On the determination of diffusion coefficients in chemical diffusion

On the determination of diffusion coefficients in chemical diffusion

ON THE DETERMINATION OF DIFFUSION COEFFICIENTS IN CHEMICAL DIFFUSION* R. W. BALLUFFIT The equations describing chemical diffusion in the general...

300KB Sizes 2 Downloads 88 Views

ON THE

DETERMINATION

OF DIFFUSION

COEFFICIENTS

IN CHEMICAL

DIFFUSION*

R. W. BALLUFFIT The equations describing chemical diffusion in the generalcase where the partial molal volumes of the diffusing components may differ and may vary with composition are reviewed and discussed. Exact and easily applied relations for obtaining chemical and intrinsic diffusivities ape derived. The results should replace the standard Darken equations and Boltzmann-Matano analysis whenever there is any possibility that differences and variations of the partial molal volumes are of significance. SUR

LA DETERMINATION

DES

COEFFICIENTS CHIMIQUE

DE

DIFFUSION

DAN9

LA DIFFUSION

L’auteur examine et discute les equations relatives au coefficient de diffusion chimique da,ns le cas general ou les volumes partiels molaires des constituants peuvent differer ou varier suivant lacomposition. 11 deduit des relations exactes et facilement applicables qui donnent le coefficient de diffusion chimique et le coefficient de diffusion intrinseque. Ces relations pourraient les equations de Darken et de BoltzmannMatano chaque fois qu’il y a possibilite de differences et de variations du volume partiel molaire. UBER

DIE

vos

BEsTIMMUNG

DIEFUSIONSKOEEFIZIENTEN DIFFUSION

BEI

CHEMISCHER

Die Gleichungen, die die chemische Diffusion im allgemeinen Fall beschreiben, wo die partiellen molaren Volurnina der diffundierenden Komponenten verschieden und van der Zusamrnensetzung abhiingig sein k&men, werden zusammengestellt und diskutiert. Zur Bestimmung chemischer und innerer Diffusionskonstant,en werden exakte und leicht anwendbare Beziehungen hergeleitet. Wenn die Mogliehkeit besteht, da13 Unterschiede und mangelnde Konstanz der partiellen molaren Volumina von Bedeutung sein konnen, sollten die Ergebnisse an die Stelle der Gleichungen van Darken und der Durchrechung nach Boltzmann-Matano treten. 1. INTRODUCTION

Most analyses

of chemical

due to differences

diffusion

have ignored

possible effects due to differences in the partial molal

2. GENERAL

volumes of the diffusing components and variations of these partial molal volumes with composition. For example,

the standard

Darken(l)

that the partial molal volumes and equal.

this

not

condition

does

then

becomes

hold

stant cross section.

com-

and

the

analysis

complicated.

diffusion

of

with the development

However,

of more accurate techniques

involved

equations

in the

data.

It turns

application

the

x < 0

t=o

Cl = cl(+co)

x > 0

t=o

with respect

\

I.

/I

\

\A)

to the non-

COfor all t. Also,

2

=

-_D.

.%

lax

+

c.v

z

2)

where i = 1 or 2. The Di are the intrinsic diffusivities of Darken(l) and Hartley and Crankc6) and measure the fluxes relative to the crystal which is mechanically flowing with a velocity, u, relative to x = 0. As Bardeen and Herring(‘) point out, equation (2) holds for a vacancy diffusion model as long as vacancy supersaturation effects are negligible. This condition

* This research was supported by the United States Air Force through the Air Force Office of Scientific Research. Received February 15, 1960. t Department of Mining and Metallurgical Engineering, University of Illinois, Urbana, Illinois. 1960

when

is

Cl = C1(-co)

J.

general

is not much greater than in the case where

VOL. 8, DECEMBER

constant

the origin are

constant and equal partial molal volumes are assumed. It is strongly recommended that the general relations be used whenever there is any possibility that effects

ACTA METALLURGICA,

essentially

W,/ax = 0 at 2 = *co. The fluxes across any section fixed with respect to

out that the

of

on actual diffusion

diffused end of the couple at x = -

of the present paper is to present a set of exact and easily applied relations for obtaining chemical diffusivities from experimental

remains

The x = 0 plane is fixed

for

measuring diffusion coefficients, these effects should be of significance in a number of systems. The purpose

labor

couple

The initial condition

by workers in the field since they are

usually small (at least in metallic systems).

Measurements

the couple is sufficiently massive in the x directions.(3-5)

An

excellent discussion of this problem has been given by Crank@). These complicating effects generally have been ignored

EQUATIONS

couples have shown that the cross section of such a

In the general case,

more

of the partial molal

We shall limit our discussion to unidimensional diffusion in a binary sandwich-type couple of con-

assume

of the diffusing

ponents are constant diffusion

equations

and variations

volumes are significant.

should be met in most diffusion 871

couples at diffusion

872

METALLURGICA,

ACTA

times of interest.(s)

The Ci are measured in moles per

+ CJ,

= 0

We first note that

of 1 of the form

(3) Since the Bi are functions

F1 dC, + v, dC, = 0 1 Pi are partial

C, = C,(A) is assumed.

= 1

C,dEp, + C&P2

where the

solution

8, 1960

the Bi are functions

unit volume and obey the standard equations CJ,

VOL.

of the composition

molal volumes.

Prager@) and Crank@) the accumulation

Following

equations

are

coefficient

However,

the 0,

are not functions of composition in the general sense since the relation 1 = 3L(C1) depends upon the particular

boundary

under

investigation.

equation where D is the chemical diffusion

of 1, they are also functions

since I = A(C,).

conditions of the diffusion couple The Matano integration of

(9) yields

which is

x dCi.

