Diffusion of copper, aluminum and boron in nickel

Scripta METALLURGICA et MATERIALIA

Vol. 29, pp. 959-962, 1993 Printed in the U.S.A.

Pergamon Press Ltd. All rights reserved

DIFFUSION OF COPPER, ALUMINUM AND BORON IN NICKEL

Masayuki Hasaka', Takao Morimura', Yasuo Uchiyama ~, Shin-ichiro Kondo', Tetsuya Watanabe 2, Kunihiro Hisatsune:' and Tatsu)i Furuse' 'Department of Materials Science and Engineering, Nagasaki University, Nagasaki 852, 3apan 2Department of Chemical and Biological Engineering, Sasebo College of Technology, Sasebo 857-11, 3apan 3Department of Dental Materials Science, Nagasaki University School of Dentistry, Nagasaki 852, 3apan

(Received July 8, 1993) Introduct ion Secondary ion mass spectroscopy(SIMS) is a valuable technique for analyzing the distributions of solutes which diffused from the surface to the interior side of metals. The high resolution of SIMS can determine small diffusion coefficients below 10- ' z cm2/s observed at low temperature,: and also measure the diffusion coefficients of various solutes including aluminum and boron isotopes which are not easily available(l-10). By means of a radioactive tracer technique, the diffusion coefficients of various solutes in nickel have been measured only at temperature hiRher than 0.7T~, where Tm is the melting temperature of nickel(t1)(12). With the benefit of SIMS, the present paper investigates the diffusion coefficients of copper, aluminum and boron in the blocks and the melt-spun ribbons of nickel from 0.7T~ to 0.36Tin. Experimental Procedure As media for diffusion of copper, aluminum or boron, the blocks and the ribbons of nickel were prepared from 99.95%Ni bulks. Some bulks were cut into the blocks of 5mmxSmm square and 4ram thickness, and the other bulks were melt-spun to form the ribbons of 90/z m thickness with a copper roll of a rotational speed of ]0m/s. The blocks were pre-annealed for 3.6ks or 360 ks at 1073K to reduce lattice defects. The crystal grain sizes of the blocks were 15 and 100~ m, depending on pre-annealing time. The crystal grain size of the ribbons was 10/zm as shown in Fig. 1. The block surfaces polished with a buff, and the ribbon surfaces as melt-spun were covered by a vapor deposited copper, aluminum or boron layer of 0.01/z m thickness. Subsequently, the copper, aluminum or boron layer was covered by a vapor deposited nickel layer of 0.05/z m thickness in order to protect the copper, aluminum or boron layer from oxidation and contamination(8). The blocks and the ribbons covered by the double layers were isothermally annealed in vacuum of 10-~Pa at temperatures of 1269K-673K for t.2ks-3370ks. The measurements of the distributions of copper, aluminum and boron along the depth direction were performed with a secondary ion mass spectrometric analyzer (Hitachi-2A) by using sputter ions of O~' An electronic aperture permitted to measure the intensity of secondary ions which came out from the 100/z m square of central portion of the eroded crater. Under the assumption of a constant sputter rate, a depth at each time was estimated from a depth observed at a final sputter time. Results and Discussion The distribution equation,

of

solute

is given

on

the

basis

of

the

thin

film

959 0956-716X/93 $6.00 + .00 Copyright (c) 1993 Pergamon Press Ltd.

solution of

the

diffusion

960

DIFFUSION

IN NICKEL

c=(M/2 ~ ' ~ [ e x p ( - x =/4Dt)+exp(-(x+H) =/#Dt)},

Vol.

29,

No.

(1)

where c is the concentration of a solute at a position x a f t e r a diffusion interval t, x is measured from the maximum of the concentration, D is the diffusion c o e f f i c i e n t of the solute, and M is the i n i t i a l amount of the solute, and H/2 is the thickness of a deposited nickel layer. The second term in this equation is the contribution from r e f l e c t i o n at the surface, namely, the concentration superimposed on the original one. The diffusion c o e f f i c i e n t D is evaluated from the slope of the plot of In c versus x 2, a In c l a x 2 =-( 114Dt)[ 1* ( H / x ) / [ ] *exp((2xH*H 2)l#Dt)}]. For the SlOpe,

region

of

H/x<< I,

0 In c / 0 x e ~ - I / ¢ D t .

the

diffusion

coefficient

(2) D

is

satisfactorily

evaluated

from

the

(3)

Fig. 2 shows the intensity profiles of copper and aluminum ions in the blocks pre-annealed for 3.6ks and the ribbon. As the ion intensities are supposed to be proportional to the concentrations, the plots of In c versus x 2 for the blocks are recognized to keep good linearity. Consequently, the diffusion coefficients in the blocks were estimated from eq.(3), in this paper. However, a plot of In c versus x z for the ribbon deviates from a straight line at the region up to x=0.7/L m which satisfies H/x<< I. This is because copper is not fully dissolved in the ribbon for the sake of a short diffusion interval. Thus, the diffusion coefficients in the ribbons were estimated f r o m the slope of the plot of the error function type, I n ( - 0 c / 0 x) versus x s, as reported previously(S). Fig.3 shows the temperature dependence of the diffusion coefficients of copper in nickel. Two straight lines of the Arrhenius type are drawn for the blocks pre-annealed for 3.6ks. The a c t i v a t i o n energy and the pre-exponential f a c t o r at high temperature were Qc==251-+7k3/mol, Do cu=0.41cm2/s at 1269K-872K in the blocks pre-annealed for 3.6ks. These values at high temperature are in good agreement with the values reported by Kadoma et al.(I 1) , Qc°=2583/mol, Do c==0.57cmS/s at 1632K-1327K. This agreement guarantees that secondary ion mass spectroscopy as well as a tracer technique is available for measuring diffusion coefficients and suggests that lattice diffusion is the predominant mechanism above 872K for the blocks pre-annealed for 3.6ks. The diffusion coefficients of copper at low temperature become smaller in the blocks when the pre-annealing t i m e increases from 3.6ks to 360ks, and become larger in the ribbons than in the blocks. This behavior means that the increase of crystal grain boundaries results in the occurrence of shortcircuiting diffusion. The a c t i v a t i o n energies and the pre-exponential factors at low temperature were

