Mössbauer effect study of b.c.c. structure alloys, FeAl and FeTi

Mössbauer effect study of b.c.c. structure alloys, FeAl and FeTi

M&SBAUER EFFECT STUDY G. OF B.C.C. K. WERTHEIMt STRUCTURE and ALLOYS, FeAl AND FeTi* J. H. WERNICKt Combined X-ray and Mdiissbeuereffect st...

619KB Sizes 0 Downloads 29 Views

M&SBAUER

EFFECT

STUDY G.

OF B.C.C.

K. WERTHEIMt

STRUCTURE

and

ALLOYS,

FeAl AND

FeTi*

J. H. WERNICKt

Combined X-ray and Mdiissbeuereffect studies have confirmed that annealed FeAl assumes the CsCl structure, and have shown that no magnetic moment is associated with the iron atoms. Disorder produced by slip inducesmagnetic moments on iron atoms with iron near-neighbors. Experiments withnon-stoichiometric FeAl indicate that iron atoms with eight iron near neighbors have a moment of 1 pa. In Fei.rA1o.a these order ferromagnetically at low temperature. The present findings explain the bulk magnetic properties observed in plastically deformed and non-stoichiometric iron-rich FeAl. In FeTi no tendency to assume a CsCl superlattice was found. Low temperature annealing (600°C) produces a non-random occupancy of the b.c.c. lattice corresponding to Fe atom clusters. ETUDE

DES STRUCTURES C.C. D’ALLIAGES FeAl ET FeTi AU MOYEN DE L’EFFET MGSSBAUER

Des etudes realisees au moyen des rayons X et de l’effet Mossbauer ont confirm6 que le FeAl recuit eat de structure de type CsCl et one montre qu’aucun moment magnetique n’est associe aux atomes de fer. Le desordre produit par glissement induit un moment magnetiquedans lea atomes de fer dont lea plus proches voisins sont dees atomes de fer. Des experiences faites aveo un compose FeAl non stoechiometrique indiquent que lea atomes de fer dont huit des plus proches voisins sont des atomes de fer ont un oment de 1 ns. Dana Fe,,,Alo,a ceux-ci s’ordonnent it basse temperature d’une f-on ferromagnettique. Les resultats de la presente etude expliquent lea proprietes magnetiques en volume observees dans lea composes non stoechiometriques FeAl riches en fer et deform& plastiquement. Le FeTi n’a pour sa part aucune tendance a former un super reseau de type CsCl. Un recuit a basse temp&ature--600”Cproduit l’occupation ordonnee du reseau C.C. correspondant & desamas d’atome de fer. UNTERSUCHUNG

DER

k.r.z. LEGIERUNGEN FeAl UND MGBBAUEREFFEKTES

FeTi MIT HILFE

DES

Kombinierte Untersuchungen durch Riintgenstrahlen und MCiBbauereffekt haben bestatigt, dalj angelassenes FeAl CsCl-Struktur annimmt, und gezeigt, da13mit den Eisenatomen kein magnetisches Moment verbunden ist. Durch Abgleitung verursachte Umordnung induziert magnetische Momente von Eisenatomen mit n&&&en Eisennachbarn. Experimente mit nichtstochiometrischem FeAl zeigen, dalj Eisenatome mit acht nachsten Eisennachbarn ein magnetisohes Moment von 1 ps haben. In Fe,,,Al,,, ordnen sich diese bei tiefen Temperaturen ferromagnetisch. Diese Ergebnisse erklaren die meisten der in plastisch verformten und nichtstiichiometrischen eisenreichen FeAl-Legierungen gefundenen magnetischen Eigenschaften. In FeTi zeigt sich keine Tendenz zu einer CsCl-Uberstruktur. Tieftemperaturerholung (600°C) fiihrt zu einer nichtstatistisohen Besetzung des k.r.z. Gitters, was Agglomeraten von Eisenatomen entspricht.

