Materials Science and Engineering, A 145
( 1991 ) 127-132
127
Segregation effects as a possible mechanism for strengthening in metallic glasses S. V. Pan, Yu. V. Milman and A. A. Malyshenko krantsevich Institute/or l'robh'ms of Materials Science, 3 Krzhizhanovsky Str., 252142 Kiev (U.S.S.R.)
Abstract As a result of examination of the temperature dependences of the mechanical behaviour of iron-based metal-metalloid type metallic glasses (MG), it has been shown that, by analogy with crystalline materials, it is reasonable to introduce for MG the concepts of characteristic deformation temperature T*, thermal and athermal components of flow stress (hardness) and ductile-brittle transition being characterized with two transition temperatures: Tdbt, the lowest temperature of macroscopic plasticity emergence, and Tdb2 .~ 0.7 T×° ( T×° is the initial crystallization temperature) for all alloys tested. It has been found that alloying of MG with chromium and molybdenum leads to an increase of the thermal hardness component (whereas nickel promotes its lowering), the alloy strength and plasticity increase, the value of Tdb I is reduced and there is delocalization of the plastic flow. Study of the effects of segregation on MG fracture surfaces allowed one to elucidate the nature of short-range order local disturbances in the sites, where the localization of plastic flow is most intensive, and to suggest the possibility of considering these processes as the basis for a phenomenological model of strengthening of MG by alloying.
1. Introduction The disordered structure of metallic glasses (MG) based on transition metals (with b.c.c, structure in the crystalline state), as has been shown earlier [1], is characterized by the availability of a covalent component in the interatomic bonds. This causes a sharp temperature dependence of the flow stress (hardness) in the inhomogeneous deformation range (low temperatures and high deformation rates), and is connected with a ductile-brittle transition [2, 3]. It is known that, in spite of the high strength of the MG, its deformation strengthening either does not occur or has a negligible value due to the very intensive strain localization [1]. The strength level of the M G depends firstly on their chemical composition (percentage of metalloid atoms, presence of alloying elements) and real alloy structure (availability and type of defects). An attempt to present a generalized scheme of *Submitted to the Proceedingsof the Seventh International Conference on Rapidly Quenched Materials. Stockholm. August 13-17, 1990.
mechanical behaviour of iron-based M G over a wide temperature range and to find out what processes affect strengthening in M G of various compositions has been done in this work.
2. Experimental details All M G specimens (Table 1) were prepared using the melt spinning method in the shape of 17-30 /~m thick ribbons. The initial crystallization temperatures Tx° were determined by using a differential scanning calorimeter "Du Pont 1090" with a heating rate of 2 . 3 x 1 0 -2 K s -t in an argon atmosphere. Investigations of the mechanical characteristics (failing stress, of, proof (flow) stress o~ ~ o0.02 and residual strain 6) in the temperature range 4.2-850 K were carried out by tensile experiments at a strain rate of 3 x 10-4 s- 1 [3]. Microhardness value measurements were done at temperatures 77-850 K with the load of 1.8 N using the method described in [2]. Many authors [1, 4-6] have obtained indirect evidence (for example, preferential etching of shear bands outlets) that under inhomogeneous plastic flow changes of chemical and topological Elsevier Sequoia/Printedin The Netherlands
128 TABLE 1 Initial crystallization, ductile-brittle transition and characteristic deformation temperatures and mechanical properties of examined MG ( T= 293 K) Alloy composition
Tx° (K)
Tdbj (K)
T* (after HV data)
HV (GPa)
aS (GPa)
of (GPa)
d (%)
HV'/HV"(")
180 225 300 320 338 362
7.4 7.85 8.8 7.7 8.3 7.95
1.77 1.43 1.90 2.26 2.36 2.21
3.49 3.08 1.93 2.78 2.85 2.96
0.27 0.13 0.03 0.10 0.18 0.34
0.46 0.72 0.45 0.97 1.17 1.35
(K)
Fe40Ni38Mo4Bls Fe70Ni8Sil0B j 2 Fe~3Bi7 F%0CrsBl5 Fe75Cr10B15 Fe70CrlsBl5
661 729 664 674 703 674
<4.2 180 293 270 260 245
aThe value of HV t was determined at T= 77 K and HV a at T* < T< Ta~2.
