UDC 839.4.019.3

ANALYSIS

OF T H E S E L E C T I V E

ACTION

OF A G G R E S S I V E

MEDIA

M . I . Chaevskii, I.N. Toropovskaya, and E.A. Kalanchuk Fiziko-Khimicheskaya Mekhanika Materialov, Vol. 4, No. 3, pp. 279-288, 1968 Force interactions of atoms in solid solutions formed in localized regions near microcrack tips as a result of reactions between aggressive melts and strained solid m e t a l s were analyzed in the framework of the statistical theory of regular solutions. Certain conclusiom were reached regarding the effects of deviations of the solutions from the Raoult law on the behavior of solid metals in contact with m e l t s , it was shown that the influence of melts substantially depends on the specific structural characteristics of the solids. Contrary to previously postulated theories it was established that weakening of solid metals m a y take p l a c e even when model solutions show negative deviations from the Raoult law. One of the causes of the absence of such a weakening m a y be the formation of stable diffused layers which hinder piastic deformation; this view was supported by the results of l o n g - t i m e strength tests and metallographic examination. According to previous recommendations [1], the influence of liqnid metals on strained solid metals can be estimated on the basis of a modeling concept of the formation of solid solutions in the defective surface volumes of metals undergoing deformation. In this connection three cases of the interaction between atoms of the solution constituents were considered; these corresponded t o the forces of interaction between dissimilar atoms being equal to, larger, or smaller than the forces of interaction between similar atoms. It was postulated that in the case of the formation of solutions obeying the Raoult law the a t o m i c interaction forces remain unchanged, becoming larger in the case of negative d e v i a tions from the Raoult law and smaller when these deviations are positive [2]. Let us carry out a more detailed analysis of the atomic interaction in such a m o d e l solution in the framework of the statistical theory of nonideal solutions. The energy of the formation of a solution can be approximately represented as a sum of the energies of dissociation of a pair of A (COAA) and a pair of atoms B (COBB) and the energy of formation of two pairs of dissimilar atoms (2~OAB):

2~ ~ AA + ~

-- 2 ~ A ~ 7

:

--

--"

(i)

Z

In accordance with the concepts of the statistical theory of strictly regular solutions, the excess energy co represents the change in the free energy during the isothermal and reversible formation of a given solution. The sign of the excess energy determines the deviation of the solution from i d e a l i t y in the following way: co > 0 means a positive deviation and < 0 a negative deviation; the coefficient ~ was introduced [3] to characterize the effectiveness of interaction between molecules of various components. If the solvent and solute molecules attract each other less strongly than similar m o l e c ules, 7] < 1 (effective repulsion). If the forces of attraction between the solvent and solute molecules are larger than those between similar molecules, 7] > 1 (effective attraction). Large values of 7) characterize c h e m i c a l reactions between the solution components. Consequently, when ~ ~ 1 we have a case of a specific atomic interaction, while 7] = 1 represents the normal interaction, Starting from the assumption that the melting point of a m e t a l characterizes to some extent the strength of its atomic bonds, it m a y be postulated that the energy of interaction of the solvent atoms A (~OAA) is lower than the energy ~5

q

0.3( 02 10

_

_

.

~

=

~

.

,

i

J

10 ~

10 8

70 ~

~. ~ec

Fig. 1. Curves showing the long=time strength characteristic of Sb in argon (curve 1) and liquid Bi-Sb alloy at 380"C (curve 2).

204

of interaction of atoms B (COBB); (it is assumed that a lower potential energy level corresponds to the more stable configuration). When a solution is formed, one m a y deal with one of the following three case~ 1) coAB = coAA - no change in the bond strength of the system; 2) COAB > COAA-- a reduction in the bond strength; 3) coAB < COAA-- an increase in the bond strength. Let us consider the first case coAB = C~

Let co = 0. From Fxt. (1) it follows that

1(

~

(2)

9 ~AA

Hence it is evident that if in this case the Raoult law is valid, we have 71 > 1, i . e . , a specific strengthening of the solution should take piace, compensating the weakening due to the normal interaction of atoms of a l o w - m e l t i n g m e t a l with a more refractory m e t a l .

a/)/ ,

(a)gg

o.8 o.6 o,4

,

,

|

,

0=

,

,

/

J 0,6

/ i

i

o,4}

i .'"