In the Fe-Al system single phase b.c.c. solid solutions can exist to approximately 54 at. ‘A Al in Fe at 1100°C. At 25 at.% Al, a disordered b.c.c. solid solution exists at elevated temperatures and on cooling orders to a solid solution which is CsCl-like at 800°C. This predominantly ordered phase transforms to the Fe&l ordered structure at approximately 500%. The order-disorder temperature for the CsCl-type structure moves to higher temperatures as the Al content increases. The ideal composition, FeAl, appears to be ordered to the start of the melting range. U-4) Fe&l can be maintained in either the ordered or disordered states by heat treatment. Each state has distinct magnetic properties, which have been studied in detail both by magnetic measurements(3s5) and more recently by Mijssbauer effect.(6-s) FeAl, although ordered up to the melting range, can be disordered to some extent by mechanical deformation, e.g. crushing or filing. The ordered

material has a small susceptibility indicating that the iron atoms do not have magnetic moments; the crushed material is weakly ferromagnetic. In the Fe-Ti system,(10-14) the maximum solid solubility of Ti in b.c.c. Fe is ~10 at.% and occurs at 1290°C. Beyond that point the hexagonal Laves phase TiFe, forms. For compositions in the range from 33.3 % Ti to 49.8 at. % Ti, TiFe, and b.c.c. FeTi coexist. Ordering of the intermetallic compound FeTi to the CsCl structure type has not been conclusively demonstrated by X-ray techniques because of the similarity of the scattering factors of Ti and Fe. The present work is concerned with the magnetic properties of ordered and disordered FeA.l and with the question of order in TiFe.

* Received April 4, 1966. t Bell Telephone Laboratories Incorporated, Murray Hill, New Jersey. ACTA

METALLURGICA,

VOL.

15, FEBRUARY

1967

297

EXPERIMENTAL

The Fe-Al alloys were prepared by induction melting the required amounts of 99.99% Fe and 99.999% Al in dense A.&O, crucibles in an argon atmosphere. The ingots (~10 g) were sealed in evacuated quartz ampules and annealed for 4 hr at lOOO”C, and then for 1 week at 800°C. Samples for

ACTA

298

METALLURGICA,

VOL.

15,

1967

99 ANNEALED

98 97

r:

94

4 e

92

if .l

90

& 100 d: u z F: 3 CJ

98

99

94

88

80

: -0.012

I -0.09

1 -0.04 DOPPLER

I 0 VELOCITY

I 0.04

1 0.08

-.

I 0.12

0.8

I 08

(CM/SEC)

FIQ. 1. Mijssbauer absorption spectra at 298’K of (a) aunealed FeAl (b) crushed FeAl (c) annealed Fel.lAl0.g. The data iu 1(a) are fitted with a single Lorentzian of width 0.040 cmfsec located at 0.0 104 cm/set; those in 1(b) could not be fitted satisfactorily by less than 4 Lorentzian components.

the Miissbauer measurements were prepared by crushing and reannealing for 48 hr at 600%. Both as-crushed and annealed specimens were examined. FeTi was prepared by inert electrode arc melting. The purity of the Ti was 99.99%. All of the above alloys were examined metallographically and by X-ray powder photographs. The latter were obtained with CrK, radiation and Straumanis-type Norelco cameras of 114.6 mm dia. The Miissbauer effect spectrometer has been previoudy described.05) The source of recoil-free gamma rays was Co5’ diffused into palladium. Positive Doppler velocity corresponds to higher energy in the absorber.

Fxa. 2. Mijssbauer absorption spectra at 4.2”K of (a) annealed FeAl (b) crushed FeAl (c) annealed Fel.lAlo.0. The data in l(a) are fitted with & single Lorenteian of width 0.043 cm/see at 0.024 cnjsec; those in 2(b) by two Loreutzian of widths 0.056 and 0.386 cmjsec located respectiveiy at 0.024 and 0.006 cm/see. The broad line contains 30% of the total area. The data in 3(c) are not least-squares fitted. RESULTS