short-range order (SRO) occur causing from their point of view the localization of plastic flow and promoting it along the slip bands, having been already formed. To find out the nature of these changes the effects of segregation on the fracture surface of MG samples strained in situ up to failure inside a JAMP-10S Auger microprobe under a vacuum of about 10 -8 Pa [7] have been studied. When fractured, the sample develops on its fracture surface a characteristic vein structure (Fig. ld). These veins are the sites of the most intensively localized plastic flow--the same kind of "micronecks" along which the M G completes its failure. Therefore, their chemical state carries information about the composition and structure of the material, of which the shear band, the socalled "fluid layer" [8], consists. Auger spectra were taken at minimized electron probe diameter ( - 80 nm) and an accelerating voltage of 10 kV at some points of the fracture surface, starting from the vein tip P1 (Fig. ld)[7]. 3. Results and discussion
As a result of the investigations carried out, the uniform character of the temperature dependence of the mechanical behaviour for all ironbased MG examined was established (Fig. la). In most cases four characteristic temperature intervals could usually be distinguished in each of these dependences. The proof stress a s and especially the microhardness HV (and, in such a way, the flow stress also) are almost temperature independent in the second interval, whereas, below a certain temperature T*, a sharp increase of HV and o s is observed. Such a kind of flow
stress (hardness) temperature dependence is usually observed for many crystalline materials having a covalent component in the interatomic bonds [9]. That is why it is reasonable to consider the value of o s (HV) as a sum of athermal asa (HV a) and thermal o~t (HV t) components by analogy with crystalline materials (Fig. la) H V = I-1Va+ HV t and to introduce the concepts of MG characteristic deformation temperature T* [9], as temperature, below which the influence of covalent component in interatomic bonds on the mechanical properties of the MG becomes significant and it results in a sharp rise of the flow stress. The data obtained (Table 1) reveal that the value of T* strongly depends on the composition of the MG and, as a rule, lies in the range of 0.3-0.5 lx ° (after microhardness measurements data). Thus, measuring the microhardness of the most traditional MG (except some recently developed Fe-Cr-B alloys having T* > 293 K) we actually define the HV a, which achieves a great value in MG and depends feebly on the alloying metals content, but depends firstly on the metalloid percentage (for instance, boron [10]). At the same time, the value of the thermal component I-IV t (connected with the angle of slope of the H V ( T ) relationship) depends on the metal and concentration of alloying elements: alloying of Fe-B alloys with group-VIA metals gives rise to H V t, while an increase of the nickel content lowers it (Table 1 ). Below T* in MG, as in crystalline materials, ductile-brittle transition (DBT) can be detected. It appears as a drop to zero of the macroscopic plasticity 6 (Fig. la) and fracture mode change
2 ~)
'1(3
t5
lO/~m
L -
5/~m
-
.
J
1-
0
0.2
0.4
0.6
0.8
1.0
T/T, Fig. 1. Schematic diagram (a) of mechanical behaviour of iron-based MG vs. testing temperature and characteristic SEM fractographs corresponding to various temperature ranges: (b) T< Td,,]; (c) T = T, lhj: (d) TabI < 7'< Tdb2:(e) 1~>:< T< T,".
