Corrosion Resistance of NdDyFeB Basic Alloys
Korozijska obstojnost osnovnih zlitin NdDyFeB
S. Kobe Beseničar1, IJS Ljubljana, Slovenija L. Vehovar, IMT Ljubljana, Slovenija B. Saje, Magneti d.d. Ljubljana, Slovenija
Prejem rokopisa - received: 1996-10-01; sprejem za objavo - accepted for publication: 1996-11-04
Nd-Dy-Fe-B-X (X = Zr, Hf) alloys were exposed to severe corrosion conditions and the corrosion rates vvere followed by various technigues (eiectrochemistry, Tafei extrapolation method). The vveight loss vvas measured over a period of 10 weeks in a wet corrosion chamber. Corrosion products were anaiysed using X-ray diffraction and the microstructures were investigated by optical microscopy and on SEM - EDS. In aggresive media, such as diluted NaCI or H2SO4, the differences between the corrosion rates vvere small. The lowest potential difference between the anodic phase (corrosion products) and the matrix, acting as cathode, vvas observed in Nd-Dy-Fe-B-Zr alloys. Corrosion rates in fresh vvater vvere 0,30 mm/year for Nd-Dy-Fe-B alloy and 0,02 mm/year for Nd-Dy-Fe-B-Zr alloy. The same trend vvas shovvn on samples exposed to conditions of simulated condensed atmospheric humidity. The highest cumulative vveight loss occurred vvith pure Nd-Dy-Fe-B alloys and the lowest vvith the alloy improved by Zr02 addition. The corrosion rates for three different alloys were 0.089 mm/year for Nd-Dy-Fe-B alloy, 0,072 mm/year for Nd-Dy-Fe-B-Hf alloy and 0,063 mm/year for Nd-Dy-Fe-B-Zr alloy.
Key words: corrosion, Nd-Fe-B alloys, permanent magnets
Osnovne zlitine Nd-Dy-Fe-B-X (X = Zr, Hf) smo izpostavili agresivnim korozijskim pogojem in zasledovali korozijski proces z različnimi metodami (elektrokemija, Taflova ekstrapolacijska metoda). V vlažni komori smo merili izgubo teže v obdobju desetih tednov. Korozijske produkte smo analizirali z uporabo X -žarkovne difrakcije ter opazovanjem mikrostrukture z optično mikroskopijo in elektronskim mikroskopom opremljenim z EDS. I/ agresivnih medijih kot sta NaCI in H2SO4 so bile razlike y korozijski hitrosti med različnimi zlitinami majhne. Najmanjšo razliko potenciala med anodno fazo (korozijski produkt) in matrico, ki deluje kot katoda, smo opazili pri zlitini Nd-Dy-Fe-B-Zr. Korozijska hitrost v vodi je bila 0,30 mm/leto pri zlitinah Nd-Dy-Fe-B in 0,02 mm/leto pri zlitinah Nd-Dy-Fe-B-Zr. Enako tendenco smo opazili pri eksperimentih, pri katerih so bile zlitine izpostavljene pogojem, ki so simulirali nasičeno zračno vlago. Najvišja kumulativna izguba teže je bila dosežena s čistimi Nd-Dy-Fe-B zlitinami in najnižja z Nd-Dy-Fe-B-Zr zlitinami. Korozijske hitrosti za različne zlitine so bile 0,089 mm/leto za zlitino brez dodatkov, 0,072 mm/leto za zlitino z dodatkom Hf02 in 0,063 mm/leto za zlitino z dodatkom cirkon oksida.
Ključne besede: korozija, Nd-Fe-B zlitine, trajni magneti
1 Introduction
Among the rare earth based permanent magnets, Nd-Fe-B magnets have assumed an important position due to their outstanding magnetic properties1,2 and their use is stili on grovving in different fields of application3. Hovvever, cotTosion has been a problem vvith Nd-Fe-B magnets, because phases rich in rare earth elements are easily oxidised in air, especially in humid air4-5. Since corrosion can deteriorate seriously the magnetic properties and on the other hand, can also be detrimental to magnetic cir-cuits, much effort has been made to improve the corrosion resistance of Nd-Fe-B magnets. Even coating and plating are not the perfect solution to this problem, because they can be imperfect and allovv the penetration of reacting species such as moisture to the magnet surface6. Searching for a better resistance of the material itself, various referred possibilities have been studied.