O'2'~li' 0

N

Fig. 2. Variation in the atomic bond strength of the solvent With increasing concentration of a low-melting component: 1) Linear weakening; 2) no weakening at iow concentrations.

,

,/

0.2

gO~a

,

N.

_

. . . .

:

/

. . . . .

R2 0,4 0.6 Zn 0 /72 0,4 0.5 NCu

Fig. 9. Concentration dependence of solutes in copper-base systems: a) Cu-Bi; b) Cu-Sb; c) C u - Z n ; d) Cu-Bi-Sb.

If N = 1, then co = 0 only when COAA = coBB = ~ this being the only condition under which both the Raoult law is satisfied and no changes in the atomic bond strength take place. IfWAA< WBB, co = 0, and N = 1 , we find from (1) that

II)AA "21- ~BB ~A8 - -

2

(3)

Consequently, the Raoult law is satisfied in the case of the normal atomic interaction, but the atomic bond strength should continuously decrease with increasing concentration of the low-melting m e t a l .

% } d00

" 30

200 lO0

?0 I

i

70

fO~

1

f0s ~, rain

Fig. 4. Long-time strength curves of steel 50Kh in argon (curve 1)

and a liquid Pb-Sn-AI alloy

l

i AL

Fig. 5. Load/strain diagrams of copper: a) in argon; 2) in liquid Zn; 3) in liquid Bi-Sb; 4) in liquid BE.

(curve 2).

205

IfT} < 1, then ~o > 0, i . e . , we have a positive deviation from the Raoult law and a more intense weakening of the solution than that possible in the case of normal interaction. And so, if a reaction between a strained m e t a l and a m e l t leads to the formation of a solid solution with a positive deviation from the Raoult law, the m e l t should weaken the solid metal. If ~ > 1, then aJ < 0, i . e . , we have a negative deviation from the Raoult law, and the weakening of the solution may, to some extent, be compensated by the specific atomic interaction. It was postulated [4] that the heat of solution, which at elevated temperatures is related in a definite way to the deviation from the Raoult law, cannot be a reliable criterion of the weakening action of liquid metals, but no concrete examples were cited which could be verified by experiment. For instance, it was concluded [4] that antimony would not be weakened by bismuth; this conclusion is not supported by the results of tests on antimony in liquid bismuth (Fig. 1). It should be pointed out that we know of no experimental data showing that no weakening takes p l a c e under the influence of metals which interact with solid metals to form solutions characterized by positive deviations from the Raoult law. This is an agreement with data cited in [5]. Let us postulate that the change in the a t o m i c bond strength of the solvent (strained m e t a l ) is proportional to the molecular proportion of the dissolved l o w - m e l t i n g m e t a l , i . e . , AO

=

tiN,

(4)

where Aaj = COAA . ~OAB, and a is constant. From the Raoult law a--N.

(5)

Substituting (5) into (4), we obtain A~

=

(6)

~a.

The above substitution is possible only in the case of linear relations in (4> and (5). In real systems relations (4) and (5) are nonlinear. Let A~o = 9Cl(N) and a = fz (N). Graphs of these functions are reproduced in Figs. 2 and 3. Equation (6) shows that the a t o m i c bond strength of a solvent will not change if the a c t i v i t y of the solute is near m zero, which is possible only at negative deviations from linearity (from the Raoult law) and at r e l a t i v e l y low solute concentrations (Fig. 2). Consequently, when solutions of sufficiently high concentrations are formed, a weakening of the solution m a y be observed even in cases of a negative deviation from ideality. Cases are known in which the interaction between dissimilar atoms A-B leads to the formation of stronger atomic bonds than those resulting from the interaction A - A . It is evident that in such cases solid metals under strain should not be weakened by the action of liquid metals.