Qc==67---3k3/mo[, Do c==1.3x10 '= 'cm=/s at 872K-722K in the blocks pre-annealed for 3.6ks, Qc==71___ 5k3/mol, Do c = = 7 . 2 x l 0 - ' S c m 2 / s at 8~0K-690K in the blocks pre-annealed for 360ks, and Qc u=45__.2k3/mol, Do c == 1.7x10- ' ' cm =/s at 823K-673K in the ribbons. Fi8.~ shows the temperature dependences of the diffusion coefficients of aluminum and boron in the blocks pre-annealed for 3.6ks. Two straight lines of the Arrhenius type are drawn for aluminum, which yield the a c t i v a t i o n energies and the pre-exponential factors at high and low temperatures, Q^,=2qS+_ 7k3/mol, Do ^ ,=0.~t~cm~/s 1269K-919K, and Q ^ , = 9 0 _+ 5k3/mol, Do ^ , = l . 2 x l 0 - g c m 2 / s at 919K-722K in the blocks p r e - a n n e a l e d for 3.6ks. The values a t high t e m p e r a t u r e a r e in good a g r e e m e n t with the values reported by Akimova e t al, (12) ,

Q^ z=257-258k3/mol, Do ^, = 1.6- ] .Scm 2/s at ] 623K- 1116K. On the other hand, one straight line of the Arrhenius type is drawn for boron because diffusion temperature may be l i m i t e d to low temperature. The a c t i v a t i o n energy and the pre-exponential f a c t o r were QB=110__.2k3/mol, Do e=2.3x10 2cm2/s at 973K-773K in the blocks pre-annealed for 3.6ks. The a c t i v a t i o n energy for diffusion of boron is not smaller than that of copper or aluminum,

7

Vol.

29,

No.

7

DIFFUSION IN NICKEL

although the diffusion coefficient temperature.

of

boron

is

larger

than

961

that

of

copper or

aluminum at

!

e 8Cu

low

References (1) W. T. Petuskey : Nontraditional Methods in Diffusion, Philadelphia, Pa. USA, The Metallurgical Socity/AIME, 179(1984), (2) F. Degreve, N, A. Thome and 3. M, Lan8 : 3, of Materials Science, 23, 4181(1988). (3) P. Dorner, W, Gust, B. Predel and U. Roll : Phil. Mag., A49, 557(1984). (#) K. Ahiborn and W. SchrO ter : Phil. Mag., A48, 661(1983). (5) K. Kanada, S. 5hinoyama and A. Katsui : 3. Appl. Phys., 55, 2881(1984). (6) D. Mathiot, and G. Edelin : Phil. Ma8., A41,447(I 980). (7) 3. Verlinden and R. Gijbels : Adv. Mass. Spectrom., 8A, 485(1980). (8) R. DO hi, M. P, Macht and V. Naundorf : Phys. Status Solidi(a), 86, 603(1984). (9) W. Gust, C. Ostertag, B. Predel and U. Roll : Phil. Mag., A47, 395(1983). (I0) P. Dorner, W. Gust, M. B. Hintz, A. Lodding, H. Odelius and B. Predei : Acta Metall, 28, 291(1980). (1 l) Kadoma, 5udo and Oikawa : 3. 3apan Inst. Metals, 28, 192(1964). (12) I. A. Akimova, B. M. Mironov and A. V. Pokoev : Izv. V. U. Z. Tsvetn. Metall, 5, 111(1985).

.......

~

I 1~Ir~-O~•~--

,

-~kO.o. O

0

I016K

in Ni

60ks

~'w"~O~o.,,

\

°o

*,,.

~ I

20K 1•44.2ks

=vAt in Ni J I

......

0 i

30~

m

Fig.1 Cross section of the Ni ribbon observed by SEM.

1 lOexS(crn s)

FIG.2 Diffusion profiles of e=Cu and =TAt in Ni. •-blocks pre-annealed for 3.6ks, O -ribbon.

2

962

DIFFUSION

10-8

I

\

\

% \

10-1o

~N

\

\

IN NICKEL

i

Vol.

i

i

Kadoma et al.

I0 -l=

! m

d 10-i4 .

lO-le

0.6

J

J

!

J

0.8

t .0

1.2

1,g

10 8 / T ( K -

l)

FIG.3 Diffusion coefficients of aSCu in Ni. O - b l o c k s pre-annealed for 3.6ks, • - b l o c k s pre-annealed f o r 360ks, [ ] -ribbons.

10-e

I%%

I

|

m

&

%%% A l d m o v a et al. I0-Io

10" l = o .{

I0-14

.% %

I0-16

I 0.6

I

I

I

!

0.S

1.0

1.2

1.#

i 08/T(K - =) F I G . , Diffusion coefficients of =TAI and t=B in Ni.

O-=~AI, • .l 'B,.

29, No.

7