quantity can be calculated approximately via the relation [17]: H ~ in --

Kl + K 2

(2)

Js The data for the intrinsic material parameters and Js for the pure Fe~aPr2B phase were drawn form Ref. [21]. The temperature dependence of /z0H~ i" of Fel4Pr2B is shown in Fig. 3. A possible influence of A1 dissolved in the hard magnetic phase on the temperature-dependent behavior of # o H ~ in has been neglected. The validity of this neglect has been verified in the Fe14NdzB system. Fig. 3 also shows the quantity # o H ~ in for the compositions Fel4Nd2B and (Fe0.9sA10.o2)14Nd2B, respectively, with data of K l +K 2

b

32

o 200

300

400 T

500

[K]

Fig. 2. The critical field /xoHcrit(T) in the composition range Fe78PrlsBTA1x (0 < x _<6).

M. Seeger, H. Kronmiiller /Journal of Magnetism and Magnetic Materials 137 (1994) 18-24

the intrinsic parameters drawn from Ref. [9]. In the temperature range in which the microstructural parameters have been determined (240 K < T < 500 K) the absolute values of the Al-containing composition are slightly reduced compared to those of the pure FelaNd2B system, but the temperature dependence does not change. According to this result, the influence of AI on the behavior of /z0H~in(T) in the Fel4Pr2B system has not been considered. The evaluation of the parameters aK and Neff can be seen in Fig. 4 for the composition Fe74PrtsB7A14 . In Fig. 5 we present the results for the dependence of microstructural parameters on the AI concentration. There are significant increases in both aK and Neff with increasing A1 content. The improvement of larger a K values and therefore a better grain surface quality, however, is overcompensated by the drastic increase in Neff, which leads to the steep decrease in/./,0ncrit a t elevated temperatures. The variation of the temperature dependence of/z0ncrit with varying AI content shown in Fig. 2 can be demonstrated more clearly by plotting the reduced quantity hcrit(Z) = l ~ o n c r i t ( T ) / l ~ o H ~ in (T), i.e. the ratio of the experimental data divided by the theoretical data. This quantity should be close to one for a perfect permanent magnet. In Fig. 6, hcrit(T) is plotted for the different compositions. It turns out that in Fe72PrtsBTAI 6

Fe78_xPr15B7AIx !

•

~K

~)

Nef f

J J

J /

Z /

i

,

__

/

--

-0

/

j/ it-/

C I

i

I

0

~

I

2

~

I

4

6

X

Fig. 5. Dependence of the microstructural parameters a K and Neff on AI concentration x in the system Fe78_xPrlsBTA1 x. The dashed lines are guides for the eye.

the critical field at T = 400 K corresponds to only 34% of the theoretical value, compared with 40% in the case of Fe78Pr15B7. These changes in the magnetic properties should also be visible in the microstructure of the magnets examined. Fig. 7 shows a SEM micrograph of the composition Fe78Pr15BT. More or less spherical grains of some few hundred nm in diameter can be seen. It should be noted that in general the grain size is not homogeneously dis-

Fe78_xPrlsByAI x

Fe74Pr15B7AI4 ,

21

i

a k = 0.91

/

.

.

/

.

.

.

i

.

.

.

.

i

.

.

.

.

i

.

.

.

.

i

.

.

.

.

d "

Neff

& T

i

200

o 2

0

•

•

.

300

400

500

4

T [K] ~oHNrnIn//Js mln Fig. 4. P,0Hcrit/Js versus //,oH N / J s to determine a/¢ and

Fig. 6. The reduced critical field h(T)=lzoHcrit(T)/iZo H~in(T) in the composition range Fe78_xPr15ByAI x (0 < x <

Nef

6).

f

of

Fe74PrlsByAI4.

22

M. Seeger, tl, Kronmiiller /Journal of Magnetism and Magnetic Materials" 137 (1994) 18-24

Fig. 7. SEM micrograph of the composition Fe78PrIsB 7.

tributed but that there is a gradient between the wheel side (small grains) and the free surface side (larger grains) of the cross-section of the ribbon

flakes [2,18]. This effect is indicated in Fig. 7, where the grains of the upper part of the micrograph can be compared with the other ones. In

Fig. 8. SEM micrograph of the composition Fe72Pr~sBTA16.

M. Seeger, H. Kronmiiller/Journal of Magnetism and Magnetic Materials 137 (1994) 18-24

contrast, in the composition Fe72PrlsB7A16 (Fig. 8) the grains exhibit more irregular shapes with very marked edges and corners, which are responsible for the large Neff values in the Al-containing compositions. Nevertheless, the average grain size has not changed significantly. However, the effect of the addition of Al to permanent magnets prepared by the melt-spinning process is not as favorable as compared to sintered magnets. In a further series we checked the influence of the annealing treatment on the magnetic properties of A1 containing melt-spun samples. In general, overquenched melt-spun permanent magnets are optimized by a short annealing treatment (/a < 15 min) during which the amorphous phase is crystallized [18,22-25]. We varied the annealing time of the composition Fe72Pr15B7A16 in the range 15 min < t a < 240 min at the vessel temperature Ta = 700°C. The intention was to smooth the surfaces of the grains by increasing the duration of the annealing treatment and therefore to reduce the demagnetization effects. The hysteresis loop of a sample annealed for t a = 240 min (Fig. 9) exhibits a ~0ncrit value comparable with that of the sample annealed for 15 min (~0ncrit = 2.2 T), but the shape of the hysteresis loop is less rectangular with a more extended field range of demagnetization. In contrast to the hysteresis

Fe74Pr15B7AI4 i

i

o o

23

Fe74Pr15B7AI4 .

.

.

.

I

.

.

.

/

O-tl--

i

.