Narasimhan et al.7 reported that raising the oxygen content to betvveen 0,6 to 3,5% significantly improved the corrosion resistance; Kim and Jacobson reported that the addition of Al, Dy or Dyz03 improved the corrosion resistance in humid air4, vvhile Tenaud, Vial, Sagavva8 and Hirosavva et al.9 used V and Mo to improve the basic
1 Dr. Spomenka KOBE BESENIČAR
Inslitul Jožef Štefan. Jamova 39 1001 Ljubljana. Slovenija
corrosion resistance of Nd-Fe-B magnets. Kobe et al. reported on the beneftcial influence of ZrC>2 addition not only to the inereased coercivity, but also to the corrosion resistance of the Nd-Dy-Fe-B magnets10. Previously Nakamura11 attained better corrosion resistance of the Nd-rich phase by the substitution of Fe vvith Co and Zr, and Sagavva et al.12 improved the corrosion resistance by addition of Co and Al. Kim et al.6 influenced the corrosion resistance by varying the amount of O, C and N in the basic composition of Nd-Fe-B magnets.
On the basis of the promising results in our previous work10, vve continued our studies on the influence of Zr02 and Hf02 additions on improving the corrosion resistance of the basic Nd-Dy-Fe-B alloy vvith the composition Ndi5DyiFe76B8. The corrosion resistance vvas fol-lovved over experimental periods during vvhich the samples vvere exposed to various severe corrosion conditions.
2 Experimental
The basic alloys used fot the corrosion experiments vvere prepared by are melting the alloys NdFe, DyFe, FeB and Fe povvder in a pure Ar atmosphere. In order to prevent the oxidation Ti sponge vvas used as a getter for oxygen. Three different batehes vvere prepared: A - samples vvithout other additives, B - 1 wt.% hafnia vvas
added before are melting, C - I wt.% of zirconia was added prior to are melting. Samples were remelted three times in order to attain a better homogeneity. Buttons of melted alloys were sliced and polished to dises, dimen-sionally appropriate for the corrosion tests.
The investigations were focused on general corrosion resistance, based on electrochemical determinations of the possible passivity of electrode surfaces, or aetive corrosion. Moreover, service conditions were simulated by exposing the test specimens in a wet corrosion facility (DIN 5017), with the aim of establishing the effect of chemical composition and microstructure on the corrosion rate and the form of corrosion.
The potentiodynamic anodic polarisation measurements were performed using an EG and G-PAR poten-tiostat and "Softcorr 352" softvvare. Experiments were carried out in fresh water and in various aqueous test-so-lutions containing low concentrations of aggressive ions sueh as Cl" and SO42*. Sueh media could only represent approximative atmospheric conditions in the industrial environment. Electrochemical determination of corrosion rates were performed by the Tafel plot technique.
After exposing the samples to various corrosion conditions they were characterised by optical and electron microscopy (SEM/EPMA JEOL, JXA 840 A). Phases in corrosion products were identified using EDS and WDS analysis facilities and an X-ray diffractometry (Philips 1710).
3 Results and discussion
3.1 Effect of the HfC>2 and ZrC>2 additives on the corrosion rate of Nd-Dy-Fe-B alloys
The example of the anodic polarisations curves pre-sented in Figure 1 indicates that ali of the three materials cannot achieve passivity. The overall shape of the curves indicates that the materials undergo aetive corrosion. It is evident that the potentiodynamic scans did not reveal any significant feature, sueh as a passive region where passi-vation is spontaneous, the pitting potentional or the eriti-cal anodic current. The conclusion from the anodic po-
tentiodynamic scans of the materials carried out in different solutions was that no significant passivation oc-curred.
Due to sueh polarisation behaviour of the materials, corrosion rate measurements were performed by the Tafel plot technique. The corrosion rates of the materials tested vvhen exposed in various media are presented in Table 1 and graphically in Figure 2.
From these results it can be concluded that chloride ions drastically promote corrosion. As their concentration inereases, so does the rate of corrosion. The corrosion process is also particularly dramatic in acid solutions containing SO42" ions, which represent very aggressive industrial atmosphere. The corrosion rates of ali materials in fresh water are relatively favourable. In addition, the results of this investigation showed that a defined trend which favours a NDFB-Zr02 material ex-ists (Figure 3, Table 1).
The same trend among the materials was observed by exposure in a wet corrosion facility, but a substantial im-provement of the corrosion properties by addition of ZrOa was not achieved. Results are presented in Table 2.
6 .386 ...........■ n-rrm].......... ................