(Fig. lb-d), accompanied by a sharp decrease of of till the level of o~ and lower at the temperature Tdb~, depending on the alloy composition (Table 1 ). For example, in the Fe40Ni38MoaBs alloy, DBT has not been detected up to liquid helium temperature (o', and o I on Fig. l a). It is interesting to note that, according to the formal definition of DBT, the point of transition from inhomogeneous to homogeneous flow mechanisms, which is about 0.7 Tx° for all alloys tested, can also be considered as the DBT temperature T~b2: a significant enhancement of 6 (Fig. la) and a change of the failure mode into viscous type (Fig. le) accompanied by a jump of of. Simultaneously with the influence on the value of Tdb1, alloying with different metals greatly effects the strength level of the MG. When the chromium percentage in the Fe-B alloy increases from 0 to 15 at.% we find an increase of more than half in of at room temperature (Table 1). Since the value of o~ under these conditions undergoes insignificant changes, growth of strength occurs due to the increase of d, the level of which under inhomogeneous deformation
depends on the quantity of slip bands being initiated before sample fracture, i.e. on the degree of plastic flow localization. The results of registration of acoustic emission signals under deformation and fracture of Fe-Cr-B MG show that the increase in their strength and plasticity is accompanied by a significant flow delocalization
II11. Thus, the process causing initiation of the new shear bands, instead of catastrophic failure of the specimen due to development of the exceeding critical size crack along the first of just forming shear bands, leads in fact to plasticity enhancement of the MG. And, as far as an increase of 6 is followed by a growth of o I (due to the growth of the difference o f - o~) and strain delocalization, at the same time, leads to an increase of the strength of the MG. What is the nature of these processes, which take place in the sites of strain localization? In what way may the atoms of alloying elements effect the degree of strain localization and, thus, on the strength of the MG? The study of segregation effects on the MG fracture surfaces has partly answered these questions,
130
In studying of binary ~e84]~16 alloy fracture surface, it has been revealed that the veins show a segregation of interstitial elements so that the concentration of boron, carbon and oxygen atoms in P1 (Fig. ld) is substantially higher than in the other areas of the surface (Fig. 2). In the immediate neighbourhood of the vein a depleted region (P2) is observed and, proportionately to the distance from the vein, the concentration of boron becomes identical to that of the stoichiometric alloy. The same segregation effects (concerning boron atoms) have been observed in all MG tested. A peculiar feature of the multicomponent FeToNi8Sil0Bt2 alloy is a competitive segregation of iron and nickel revealed by absence of nickel peaks in the Auger spectra taken at P1 (Fig. 3). This may be explained by the substitution of part of the nickel by iron atoms in the region of flow localization. The common B C
0
tendency of vein enrichment with atoms of alloying metals has been determined under testing of other alloys: Fe70CrlsB15 and Fea0Ni38Mo4Bz8 (Fig. 4), but these processes seem to be much more complicated. One of the possible explanations of the nature of the observed segregation effects may be supplied by the following model based on assumption of the correctness of the idea which postulated the increase of free volume within shear band under inhomogeneous plastic flow in MG [1, 4-6, 8, 12, 13 et al.], i.e. dilation, which leads to a local decline of viscosity and density.
B C
O
Fe
Ni
Fe
% ILl
20
Z "I3
300
600 E (eV}
900
Fig. 3. Auger spectra accumulated at points P1, P2 and P3 on the fracture surface of FeToNisSi,~B~2metallic glass.
(a)
2o'
2;o
'
4;o ' 6;o E(eV)
+oo
4,~. " / /
(o)
I l l l l l
0.12
"
~
~]
]
1 / 1 1 / / z
,~
it.
)
(J - C r
08
Fe84016 o-B
2~.~//~
/
FeT0CrsB15 o-B
/ /
///,~, / / / / ~
~-C
~
o
o-0
\
/01///
P3
o12l,~,..~///] P+2
J
_
•
(b]
f]j
Fe4oNi38Mo4BiB
o++ o + ,i,?,
/,VEI N
0" / P l / / 0 (b)
0 1
2
3 r (~umJ
'4
Fig. 2. Characteristic Auger spectra (a) and segregation profiles (b): relative Auger peak intensities (I~:IF~)/Ie~ of boron, carbon and oxygen vs. distance r from the vein tip P 1 on the fracture surface of Fe84Bj6 metallic glass.
1
2
3 r (~um)
4
Fig. 4. Segregation profiles: relative Auger peak intensities Ix/1Fe of boron and chromium for Fe70CrlsBi5 metallic glass (a) and boron, nickel and molybdenum for Fe40Ni3~Mo+B~8 metallic glass (b) vs. distance r from the vein tip PI on the fracture surfaces.