AND

DISCUSSIONS

The simplest Mijssbauer spectrum w&s obtained with annealed FeAl. It consists of a single line, Figs. 1(a) and 2(a), whose shape is approximately Lorentzian even though the width is considerably in excess of the natural width. The performance of the equipment with other absorbers such as metallic iron indicates that the excess width originates in the sample itself. The results to be described below will make it clear that the distorted line shape and increased width are due to imperfect order, i.e. Fe atoms in A.l sites, or else due to Fe atoms near antiphase domain boundaries. The width increases only slightly at 4.2”K. The absence of resolved splitting is consistent with

WERTHEIM

AND

WERNICK:

MOSSBAUER

the (ordered) CsCl structure and shows that there is no magnetic transition in this compound. The crushed material in which the ordered superlattice is partially destroyed has a distinctly different spectrum, Figs. l(b) and 2(b). At 298°K the absorption line is asymmetrically broadened, and the centroid is shifted slightly toward lower energy relative to that of the annealed material. At low temperature the components which contribute to the broadening at 298’K spread out into a very broad background absorption without any defined features, Fig. 2(b). The width of this absorption is such that it can only originate from magnetic hyperfine interaction. (X-ray pattern of crushed material shows superlattice lines, in addition to broad lines, indicat(ive of large elastic distortion.) In order to estimate the fraction of iron atoms involved in this broad absorption the spectrum was fitted with two Lorentzian lines by the method of least squares. The width of one line is close to that of the annealed material; the other is 30 times greater. No particular significance is attached to the use of a Lorentzian line shape for the broad line; it is merely an arbitrary way of representing a broad spectrum resulting from superparamagnetism or inhomogeneous magnetic h.f.s. interactions. The area under the broad curve contains 0.30 of the total area. Interpretation of the broad component as due to superparamagnetic clusters of iron atoms requires a moment of about 1 ,u~ per atom. Interpretation in terms of inhomogeneous magnetic h.f.s. requires that the average iron atom in the disordered region have a moment of ~0.7 lu,. These results are generally consistent with the well-known enhanced susceptibility produced by cold [email protected]) In order to relate the magnetic properties to the local environment of the iron atoms we will examine the effects of slip in an ordered b.c.c. phase. Two effects dominate: (1) the passage of dislocations will destroy the ordered occupancy of the b.c.c. lattice in the vicinity of the slipped region, producing for example antiphase domain [email protected]) or stacking slip the slip plane normal faults .cl’) For {llO)(lll) contains excess like-atom [email protected]) (2) The regions associated with dislocation lines and tangled dislocations have large elastic distortions. In the following analysis we will be concerned with the former which is expected to have a more pronounced effect on the magnetic properties of the atoms in the alloy. The most important slip system in b.c.c. material is {110}(111). However, slip has been observed to occur on (211) and (321) planes in some b.c.c. metals.

STUDY

B.C.C.