- NFB
e .iee----- NFB-Hf
NFB-Zt
S -e.iae -
tn
3 -8.388 u
-6.986 ..........^ ........1 ................—.........—.......
-6.886 -7.888 -6 888 -5.888	BB8 3 B8B -Z.886
I/area (18**
Figure 1: Potentiodynamie polarisation curves for three types of alloys tested in fresh water, 20°C
Table 1: Corrosion rates of alloys in different media at 20°C
Material	Media	Corrosion rate (mm/year)
NFB	fresh vvater	0,300
NFB-HfO:	fresh water	0,530
NFB-ZrCb	fresh water	0,022
NFB	0,09 M NaCl	2,120
NFB-Hf02	0,09 M NaCl	2,710
NFB-Zr02	0,09 M NaCl	2,650
NFB	0,17 M NaCl	2,650
NFB-HfCh	0,17 M NaCl	3,260
NFB-Zr02	0,17 M NaCl	3,150
NFB	0,5 M H2S04	303,0
NFB-HfOj	0,5 M H2SO4	274,0
NFB-ZrOj	0,5 M H2SO4	237,8
546		
E E
Figure 2: Corrosion rates of the materials exposed in various media presented graphically
Figure 5: Microstructures (cross sections) of sample with Hf02 (B) and sample vvith ZrC>2 (C) (385 x)
proceeds in samples A. In samples B and C the corrosion products are located mainly on the surface, especially in samples C, vvhere no deep corrosion in the bulk material was observed. The reason for such local corrosion is sup-posed to be the presence of particular phases.
More detailed analyses of the phases present vvere obtained by electron microscopy. Figure 6 shows the combined BS/SE image of an SEM micrograph of sample A and spectra of phases Pi and P2. The phases present in the corrosion products of sample A vvere found to be combined Nd, Dy and Fe oxides. The ratio betvveen Nd and Fe oxides differs in the phases Pi and P2. The results of standardless quantitative analyses (ZAF correc-tion program) are presented in Table 3.
Table 3: The results of standardless quantitative analyses of the oxide phases
	Nd203 (wt.%)	Dy203 (wt.%)	FeO (wt.%)
Phase P,	43,35	30,18	24,47
Phase P2	07,69	-	92,31
Phase P n	37,55	24,73	37,72
Phase P|2	11,99	-	88,01
In samples B a Hf-Fe rich phase vvas detected. The combined BS/SE image of the SEM micrograph of sample B and the corresponding spectrum of phase P6 are shovvn in Figure 7. Other phases present are the matrix phase P5 (REjFenB) and RE -rich phase P7.
In samples C a Zr-Fe -rich phase vvas found, mostly on the phase boundaries betvveen the hard magnetic RE2Fei4B phase (P5) and the RE -rich phase (P7). A combined BS/SE image of the SEM micrograph of sample C and the corresponding spectrum of Zr-Fe -rich phase P9 are shovvn in Figure 8. The SEM micrograph of the same sample shovving different phases in the cor-roded area and the corresponding spectra of these phases are presented in Figure 9. In samples C, the barrier based on the Zr-Fe -rich phase, vvhich exists betvveen the
0123466789 10 Time (weeks)
Figure 3: Cumulative mass-loss of different alloys during 10 vveeks of exposure in a wet corrosion chamber
Table 2:Corrosion rates of alloys exposed in a wet corrosion cabinet
Material	Environment	Corrosion rate
_(mm/year)
NFB Wet corrosion chamber 0,089 NFB-HfO: Wet corrosion chamber 0,072 NFB-ZrO;_Wet corrosion chamber_0,063
3.2 Microstructural study
Cross section of the samples A, B, C vvere ground and polished vvith diamond paste. The polished surfaces vvere examined by optical microscopy and electron mi-croscopy (SEI and BSEI). The phases present vvere ana-lysed using EDS standardless quantitative analyses.