7

ql

206

i I0 ~

6

0

i 10 ~

' ~

i

~0 ~

~,

sec

Fig. 6. Long-time strength curves of copper at 500* C: 1) In argon; 2) in liquid Zn; 3) in liquid Bi-Sb alloy; 4) in liquid Bi. At 600 ~ C: 5) in argon; 6) in liquid Zn; 7) in liquid Sb; 8) in liquid Bi.

In practice one often observes an increase in the strength of metals undergoing deformation under the influence of liquid metals, and so it is known that under the influence of liquid tin or Pb-Sn eutectic the fatigue strength of notched steel specimens m a y be increased b y 50% [6]. A considerable increase in the endurance of 40 Kh steel under t h e influence of a liquid AI-Pb-Sn alloy was observed [7]. Data on the l o n g - t i m e strength of steel 50 Kh in a liquid AI-Pb-Sn a l l o y ( 3 . 8 wt. % A1, 86o 7 wt. % Pb; 59.5 wt. % Sn) is reproduced in Fig. 4, showing a significant increase in the t i m e - t o rupture of specimens under the influence of the m o l t e n alloy. Our experimental studies were undertaken to verify the above postulates and to determine the causes of the absence of weakening of solid m e t a l s deformed in c o n t a c t with liquid m e t a l s . The short-time strength tests were carried out on hollow copper specimens (D/d = 10/5) with a gage portion 50 m m long. To study the effect of liquid metals (Bi, Zn, Sb) and alloys (50% Bi-Sb) on the l o n g - t i m e strength of copper, tests were carried out on hollow specimens (with one end sealed by welding) filled with these l o w - m e l t i n g metals, or on specimens coated with these metals by vacuum deposition. The l o n g - t i m e strength tests were carried out on type Zst3/3 machines in specially m a d e ampules filled with technically pure argon. The short-time strength tests were carried out on a tensile testing m a c h i n e equipped with a vacuum chamber [8].

Fig. 7. Photomicrographs (x 200) of a copper specimen tested in liquid Zn at 600 ~ C: a) the external part; b) the internal part in contact with the m e l t . Tensile load-strain diagrams of copper specimens are shown in Fig. 5. It will be seen that liquid zinc has no harmful effect on the strength and ductility of copper which is both weakened and embrittled by a liquid Bi-Sb alloy and even more so by liquid bismuth. Long-time strength tests gave similar results (Fig. 6). Liquid zinc produces a slight increase in the t i m e - t o - r u p t u r e of copper at 500 ~ C and a slight reduction at 600 ~ C. Bismuth, antimony and the Bi-Sb a l l o y produce a substantial reduction in the l o n g - t i m e strength of copper at both 500 and 600 ~ C. A positive deviation from the Raoult law is observed in the system Cu-Bi (see Fig. 3) and negative deviations in systems C u - Z n and Cu-Sb. On the basis of recommendations given in [1] one could conclude that copper should be weakened by liquid bismuth but not by antimony and a Bi-Sb alloy. As shown by data in Figs. 5 and 6, this view is not supported by experiment. The cause of the weakening of m e t a l s under strain by the action of l o w - m e l t i n g melts in the case of systems showing a negative deviation from the Raoult law was previously explained. The character of the interaction between solid and liquid m e t a l s depends to a large extent on the concentration of the resulting solution, i . e . , on the kinetics of the process and on the character of the a c t i v i t y curve. Figure 3 shows that the a c t i v i t y of the l o w - m e l t i n g component in certain concentration ranges is near to zero in a l l the systems studied. Consequently, the fact that certain metals are not weakened by the action of melts should be a t t r i buted to factors inhibiting further increase in the concentrations of the solutions formed. To determine the nature of these