ALLOYS

299

Considering the former it is readily shown that a unit displacement in a (111) direction replaces iron atoms with aluminum atoms (and conversely) in the slipped region. From the point of view of iron atoms adjacent to the slip plane the chief effect is the replacement of two near-neighbor Al atoms with Fe atoms and the replacement of two next-neighbor Fe atoms with Al atoms. More complex disorder will be introduced at regions where slip planes intersect after the passage of dislocations but the essential feature of disorder is the production of iron atoms with iron atoms in near-neighbor sites. The immediate hypothesis is that the weak ferromagnetism and the broadening observed by Mijssbauer effect are due to these iron atoms. The shift of the broad component is in accord with this model since it is away from FeAl toward metallic iron. (Both have the same fundamental b.c.c. structure.) As a check on this hypothesis it is desirable to prepare a material in which iron-iron neighbors are introduced into FeAl in a controlled fashion. This circumstance can be achieved by preparing an alloy with excess iron, i.e. Fel+,Al_,. For quite sizable deviations from stoichiometry X-ray evidence shows that the material still attains an ordered superlattice. The resulting structure is conveniently described as a cubic iron lattice with an interpenetrating cubic aluminum lattice, in which the iron lattice is entirely occupied by iron atoms, but the aluminum lattice is randomly occupied by aluminum and iron atoms in the ratio 1 - x to x. Restricting ones attention to the immediate neighbors there are now a number of distinct “neighbor configurations”, which are here specified by listing the number of iron atoms in near-, and next-near-neighbor positions. All iron atoms on the aluminum lattice have 8 iron near-neighbors, and dominantly Al next-near-neighbors. Iron atoms on the iron lattice may have 0, 1, 2, . . . , 8 iron nearneighbors, and all Fe next-near-neighbors. For small deviations from stoichiometry the number of iron atom neighbors will be small, i.e. 0, 1 or 2. As a result such an ordered alloy has iron atoms with all iron near neighbors and iron atoms with dominantly aluminum near neighbors and almost none with equal numbers of iron and aluminum neighbors. (See the Appendix for more detail.) It is not surprising that the magnetic properties of these iron atoms are distinct. At 4.2OK the Miissbauer spectrum of Fe,,lAl,,, shows the following components [Fig. 2(c)]: (1) iron with a hyperfine splitting of 150 kOe, (2) a single line at the position of the line in stoichiometric FeAl, and (3) satellites at lower energy. The hyperfine splitting has been identified with the iron atoms with eight iron neighbors because

ACTA

300

METALLURGICA,

VOL.

15,

1967

TABLE 1. Magnetic properties and Isomer Shift of iron atoms b.c.c. metals and alloys Magnetic moment (pa)

Fe neighbors near, next-near

Material

I.S. * Heri (T = 0) (koe)

(==2f;W

Ref.

Fe Fe,Al Ordered D site

696

2.18

340

-0.0182

890

2.14t

-0.019t

A site

4,6

FeAl

0,6

336 f 10 318 246 f 10 242 0

Fe,+&-,

690

1.46.t 0

-0.001$ 0.0104 0.007: -0.017

150

-1

Ref. Ref. Ref. Ref.

6 8 6 8

Ref. 6

* The Isomer Shift is expressed relative to Pd(Co6’) at 298°K. t R. NATHANS, M. I. PIQOTTand C. G. SHULL, J. Whys. Chem. Solids 6, 38 (1958). z 0.029 cm/set subtracted from published values to bring them to Pd reference.

the fraction

of the area under

corresponds

closely

aluminum

sites, -9%.

considerable

and iron atoms (see Appendix). all Al

near

produce Those

spectrum

presumably

sites are occupied neighbors

and

the absorption

spectrum

example

of iron atoms in

The h.f.s.

broadening,

next-neighbor

the h.f.s.

to the fraction

shows

because the six

that of metallic

shift is mainly

deter-

atoms and is similar to

iron if the near neighbors

are iron

atoms.

The iron atoms with next

(2) the isomer

mined by the near-neighbor

by both aluminum

all Fe

the two cases with 8 near, and 0 next-near

Fe neighbors),

105

neighbors

line identical to that of FeAl.

with one or more iron atoms in the Al near

neighbor shell produce the low energy satellites which

FeTi CRUSHED 4.2-K

may actually be the result of small hyperfine structure splittings. The

ferromagnetic

behavior

of this

nonstoichio-

metric alloy requires further discussion.

It is apparent

from the data that only a small fraction atoms

of the iron

(those in Al sites) have appreciable

moments.

The other iron atoms

magnetic

have no magnetic

moment, or one too small to produce resolved splitting, i.e. less than 0.2 pB.

