FIM-ATOM PROBE INVESTIGATION OF MELT-SPUN PERMANENT MAGNETS BASED ON NdFeB

HAL Id: jpa-00229955

https://hal.archives-ouvertes.fr/jpa-00229955

Submitted on 1 Jan 1989

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

FIM-ATOM PROBE INVESTIGATION OF

MELT-SPUN PERMANENT MAGNETS BASED ON NdFeB

A. Hütten, P. Haasen

To cite this version:

A. Hütten, P. Haasen. FIM-ATOM PROBE INVESTIGATION OF MELT-SPUN PERMANENT MAGNETS BASED ON NdFeB. Journal de Physique Colloques, 1989, 50 (C8), pp.C8-337-C8-342.

�10.1051/jphyscol:1989857�. �jpa-00229955�

COLLOQUE D E PHYSIQUE

Colloque C 8 , Supplkment au n 0 l l , T o m e 50, novembre 1989

FIM-ATOM PROBE INVESTIGATION OF MELT-SPUN PERMANENT.MAGNETS BASED ON NdFeB

A. H ~ T T E N and P. HAASEN

Institut fiir Metallphysik d e r Universitdt Gdttingen and Sonderforschungsbereich 126, D-3400 Gdttingen, F.R.G.

Abstract

-

The microstructure of Nd15Fe7,B8 in the optimum magnetic state is compared to that of overquenched and heat treated specimens using Field Ion Microscopy (FIM) and Atom Probe (AP) techniques. Apart from the hard magnetic Nd2FeI4B - ( (P ) - and the nonmagnetic Nd,,,Fe4B4 - ( 11 ) - a metastable Nd7Fe3 - phase controls the magnetic properties of melt-spun NdFeB - ribbons. Furthermore the influence of a partial F e substitution by CO on the microstructure was investigated and is presented in this report.

I - INTRODUCTION

The technological potential of NdFeB permanent magnets is based on the Nd2Fe14B line compound which was discovered in 1983 /l. 2/. Boron in combination with Nd and Fe stabilizes the tetragonal Nd2Fe14B - structure /3/. Whereas its Fe - sublattice in principle determines the Curie temperature and the saturation polarization the Nd

-

sublattice is mainly responsible for the huge uniaxial magneto- crystalline anisotropy. The efficiency to which these quantities lead to good permanent magnetic properties as coercivity. remanence and maximum energy product depends very strongly on the microstructure of the particular material. Therefore it is necessary to determine the exact nature of minority phases for understanding the role which each phase plays in the magnetization reversal process.

In the present study we report on the microstructure of isotropic Nd15Fe77B8 melt

-

spun ribbons.

These specimens made by rapid solidification are metallurgically interesting because this technique provides the possibility to bypass the phase relations given by the equilibrium phase diagram. Fwther- more the change of the microstructure caused by partial substitution of Fe by CO in a series of isotropic N ~ ~ ~ ( F ~ ~ - , C O , ) ~ , B ~ melt - spun ribbons with ^{X g} 0.6 is discussed.

I1 - EXPERIMENTAL

Melt

-

spun ribbons of N ~ ~ ~ ( F ~ ~ - ~ c o ~ ) ~ ~ B ~ with X ranging from 0 to 0.6. 15 - 40 um thick and about 20 - 100 mm in length were provided by Siemens AG, Erlangen. The coercivity and hence the micro- structure of the specimens produces by this technique a r e sensitive to the quench rate which is controlled

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1989857

by the surface velocity of the copper wheel as illustratred in fig.1. Ribbons quenched into the optimum magnetic state which is characterized by a maximum coercivity were obtained at substrate velocities of about vWheel = 15 &S. The specimens quenched at the highest rate were almost amorphow as indicated by a diffuse maximum in the X - ray diffraction pattern /4/ and henceforth referred as "overquenched"

in the text. As sketched in fig.2 FIM tips were made from ribbons by first mechanically grinding them to a quadratic cross

-

section followed by electropolishing in perchloric acid and ethyl alcohol in the proportion 1:30. Due to the inhomogeneous cooling across the thickness of ribbons in the optimum magnetic state grains of up to 300 nm in diameter can be found at the free surface /4/ whereas grains of about 20