I

'

~

I~-

L

,

i

i

--

I

.0

'

~

'

Nef f

- -

,

,

i

1O0

i

I

200

i

i

I

i

300

k [min] Fig. 10. The microstructural parameters aK and Neff of the composition Fe74Pr15B7A14 in dependence of the annealing time t a. The dashed lines are guides for the eye.

loop of Fig. 1, the irreversible demagnetization process starts at small reversed fields o f / z 0 H = -0.5 T. Furthermore the virgin curve indicates grain growth during the annealing treatment, the second magnetization step has nearly disappeared in favor of a large initial susceptibility due to multidomain particles. The result of the evaluation of the microstructural parameters for the different annealing times is represented in Fig. 10. Up to t a = 120 min Ncff increases without any change in t~r, indicating a deterioration of the magnetic properties. For t a ---240 min both parameters decrease. It should be noted that the determination of the microstructural parameters is restricted to the critical field values and the poorer quality of the whole demagnetization curve is not considered in this method. From these results we conclude that a short annealing treatment ( t a = 15 min) is most beneficial with respect to the magnetic properties in the case of Al-containing FePrB-based magnets.

o I

4. Summary 15 -

,

0

,

,

,

I 5

IZoH [T] Fig. 9. Hysteresis loop at T = 300 K of the composition Fe74Pr15B7AI4 annealed for t a = 240 min.

We have investigated magnetic and microstructural properties of rapidly solidified FePrBbased permanent magnets with the addition of A1 in the composition range Fe78_xPrlsB7Alx

24

M. Seeger, tl. Kronmiiller /Journal of Magnetism and Magnetic Materials 137 (1994) 18-24

(x = 0, 4, 6). The main results are as follows: A1 addition does not change the critical field values at room temperature. The remanence of Al-doped melt-spun magnets decreases with increasing AI concentration. The temperature coefficient (dHcrit/dT)/Hcrit at room temperature increases from - 0 . 6 9 % / K to - 0 . 8 1 % / K with increasing AI addition. This difference is correlated with changes in the microstructural parameters a K and Neff, both of which increase after the addition of Al. The shape of the grains in the Al-containing states is less spherical, thus leading to remarkable effective demagnetization effects. A long annealing treatment does not reduce these demagnetization effects. -

-

-

-

-

-

A c k n o w l e d g e m e n t s

The authors wish to thank Dr R. Reisser, Dr D. K6hler and Dipl.-Phys. J. Bauer for helpful discussions, and M. Kelsch for performing the SEM investigations. This work was supported by the Commission of the European Community within the B R I T E / E U R A M programme.

R e f e r e n c e s

[1] M. Sagawa, S. Fujimura, M. Togawa, H. Yamamoto and J. Matsuura, J. Appl. Phys. 55 (1984) 2083. [2] J.J. Croat, J.F. Herbst, R.W. Lee and F.E. Pinkerton, J. Appl. Phys. 55 (1984) 2078.

[3] J.F. Herbst, Rev. Mod. Phys. 63 (1991) 819. [4] M. Zhang, D. Ma, X. Jang, S. Liu, Proc. 8th Int. Workshop on Rare Earth Magnets and their Applications, Dayton, OH (1985). [5] P. Schrey, IEEE Trans. Magn. 22 (1986) 913. [6] S, Hirosawa, Y. Yamaguchi, K. Tokuhara, H. Yamamoto, S. Fujimura and M. Sagawa, IEEE Trans. Magn. 23 (1987) 2120. [7] T. Mizoguchi, I. Sakai, H. Niu and K. Inomata, IEEE Trans. Magn. 23 (1987) 2281. [8] M. Tokunaga, H. Kogure, M. Endoh and H. Harade IEEE Trans. Magn. 23 (1987) 2287. [9] S. Hock, Thesis, University of Stuttgart (1988). [10] B. Grieb, K.G. Knoch, E.-Th. Henig and G. Petzow, J. Magn. Magn. Mater. 80 (1989) 75. [11] J. Chen and H. Kronmiiller, Phys. Stat. Solidi (a) 120 (1989) 617. [12] J. Fidler, K.G. Knoch, H. Kronmiiller and G. Schneider, J. Mater. Res. 4 (1989) 806. [13] W. Rodewald and B. Wall, J. Magn. Magn. Mater. 101 (1991) 338. [14] B. Grieb, C. Pithan, E.-Th. Henig and G. Petzow, J. Appl. Phys. 70 (1991) 6354. [15] H. Kronmiiller, K.-D. Durst und M. Sagawa, J. Magn. Magn. Mater. 74 (1988) 291. [16] G. Martinek and H. Kronmiiller, J. Magn. Magn. Mater. 86 (1990) 177. [17] H. Kronmiiller, J. Magn. Soc. Jpn. 15 (1991) 6. [18] M. Seeger, D. K6hler and H. Kronmfiller, J. Magn. Magn. Mater. 130 (1994) 165. [19] E.C. Stoner and E.P. Wohlfarth, Phil. Trans. R. Soc. 240 (1948) 599, [20] H.F. Schmidts, Thesis, University of Stuttgart (1992). [21] S. Hirosawa, private communication. [22] A. Manaf, H.A. Davies, R.A. Buckley and M. Leonowicz, J. Magn. Magn. Mater. 104-107 (1992) 1145. [23] J.M. Gonzalez, F. Cebollada, V.E. Martin, M. Leonato, A. Hernando, E. Pulido and P. Crespo, J. Magn. Magn. Mater. 104-107 (1992) 1179. [24] M.T. Clavaguera-Mora, J.A. Diego, M.D. Bar6, S. Surifiach, A. Hernando, P. Crespo and G. Rivero, J. Magn. Magn. Mater. 119 (1993) 289. [25] D. K6hler, Thesis, University of Stuttgart (1992).