207

factors, we carried out comparative m e t a l l o g r a p h i c examination of copper specimens strained in liquid Zn and Bi a t 600~ Zinc was placed inside hollow specimens whose external surface was exposed to argon (Fig. 7a). It was found that the external surface layer was covered by a network of cracks, while a diffuse layer of a C u - Z n alloy was formed on the internal specimen surface (Fig. 7b). In the latter case only one crack formed underneath the diffused layer was observed. It m a y be postulated that the role of the diffused layer consists m a i n l y in that it retards the diffusion process and inhibits the formation of a concentrated solution and that it forms a barrier preventing the emergence of dislocations on the m e t a l surface, i . e . , preventing plastic deformation. Element

Bi

Atomic radius, ,~ 1 8 2 Solid solubility in less Cu (at.%) ~ 0.5~ ~Cu - - "[i

..... "[cu

100 oA 420A

Sb

Zn

Cu

1.61

1.37

1.28

7~6

30~6

26N

7%

No diffused layers were forme, on copper in contact with liquid antimony, bismuth and Bi-Sb a l l o y . Bismuth a c t i v e l y penetrated the grain boundaries causing intergranular cracking of copper (Fig. 8). The following are the necessary conditions for the formation of the diffused layers [9]. 1) The atomic radius of the diffusing e l e m e n t must not exceed the atomic radius of the solvent by m o r e than 15%. 2) The diffusing e l e m e n t should have a sufficiently high sotid solubility in the solvent. Data reproduced in a table show that the formation of diffused layers is possible in the system C u - Z n but not in systems Cu-Bi and Cu-Sb. Fig. 8. Photomicrograph (x 200) of a The interaction of liquid zinc with eopper takes p l a c e by the copper specimen after a test in liquid mechanism of reactive diffusion as a result of which c l e a r l y defined Bi at 600 ~ C. layers of different phases are formed at the c o p p e r / z i n c in•ffaee. The formation of diffused layers of this kind in the system Cu-Zn was oftenobserved and extensively studied [10,11]. Summary 1. When m o d e l solid solutions showing positive deviations from the Raoult law are formed a t the crack tips of solid metals deformed in liquid metals, a weakening of the solid m e t a l should take place in accordance with conclusions reached in [1]. 2) It was established that, contrary to concepts formulated in [1], no weakening of the solid component of a solidliquid m e t a l system showing a negative deviation from the RaouIt law m a y occur if the concentration of the solid solution formed is sufficiently low; this happens when a diffused alloy layer is formed on the surface of a solid m e t a l in contact with a liquid low-meIting m e t a l or alloy. REF ERENC ES 1. M . I . Chaevskii, FKhMM [Soviet Materials Science], no. 6, 1966. 2. O. Kubashevski and E. Evans, Thermochemistry in Metallurgy [Russian translation], I . L . , 1954. 3o M . I . qhakhparonov, Molecular Theory of Solutions [in Russian], Moscow, 1956. 4. V.M. Zalkin, FKhMM [Soviet Materials Science], no. 1, 1968. 5. M~ Chaevskii and V~ V. Popovich, FKNMM [Soviet Materials Science], no. 2, 1966. 6. M . I . Chaevskii, FKhlvIM, no. 7, 1960. 7. M. I. Chaevskyy and V. F. Shatins'kyy, DAN URSR, no. 11, 1962. 8. V. V. Popovich, V. F. Shatinskii, I. V. Kondratenkov, and M. I. Chaevskii, FKhMM[Soviet Materials Science], no. 5, 1968. 9. N.S. Gorbunov, Diffused Coatings in Iron and Steel [in Russian], Moscow, 13, 1958. t 0 . V.I. Arkharov et. a l . , FMM, no. 4, 580, 1966. 11. V . Z . Bugakov, Diffusion in Metals and Alloys [in Russian], Leningrad-Moscow, Oostekhizdat, 1949. 29 November 1967

208

tnstimte of Physics and Mechanics, AS UkrSSR, L'vov