(12)

given by

D = D&7, The velocity v=

(5)

+ D,V,C,.

of flow is

Using equations (12), (10) and (3) we solve for D in the convenient form

D)%+ 2 ax

V(D,1

(6) We note that the macroscopic terms

of

However, not

a

single

diffusion is described in

chemical

diffusion

coefficient.

the usual simple forms of Pick’s

pertain.

The

is a small velocity

integral

term

in

laws do

equation

term arising from variations

partial molal volumes tions will generally

with composition.

cause the couple,

(6)

of the

Such variaas a whole, to

swell or shrink. 3. THE

DETERMINATION AND INTRINSIC

expedited

scaling down the x coordinate by the factor (VI

of the penetration

curve

It is stressed that the origin

- V,)/V,.

is located at the x = 0 plane under all circumstances. C,( +m)

OF THE CHEMICAL DIFFUSIVITIES

the integral in equation

two approximate the following

small compared to the first and is readily evaluated by

The plane delined by

The determination of the chemical diffusivity, D, is somewhat complicated by the fact that it appears behind

Equation (13) may be solved graphically by the usual means. The second integral will generally be quite

methods

(4).

we solve for D exactly.

by introduction

Crank(s) reviews

for its determination.

In

The solution is

of the functions

s C,( - m) will not generally Inspection

xdC,=O

coincide

with the x =

0 plane.

of equations

(7) and (12) shows that coinci-

dence is only obtained

in the special case when the

couple as a whole does not expand or shrink. This result may also be demonstrated

by integrating

under the

diffusion curve to find the amount of either component which has crossed the x = 0 plane during the diffusion. When intrinsic

Then

j. z = __8.

a at

-z

a

data are available, be obtained from

the the

relations

!!!3

8

marker movement diffusivities may

ax ’

( 1 eiz,

(9)

and

D = c,v,e,

+ c,v,e,.

(10)

For the present boundary conditions, the Bi may be determined by the usual Boltzmann-Matano method

The slopes and integrals

where the variable

are the same as in equation

il E xt-*

is introduced

and the

in equations

(14) and (15)

(13) ; and, therefore, the

BALLUFFI:

Di are readily determined

DIFFUSION

COEFFICIENTS

if the marker velocity,

vn, is

IN

CHEMICAL

approximation.

by

showing

that

an inert

marker

initially placed at x = 0 will move parabolically

with

and

V, are

quite simple. vanishes,

displacement:

V2 is 2tv, = xm.

The instantaneous equation

(6).

velocity

Making

VP, - 4)

v=

(16)

of any marker is given by

the A substitution

we have

unknown,

it is suggested

forms

of

behavior

of

constant

method

system

of coping

with volume

is to employ where

equal

is satisfied

by

a particular

marker

traveling at constant composition at a constant value of 1, say il,, whose motion is given by xm = ;l,t*. In this case v, = dx,&?t &(D,

+/ym&)

-

increments

of diffusion

ordinate

in the direction

numbers

of atoms.(296J10) In this type of space, the

contain

(2)

when

previous

may

coefficients

is not markedly Darken

be readily

applied

as functions of composition.

In such a case, the labor standard

= constant.

REMARKS

equations

VI and Vz are known

d1]

equation0

in determining greater

diffusion

than when the

are employed

must

be known ;

gration

must be carried

lations are required.

However,

D by this method,

a Boltzmann-Matano out;

VI

and

type

and further

The advantages

equal

V,

inte-

manipu-

of this method,

therefore, appear open to question. 1. L. S. DARKEN, Trans.

D,)%‘(U

(2)

coor-

of the co-

REFERENCES

;l,,,/2tg where

=

4. CONCLUDING

The

(18)

changes

a non-laboratory

simple form of Fick’s laws always pertain. equation

and

that the investigator

VI and V2.

Another

in order to determine This

Fick’s

VI

measure the density of the two terminal alloys employed in the couple and use the approximation of

dinate

an

and the usual

When the detailed

during diffusion

acl

VI

The equations then become

The integral term in equation (4) and (6)

19~= Di,

laws pertain.

i.e.

is obtained when it is assumed that

each constant.

time. For this marker then, the term 2tv, in equations (14) and (15) is simply represented by the marker

873

In many cases, an excellent first order

approximation

known. We conclude

DIFFUSION

as an

Awwr. In& Min. (Metall.) Engrs. 175, 184 (1948). 2. J. CRANK, The Mathematics qf Diffusion p. 219. Oxford University Press (1956). _ __ _ 3. L. C. C. DA SILVA and R. F. MERL, Trans. Awaer. Inst. Min. (Metall.) Engrs. 191, 155 (1951). 4. R. W. BALLUFFI and L. L. SEIQLE, J. Appl. Phys. 25,607 (1954).

5. R. RESNICK and R. W. BALLUFFI, Trans. Awser. Inst. Min. (M&all.) Engrs. 203, 1004 (1955). 6. G. S. HARTLEY and J. CRANK, Trans. Faraday Sot. B42, 12 (1946). 7. J. &D&EN and C. HERRING, Atom. Movements p. 87. American Society for Metals, Cleveland (1951). 8. H. FARA and R. W. BALLUFFI, J. AppZ. Phys.

30, 325 (1959). 9. S. PRAGER, J. Chem. Phys. 21, 1344 (1953). 10. M. COHEN, C. WAQNER and J. E. REYNOLDS, Trans. Amer. Inst. Min. (Metall.) Engrs. 197,1534 (1953); 200,702 (1954).