131 TABLE 2 Atomic radii of elements according to Goldschmidt ( 14] Elements
Fe
Ni
Cr
Mo
B
C
O
R~, (nm)
0.126
0.125
0.128
0.14(/
0.097
0.077
0.060
Tensile stress under these conditions can stimulate and provide energy advantages to transference (due to the availability of numerous sites "expanded" by dilation) of interstitial atoms with small Goldschmidt radius Ra (Table 2) according to the vacancy mechanism, so that substitution of atoms with bigger R,, values instead of the same size but smaller atoms to the regions of excess free volume formation. However, the size factor alone cannot explain all the segregation phenomena observed; that is why it seems to be reasonable to take into account also the peculiarities of the electronic structure of the alloys. On the basis of these data, the process of inhomogeneous plastic flow may be presented (without details) as follows: the initiation and formation of shear band, consisting of the so-called "fluid layer" [8], confined between two "solid" surfaces, involved a disturbance of topological SRO which is manifested in the formation of the sites of excess free volume, causing transference of interstitial and alloying metal atoms into these regions. As a result of the local redistribution of atoms, such changes of chemical SRO may occur which is revealed in partial compensation of the excess free volume formed and an increase of strong bonds within the shear band and across its boundaries; in other words, leading to its strengthening. When the stress level within the slip band reaches a critical value and fracture of "fluid layer" is initiated (for instance, by the Spaepen mechanism [8]) causing fracture surface formation, i.e. formation of the veins within which plastic flow localization reaches its maximum level. Thus the veins are the areas where after failure the structure corresponding to the greatest disruption of SRO, is "frozen". The observed segregation effects [7] upon the fracture surface may be the result of such processes. As a confirmation of such a possible mechanism of MG strengthening one may consider the experimental data of a significant increase of strength and plasticity in MG, alloyed with chromium or molybdenum against initial Fe-B alloys.
4. Conclusion The results obtained from mechanical and Auger spectroscopy investigations lead to the following conclusions. (1) It is shown that the idea of characteristic deformation temperature T*, below which the effect of covalent component in the interatomic bonds on the temperature dependences of o 2 and HV becomes significant, can be expanded over iron-based MG. (2) Values of HV and o~ are presented as a sum of thermal and athermal components (at T<0.7T~°); the metalloid content is responsible for the HV ~' level, whereas HV ~ is defined by the type of metal component. (3) The concepts of two ductile-brittle transition temperatures 7~tbl depending on the alloy composition and Tdb: = 0.7 T, ° is introduced for MG. (4) Segregation effects in sites of most intensive strain localization have been detected. An explanation of the nature of these processes is proposed, which may be used as the base of a phenomenological model of MG strengthening.
Acknowledgments The authors consider as a pleasant duty to thank Academician V. I. Trefilov for his support during the carrying out of this work, Dr. V. P. Ovcharov for help in sample preparation, Dr. Yu. N. Ivashchenko for helpful discussions, Dr. A. I. Pokhodnya, S. V. Postoy and S. S. Ponomarev for assistance in experimental work.
References I C.A. Pampillo, J. Mater. Sci., 10(1975) 1194. 2 Yu. V. Milman, V. P. Ovcharov, S. V. Pan and A. P. Rachek, Soy. Powder Met. and Metal Ceram., 23 (1984) 961. 3 Yu. V. Milman, S. V. Pan and A. P. Rachek, Phys. Metals' (Sov. J.), 7(1987) 195. 4 C. A. Pampillo and D. E. Polk, Acta Metall., 22 (1974) 741. 5 F. Spaepen, Acta Metall., 25 (1977) 407.
132 6 P.S. Steif, F. Spaepen and J. W. Hutchinson, Acta Metall., 30 (I 982) 447. 7 Yu. N. Ivashchenko, Yu. V. Milman, S. V. Pan and S. S. Ponomarev, Phys. Metals (Soy. J.), 7(1989) 1107. 8 E Spaepen, Acta Metall., 23 (1975) 615. 9 V. I. Trefilov, Yu. V. Milman and 1. V. Gridneva, C~st. Res. Technol., 1911984)413. 11) A. I. Taub and E Spaepen, J. Mater. Sci., 1611981 ) 3087. I1 A. Yu. Vinogradov, A. M. Leksovsky, Yu. V. Milman,
S. V. Pan and V. P. Ovcharov, Metallofizika, 11 (1)11989) 66. [In Russian.] 12 A.S. Argon, Acta Metall., 27 (1979) 47. 13 J.J. Gilman, J. AppL Phys., 4611975) 1625. 14 V. M. Goldschmidt, Geochemische Verteilungsgesetze der Elemente (Oslo, 1923) p. 1038; Kristallchemie in HandwOrterbuch d. Naturwissensch. (Jena, 1934). (Cited from Smithells Metal Reference Book, ed. E. A. Brandes, 6th edn. Butterworths, London, 1983).