The interactions

magnetic

take place

duction

atoms

may

between

via polarized

electrons, but the present experiments

the

‘\ i;

P

condo not

define the mechanism. A comparison FeAl

makes

it possible

with the annealed

former

with iron atoms having

to identify

These results also support anomalous,

temperature

the heat capacity terms

in Fig.

independent

1

The

is in accord

of the FeAl

imperfect

to in line

shape in annealed

FeAl

clusions concerning

residual disorder in annealed FeAl

with the con-

in Table

I

80 -08

found in [email protected]) The summary

I

in the

contribution

stoichiometric

clusters.(lg)

K

iron near-neighbors.

the interpretation

of nearly

of magnetic

Fe&l,,,

the broadening

4.2’

I

of the data for the crushed stoichio-

metric

FeTL ANNEALED

1 compares

the magnetic

moments, hyperfine field and isomer shifts of Fe atoms in three b.c.c. metals and alloys, Fe, FeaAl and FeAl. A number of additional facts are apparent. (1) The near and next-near neighbor environment do not suffice to specify the hyperfine field (see for

I -0.6

I

I

-0.4 -0.2 DOPPLER

“ELOkY

I

I

I 0.4

I 0.6

I 0.

::,SEC)

Fm. 3. MGssbauer absorption spectra at 4.2”K of (a) crushed FeTi (b) annealed FeTi. The data in 3(a) are fitted by two Lorentzians with widths 0.038 and 0.271 cm/set located respectively at -0.0198 and -0.0088 cm/set. The data in 3(b) contain a narrow component coincident with the narrow component of 3(a) and a “pedestal” corresponding to a hype&e interaction of 100 kOe.

WERTHEIM

ALNDWERNICK:

MOSSBAUER

FeTi

The absorption spectrum of crushed FeTi at 4.2”K, Fig. 3(a), closely resembles that of crushed FeAl. The only significant difference is in the Isomer Shift, Table 2. A least-squares fit to the low temperature

STUDY

M&&e1

-FeAl, annealed FeAl, crushed

+0.0104 Not analyzed

FeTi crushed

-0.0317

Tq.&‘eo.ol

301

ALLOYS

assume the ordered CsCl structure on annealing. On the contrary, there does appear to be a tendency for the clustering of iron and titanium atoms during low temperature annealing (600°C) but without destruction of the b.c.c. lattice.

TABLE 2. Comparison of Isomer shifts of iron in FeAl and

FeTi. For crushed FeAl and FeTi the broad line is shifted towards the I.S. of metallic iron

B.C.C.

ACKNOWLEDGMENTS are indebted to D. Dorsi, D. N. E. Buchanan and E. Berry for assistance with these experiments. The authors

4.2”K

Remarks

APPENDIX

+0.0240 +0.0239 $0.006

Single line Narrow line Broad lie, 30% of area Narrow line Broad line, 30% of area

Fel+,$ll_Z, the fraction atoms are on the Al lattice. These atoms have 8 iron near neighbors and a probability P,(n) of having n iron next-near neighbors:

-0.0204

-0.01

-0.021

In

the

Quadrupole split line

* The Isomer Shift is expressed relative to Pd(CoS’) at 298°K.

again indicates that ~30% of atoms have magnetic moments and exist in superparamagnetic clusters or else that inhomogeneous h.f.s. with average moments of No.5 PB are responsible for the broadening. These results are generally compatible with the magnetic measurements on crushed material reported by NevittJ2n and with the interpretation of both heat capacity and magnetization on the basis of superparamagnetic clusters. (22~ss)The present analysis indicates smaller magnetic moments than the 2.2 ,u~ postulated in Ref. 22, and also shows that a larger fraction of the volume is included in the magnetic clusters. By analogy with the behavior of FeAl, the unusual bulk magnetic properties of FeTi are then attributed to iron atoms which have appreciable magnetic moments because of the presence of iron near neighbors, i.e. clusters. (The suggestion that FeTi may be an antiferromagnet is ruled out by the Mijssbauer spectra.) Annealing at 600°C for 48 hr does not produce the single narrow Mossbauer absorption line characteristic of an ordered phase. Instead, the background changes into a broadened h.f.s. pattern corresponding to an effective field of ~100 kOe, Fig. 3(b). The magnitude of this field corresponds closely to that of the site with greater magnetic moment in Fe,Ti, but the pattern shows none of the detailed structure expected for that compound. X-ray examination of the annealed data

ordered

alloy,

x/l + x of the iron

P&)