-

30 nm in diameter were typically for the wheel side of the ribbons /S/. The big grains were preferentialy removed from the tip during electropolishing by a preferential thinning at Nd-rich grain boundaries. As a consequence the microstructure of the ribbons which is available for FIM and AP techniques is restricted to a region close to the bottom side of the ribbons. The FIM and AP studies were performed on the Gottingen instrument described by Wagner /6/. The tips were cooled to about 8 0 K. For the AP studies a high

-

voltage pulse ratio of about 15% was sufficient to prevent preferential evaporation of one component. To improve the imaging conditions and hence the phase contrast of specimens in the optimum magnetic state a (l-X) Ne + X Ar

-

gas mixture of 5.10-4 Pa with X about 0.1 was used, whereas only Ne was used during AP analysis. As shown in fig.3a no well developed crystalline ring structures a r e visible in the field

-

ion micrograph of the single - phase tip. The question arises for what reason the field

-

ion image looks amorphous, although the specimen certainly was crystalline from the magnetic behaviour /4/. lkerefore computer

-

simulations of field

-

ion micrographs were performed /S/. A simulated image of Nd2Fe14B in C0021

-

direction with about the same tip radius and the same brightly imaging spot density as in fig.3a is shown in fig.3b. Apart from the (002) - pole in the middle no other poles a r e visible. It must be concluded that the complexity of the Nd2Fe14B unit cell is responsible for the underdeveloped ring structures of the field

-

ion micrograph. This effect is probably magnified by the great difference in atomic sizes of the components of the alloy. Therefore, no information about the crystalline state of the specimens could be obtained from field

-

ion images.

In the AP mass spectra B2+. B+. Fez*. Fe+, Nd3+. Nd2+ and Nd+ were usually present. Most of the Nd atoms were field evaporated as Nd2+ and Nd3+. In spite of the fact that the imaging conditions were comparable in some of the mass spectra additional Nd - complexes were present as ions. In order to detect the right composition of those ions mass spectra were simulated /5/ and shown in fig.4. Evidently the appearance of Nd

-

complexes is correlated with the presence of ions with mass

-

to - charge ratios between 12 to 22. For the simulation of these ions their percentage in the main chamber vacuum determined by a residual gas analyzer was taken into account. The qualitative agreement with the measured mass spectrum in the same range led to theconclusion that residual gas atoms a r e absorbed on the tip. Therefore the presence of Nd - complexes could be explained by chemical interactions between absorbed and substrate atoms influenced by the electric potential close to the tip surface. On the other hand these Nd - complexes also were detected in sintered NdFeB specimens /7/.

I11 - RESULTS

AND

DISCUSSIONS

Fig.5 shows the microstructure of Nd15Fe7,BB in the optimum magnetic state. According to selected area analysis the darkly imaging area 1 has the stoichiometric composition of the hard magnetic Nd2Fe14B phase. Area 2 which is brightly imaging surrounds the first one and was detected as Nd7Fe3 As one can see in fig.5 the hard magnetic @

-

grains are separated by the Nd

-

rich phase acting as a grain boundary phase. It is commonly agreed that this phase is nonmagnetic a t room temperature due to the high amount of Nd. Hence this Nd7Fe3 phase diminishes intergranular magnetic exchange interactions of hard magnetic grains. By comparison the morphology of overquenched NdlSFe7,B8 specimens is given in fig.6al. Despite the quenching rate it i s already decomposed on a very fine scale. Again darkly imaging regions which were Fe

-

rich a r e separated by brightly imaging ones corresponding to I1 in fig.6b. Additional B - enriched areas which are indicated by brightly imaging spots (see fig. 6 a 4 )

were obtained during random area analysis. The Fe

-

rich areas (see I in fig. 6b) have already compositions comparable to the Q, - phase, whereas those of the Nd

-

enriched regions are comparable with the liquid phase composition which is determined in sintered specimen cooling from 1060°C - 1000°C /S/.

Therefore the concentration fluctuations seen in fig. 6b may be interpreted as early states of decomposition during melt - spinning. As melt is cooling down to 1000°C small ares of Nd2Fe14B are formed simultaneously and cause an increase of Nd

-

concentration in their neighbourhood. Then, at about 700°C Nd

-

enriched regions are frozen into the detected composition. The sluggishness of the formation of the B

-

rich q

-

phase which is presented in the equilibrium diagram (see fig. 7) compared to those of the Nd2Fe14B

-

and Nd

-

enriched phases can be understood because it is kinetically suppresed in the quench process due to a very complicated unit cell / 9 /

.

This is verified by investigations of overquenched Nd15Fe77B8 - specimens which were annealed at 530°C for 30 min. Apart from the @ - and Nd7Fe3

-

phases a third one described as Ndl.lFe4B4 was determined / S / . Obviously the B

-

rich phase is now due to long

-

range diffusion which is necessary for the formation of this phase. The Curie temperature of the 9 - phase is about 14 K /10/. Hence these room temperature nonmagnetic grains produce magnetic stray fields opposite to the spontaneous magnetization in the neighbouring Q!

-

grains, thus reducing their coercivity / S / . Random or selected area analysis of Nd15Fe77B8 - specimens in the optimum magnetic state as well as overquenched and annealed ribbons revealed that the intergranular Nd

-

rich phase is not one of the equilibrium phases in the NdFeB isothermal diagram (fig. 7). Hence the question arises whether the detected Nd7Fe3 phase is stabilized by impurities or is a metastable one of the NdFeB - system. Taking into account that the origin of the sometimes detected Nd - compounds is not clear, as discussed above, we suppose that Nd7Fe3 is metastable in contrast to the stable Nd95-97Pe3-5 phase in sintered magnets /U/.

With CO substitution the formation of N ~ ~ ( F ~ ~ - , C O , ) ~ ~ B and N ~ ~ ( F ~ ~ - , C O , ) ~ is expected especially for small CO

-

additions assuming a complete substitution of Fe by CO in these phases. For specimens near ^X

*

1 other phases than the above should be found due to the fact that the NdCoB

-

system /12/ looks quite differently from NdFeB

.

Fig. 8 shows the microstructure of specimens in the optimum magnetic state with increasing CO

-

content. A similar phase contrast in the field - ion micrographs 1-4 allowed us to distinguish a darkly imaging phase from a brightly imaging one surrounding the former. The measured Fe concentration of the darkly imaging phases was found to decrease linearly by simultaneous incorporation of Co. These results were obtained by selected area analysis are in good agreement with those expected from N ~ ~ ( F ~ ~ - , C O , ) ~ ~ B . This means that the degree of the Fe substitution by CO in the hard magnetic phase is controlled by the nomial composition in the studied concentration range. As can be seen from fig. 9 the composition of the brightly imaging second phase cannot be described in one formula which takes into account just the Fe

-

decrease as well as the CO

-

increase.

It is roughly described as N d 2 3 ( ~ e l - x ~ o , ) 7 5 ~ 2 concerning the Fe - and CO

-

ratio. This phase is ferromagnetic for ^{X L} 0.07 after DSc investigations /S/. Thus is may be possible that N ~ ~ ~ ( F ~ , - , c o , ) ~ ~ B ~ produces a magnetic shielding which weakens the influence of magnetic stray fields caused by neighbouring 9

-

grains. For X 0.025 the N d 2 3 ( ~ e l - x ~ ~ x ) 7 5 ~ 2

-

is replaced by a N ~ ~ ( F ~ , C O ) ~

-

phase. Therefore it may be concluded that the former is stabilized by CO and changes to the latter caused by a CO

-

content less than 10% (see fig. 9). In fact the N ~ ~ ~ ( F ~ ~ - ~ c ~ ~ ) ~ ~ B ~ - phase is metastable as evidenced by N ~ ~ ~ ( F ~ ~ - ~ c ~ ~ ) ~ ~ B ~ - specimens which were annealed at 630°C up to 10 min. AP analysis yielded the presence of ~ d ( ~ e ~ - , C o ~ ) ~ instead of Nd23(Fel-xCox)75B2. Hence the latter is stabilized by the kinetics of melt

-

spinnig into the optimum magnetic state. Apart from N ~ ~ ( F ~ ~ - , C O , ) ~ ~ B and N ~ ~ ~ ( F ~ ~ - ~ C ~ ~ ) ~ ~ B ~ a third phase determined as Ndl.l ( F ~ ~ - , C O , ) ~ B ~ which is indicated by brightly imaging spots (see fig. 8, 5) was found only for ^X = 0.05.

We therefore conclude that three phases are present and responsible for the magnetic values of the permanent magnet and at least one of them changes drastically with CO substitution.

The authors thank Dr. J. Wecker ( Siemens AG, Erlangen ) for the provision of the material and Dr.

R. Wagner and Dr. M. Oehring for stimulating discussions.

References

Fig. 1 - Coecive force HCi as a function substrate velocity vWheel /4/.

/ l/ M. Sagawa, S. Fujimura, N. Togawa, H. Yamamoto and Y. Matsuura (1984) J. Appl. Pyhs. 55. 2083.

/ 2 / J . J. Croat. J. F. Herbst. R. W. Lee and F. E. Pinkerton (1984) J. Appl. Phys. 55, 2078.

/ 3/ 5. F. Herbst, J. J. Croat, F. E. Pinkerton and W. B. Yellon (1984) Phys. Rev. B29, 4176.

/ 4 / 5. Wecker and L. Schultz (1987) J. Appl. Phys. 62. 990.

/ 5 / A . Hiitten (1989) Ph. D. Thesis, Gottingen.

/ 6 / R . Wagner (1982) Field Ion Microscopy in Material Science, Crystals Vol. 6. Springer Verlag, Berlin.

/ 7/ J. Fidler (1987) IEEE Trans. Magn. MAG-23, 2106.

/ 8 / G. Schneider. E.

-

T. Henig, H; H. Stadelmaier and G. Petzow (1987) Proc. 9th Int. Workshop on Rare - Earth Magnets and their Applications, Part 11, Bad Soden

.

^347.

/ 9 / R . K. Mishra (1989) J. Appl. Phys. 64, 5562.

/10/ D. Givord, J. M. Moreau and P. Tenaud (1985) Solid State Commun. 55. 303.

/11/ M. Sagawa, S. Hirosawa, H . Yamamoto, S. Fujimura, Y. Matsuura (1987) Jap. J. Appl. Phys. 26, 785.

/12/ N. S. Bilonizhko and Yu. B. Kuz' ma (1983) Izv. Akad. Nauk SSSR Neorg. Materialy 19, 487.

Nd15Fe77 BB ^me L t-spun r i bb

Fig. 3

-

Comparison between

a.) 0.92 Ne + 0.08 Ar field

-

ion image of the Nd2FeI4B - phase b.) computer simulated image of the Nd2FeI4B

-

phase in [002] - direction.

The bigger spots indicate the B - atoms.

30

20

-

z

a _-

2

10 -

0

0 10 20 30 40

/ 0-

[ 1

^"\\

/

_\

opt ;mum magnet i c

'

s t a t e \

\ \

\ overquenched

\ '\

\

1

,

^{I Y}

-

c e n t r a l p o r t ion d9 =80 nrn

FIM t i p

Fig. 2

-

FIM tip preparation

Fig. 4

Mass spectra of Nd15Fe77Bs obtained during AP analysis compared to those

obtained by

n simulated

J 200 Z

Fe'lFeH' simulations.

N~IOHI:' 100 -

0-+4

..,...,...,...,III.IIIIII...~,.

0 L0 60 80 100 170

mln [amul

Fig. 5

-

Field

-

ion micrograph of Nd15Fe77B1) in the optimum magnetic state.

1

10 nm

Fig. 6

-

a.) Field which correspond to the marked positions - ion images of overquenched Nd15Fe77B8 1 an 4 ⁰

1,

^{,, , ,,100 ,,}

t2

^{G ,, m200 m 7 G ,r 300 n n m}

1,

^{,m ,400 ,7 , ,}

l,,

^{, 500}

in the concentration profiles. b.) Number o f desorbed layers b.) Concentration profiles of overquenched Nd15Fe77B8.

Nd-Fe-B Phasediagram

Fig. 7 - NdFeB - equilibrium diagram (continuous straight lines) in contrast to AP

-

results (dashed lines).

B

Fe 2 0 4 0 60 80 Nd

Nd7Fe,

10 nm 10 nrn 10 nm

Fig. 8

-

Field

-

ion micrographs of N d 1 5 ( F e l - x ~ ~ x ) 7 7 B s - ribbons in the optimum magnetic state.

Fig. 9

-

Nd -, Fe -, CO -, B

-

concentration in the N d 2 3 ( ~ e l - , ~ o , ) 7 5 B 2 - phase as a function of CO - content.