Figure 4 shovvs a comparision of the microstructures (cross sections) of sample vvithout any addition (A) and sample vvith 1 wt.% of Hf02 addition (B). Figure 5 shovvs the cross sections of the polished surfaces of samples vvith Hf02 (B) and Zr02 (C) addition. There is an obvious difference in the level of corrosion attack betvveen the three samples. The most aggressive corrosion
Figure 4: Microstructures (cross sections) of sample without any addition (A) and sample vvith 1 wt.% of Hf02 (B) (385 x)
Phase P9 44,22 54,02 05,76 Phase P|»_37,28_54,35_08,38
4 Conclusion
The results of the corrosion experiments and analyses of RE-Fe-B-X aIloys, as well as analyses ot the corrosion products and microstructural observation and analyses show, that zirconia addition gives the most promising re-
Figure 7: Coinbined BS/SE image of SEM micrograph of sample B and corresponding spectra of Hf-Fe -rich phase Ps
p ODO	VFS ■ 4036 10 240
Figure 8: Back scattered image of SEM micrograph of sample C and the corresponding spectrum of Zr-Fe -rich phase Pio
corrosion products (in the RE -rich phase) and the hard magnetic matrix phase, prevents the propagation of corrosion. Phase Pio shovvn on SEM micrograph (Figure 9) illustrates this tentative explanation. The results of stan-dardless analyses of the Zr-Fe -rich phases found are presented in Table 4.
Table 4: The results of standardless quantitative analyses of Zr-Fe -rich phases
Figure 6: Combined BS/SE image of SEM micrograph of sample A and spectra of phases Pi and P2
sults in corrosion protection of the basic material. Corrosion rates in fresh water were 0,30 mm/year for Nd-Dy-Fe-B alloy and 0,02 mm/year for Nd-Dy-Fe-B-Zr alloy. The same trend vvas shovvn vvhen the samples vvere ex-posed to conditions vvhere condensed atmospheric hu-midity vvas simulated. The highest cumulative vveight loss occured vvith pure Nd-Dy-Fe-B alloys and the lovv-est vvith the aIloy improved by ZrO: addition.
A tentative explanation for the difference is that the change in microstructure is obviously responsible for im-proving the corrosion resistance of Nd-Dy-Fe-B-Zr a!loy. The reason for local corrosion is the presence of particu-lar phases, (Fe-Hf, Fe-Zr) acting as an anode, vvith con-siderable potential difference betvveen these and the ma-trix. A tentative explanation for the formation of Fe-Hf and Fe-Zr -rich phases is that in the samples vvith HfC>2 and ZrOa addition, during the are melting process most probably Nd from Nd -rich phase reduces both oxides and Hf or Zr -rich phases are formed. They act as the barrier betvveen the corrosion products (in the RE -rich phase) and the hard magnetic matrix phase and to some extent prevent the propagation of corrosion.
The improvement of the corrosion resistance of basic material itself can contribute significantly to the stability of coated magnetic material.
j
O 0CB	VF5 ■ 1024 10 24«
	p„
	
	] 1
(j .	
k_________________jM.	1 LA::.-
0 eea	v^s ■ zo*9 10 2*0
Figure 9: SEM micrograph of sample C showing various phases in the corroded area and the corresponding spectra of phases Pjo, Pu and P12
5 References
1	M. Sagawa, S. Fujimura, N. Togawa, H. Yamamoto, Y. Matsuura, J. of Appl. Phys„ 55, 1984, 2083
2	J. Croat, J. Herbst, J. Lee, F. Pinkerton, J. of Appl. Phys., 55, 1984, 2078
3	M. Hersch, Permanent Magnet Demand Fuels Continued Market Grovvth, PCIM, July 1994, p 26
4	A. Kim, J. Jacobson, IEEE Trans, on Mag., 23, 1987, 5,2509
5	A. Kim, Journ. of Material Eng., 11, 1989, 1, 95
6	A. S. Kim, F. E. Camp, S. Constantinides, Corrosion of Electronic and Magnetic Materials, ASTM STP 1148, Ed. P. J. Peterson, Am. Soc. for Testing and Materials, Philadelphia, 1992, p 68
7	K. Narasimhan, C. VVillman, E. Dulis, US Patent, No. 4588439, 1986
8P Tenaud, F. Vi al, M. Sagawa, IEEE Trans, on Magn., 26, 1990, 5,
1730
9	S. Hirosawa, S. Tomizawa, S. Mino and A. Hamamura, IEEE Trans, on Mag., 26, 1990, 5, 1960
10	S. Kobe Beseničar, J. Holc, G. Dražič, B. Saje, IEEE Trans, on Mag., 30, 1994, 2, 693
" H. Nakamura. A. Fukuro, T. Yoneyama, Proc. of the IOth Inter. Work-shop on Rare Earth Magnets and Their Application, Kyoto, Japan, 1989, p 315
12 M. Sagawa, S. Fujimura, H. Yamamoto, S. Hirosavva, Japanese Patent, No. 6338555, 1988