6! = (6 _ n)l %l x”(l - XY,

where O
X)*--n>

where OlnL8. The probabilities of the most common near, and next near neighbor arrangements for x = 0.1 are given in Table Al. TABLE Al Iron neighbors (near, next year) 076 196 2,6 396 436 870 891 8,2 8,3 -

Probability 0.391 0.347 0.135 0.030 0.004 0.048 0.032 0.009 0.001 0.997

REFERENCES

(The Ku, and

1. For a discussion of this zdloy system see M. HANSEN, Constitutiorrof Binary Alloys, pp. 90-95. McGraw-Hill (1968); R. P. ELLIOTT, Con&ution of Binary Alloys, pp. 36-37. First Supplement (1965). 2. A. J. BRADLEY and H. H. JAY, Proc. R. Sot. Al%, 210

The essential conclusion which can be drawn from the present work is that FeTi has no tendency to

3. A. TAYLOR and R. M. JONES, J. Phys. Chem. Solid8 6, 16 (1958). 4. F. LIHL and H. EBEL, Arch. EisenhQtt Wes. 52,483 (1961). 5. H. SATO and A. ARROTT,Phys. Rev. 114, 1427 (1959).

material gave only lines corresponding to the b.c.c. structure in agreement with the work of Refs. 13, 24 and 25 but not with that of Ref. 26. Km, reflections were clearly resolved.)

(1932).

302

ACTA

METALLURGICA,

VOL.

15, 1967

6. K. ONO, Y. ISHIKAWA and A. ITO, J. phys. Sot. Japan 17, 16. M. J. MARCINKOWSKIand N. BROWN, Phil. Mag. 6, 811

1747 (1962). 7. E. A. FRIEDMAN and W. J. NICHOLSON, J. appl. Phys. suppz. 84, 1048 (1963). 8. C. E. JOHNSON,M. S. RIDOUT and T. E. CRANSHAW,Proc. phys. Sot. Lond. 81,1079 (1963). 9. L. CSER et aZ., Phys. Lett. 19, 99 (1965). 10. For csdiscussion of the Fe-Ti alloys system see M. HANSEN, Constitution of Binary Alloys, pp. 723-727. McGraw-Hill, New York (1958). 11. H. WITTE and H. J. WALLBAUM, 2. MetaUk. 30,100 (1938). 12. H. W. WORNER, J. Inat. Metals 79, 173 (1951). 13. D. H. POLONIS and J. G. PARR, J. Metals 6, 1148 (1954). 14. I. I. KORNILEV and N. G. BORISKINA, Dokl. Akad. Nauk SSSR 108, 1083 (1956), translated in Proc. Acad. Sci. USSR, Chem. Section 108,323 (1956). 15. G. K. WERTHEIM,Miisabauer Effect Principles and Applications, Chapter II. Academic Press, New York (1964).

(1961). 17. R. J. WASILEWSKI, Acta Met. 18, 40 (1965). 18. G. Y. CHW, J. Mater. Sci. Engng, to be published. 19. C. H. CHENC, K. P. GUPTA, C. J. WEI and P. S. BECK, J. Phys. Chem. Solids 25, 759 (1964). 20. J. A. SEITCHIKandR. H. WALMSLEY, Phys. Rev. 137,Al43 (1965). M. V. NEVITT, J. appl. Phya. 31, 155 (1960). f K. SCHRBDERand C. H. CHENQ, J. appl. Phys. 81,2154 (1960); K. SCHR~DER,J. appl. Phys. 52, 880 (1961). 23. E. A. STARKE, JR., C. H. CHENQ and P. A. BECK, Phys. Rev. 126, 1746 (1962). 24. T. V. PHILIPS and P. A. BECK, Trans. Am. Inst. Min. metall. Engrs 209, 1269 (1957). 25. A. E. DWIQHT, Trans. Am. In&. Min. metall. Engrs 215, 283 t19591. 26. P. P;ETR;KOWSKY and F. G. YOUN~KIN, J. appl. Phys. 81, 1763 (1960).

B: