4111066015 材料二 邱禹儒 物理冶金Chapter12

Physical metallurgy principles

Diffusion in substitutional solid solutions

Content
 Diffusion in an ideal solution
 Fick’s first law
 Fick’s second law
 Kirkendall effect
 Matano method
 Darken’s equations
 Intrinsic diffusivities
 Self-diffusion
 Chemical diffusion
 Grain boundaries and free surfaces
diffusion

 the average frequency with which an A atom jumps from X to Y is 1/6

Diffusion in an ideal solution  the flux of solute atoms from plane X to plane Y is
1
 Ideal solution: there is no interaction between solute J X Y  (n A a)
6
and solvent atoms or that the two forms act inside J X Y : flux of solute atoms from plane X to Y per unit cross - section
the crystal as though they were a single chemical  : mean time of stay of a solute atom at a lattice site
species. n A : number of A atoms per unit volum e
a : lattice constant of crystal

 concentration of A atoms in plane Y is :

dn
(n A )Y  n A  (a) A
dx
dn a
 J Y  X  [n A  (a) A ]
dx 6
 net flux ( J ) J  J X Y  J Y  X
a dn a
ex. a simple cubic with  100  direction along the axis of the bar  (n A )  [n A  (a) A ]
6 dx 6
the difference in compositio n between two adjacent t ransverse 2
a dn A
dn 
atomic planes is  A 6 dx
dx a2 dn
a : the interatomi c (or lattice spacing) if D   net flux J   D A ......Fick' s first law
6 dx
 : the mean time of stay of an atom in a lattice side D : diffusivit y ; or diffusion coefficient
( the average frequency with which the atoms jump :1/ ) J : quantity per second, of diffusing matter passing normally t hrough a unit
area under the action of a concentration gradient dn A / dx

The diffusivity D in Fick’s first law

was assumed to be constant for
measurements made at a fixed
temperature.

Variation of the diffusivity with

composition in a binary gas solution
Variation of the diffusion
of oxygen and hydrogen.
coefficient with composition.
Diffusion rate: vapor > liquid > solid
Variation of the diffusion : solid > liquid > vapor

1
Fick’s second law
 Fick’s second law is the basic equation for the
experimental study of isothermal diffusion.
 Two standard methods of measuring the diffusion
coefficient :
 1. Grube method: the diffusivity is assumed constant,
is strictly applicable only to those cases in which the
diffusivity varies very slightly with composition.
 2. Matano method: the diffusivity is assumed as a
function of composition.

Kirkendall effect
 In a binary solid solution, each of the two atomic
forms can move with a different velocity.
 Heated to a temperature close to the melting point
of the metals comprising the bar.

 The wires moved during the diffusion process.

 The distance has been found to vary as the square
root of the time during which the specimen was
maintained at the diffusion temperature.
x t

• The movement of the wires during the diffusion process is

for the A atoms to diffuse faster than the B atoms.
Mechanisms:
• 1. The motion of a single atom at a time – vacancy and
interstitial diffusion.
• 2. The cooperative movement of two or more atoms.

2
  Direct interchange diffusion  The vacancy mechanism is the correct mechanism
mechanism. for diffusion in FCC metals.
(substitutional solid solution diffusion)
 1. The various methods that have been proposed to
explain the movements of atoms in this type of
crystal, it requires the least thermal energy to
activate it.
 Zener ring mechanism for diffusion
(substitutional solid solution diffusion)  2. The Kirkendall effect in has now been observed in
 Preferred diffusion mechanism in a great many diffusion experiments involving
BCC metals because their structure couples composed of FCC metals.
is more open than that of close-
packed metals (FCC, HCP)  BCC metals: that diffusion by a ring mechanism
 Ring mechanism is more probable might require less thermal energy than vacancy
mechanism than the direct diffusion, the discovery of a Kirkendall effect in BCC
interchange metals indicates that these metals also diffuse by a
 But BCC structure has the Kirkendall vacancy mechanism.
effect has been observed.  HCP metals: is not quite as clear , results of
diffusion experiments are in accord with a vacancy
mechanism.

Porosity  Net flow of vacancies occurs from the Ni-

 In Kirkendall experiment demonstrates that the rate rich side of the bar toward the Cu-rich side.
at which the two types of atoms of a binary solution
diffuse is not the same.  Vacancies must be created on the side of a
 The element with the lower melting point diffuse couple that gains mass (Ni-rich side), and
faster. (move rate: Zn>Cu; Cu>Ni) absorbed on the side that loses mass (Cu-
rich side).
 Grain boundaries and exterior surfaces are
probable positions for both the creation and
elimination of vacancies.
 The most important mechanisms for the
creation of vacancies is dislocation climb,
and the dislocation climb and void
formation account for most of the absorbed
Right region: loss mass; vacancies (pores); 2-D tension stress
vacancies, positive climb being associated
Left region: gains mass; 2-D compressive stress
with the removal of vacancies and negative
Stress field  plastic deformation  structural changes 
climb with the creation of vacancies.
formation of substructures, recrystallization, grain growth.

3
 Darken’s equations – to determine the intrinsic diffusivities
Darken’s equation experimentally.

n A n  Assumptions:
J A   DA and J B   DB B  1. it is assumed that all of the volume expansion and
x x
J A and J B : the flux (number of atoms per second passing through a unit contraction during diffusion, due to unequal mass flow,
occurs only in the direction perpendicular to the weld
cross - section area) of the A and B atoms interface.
D A and DB : the diffusivit ies of the A and B atoms  2. The total number of atoms per unit volume is a const.
(intrinsic diffusivit ies; function of compositio ns) nA + nB = constant
n A and nB : the numbers of A and B atoms per unit volum e
The velocity v of the Kirkendall
markers may be written :
x  x' volume 1
v  
dt seconds unit area
The marker velocity is equal, but
opposite in direction to the volume
of matter flowing past the markers
per second divided by the cross -
section area A of the bar at the marker.

volume 1 J net
 
seconds area nA  nB
1/ nA  nB : the volume per atom
nA n
J net  J A  J B   DA  DB B
x x
the marker velocity gives us the following equation:
n n mm : cross - section at x;
( DA A  DB B )
volume 1 x x nn : cross - section at x  dx
v  
seconds area (nA  nB ) the two cross - sections define a volume :1  dx
nA nB * consider the rate at which the number of A atoms
NA  ; NB  (nA  nB : constant) changes inside this volume.
nA  nB nA  nB
v : the volume of metal flowing past the cross - section
N B N
 N B  1  N A and  A at x in a second.
x x
n A : the number of A atoms per unit volum e
N
 marker velocity: v  ( DA  DB ) A -- first of Darken's eq. n Av : a number of A atoms is carried each second through
x
the cross - section at x.
Above equation relates the two intrinsic diffusivities to
 the total number of atoms per second crossing the
N
the marker velocity (v) and the concentration gradient ( A ) n
x boundary at x is : ( J A ) x   D A A  n A v
x
two quantities that can be experimentally determined.

Substitutional diffusion
The flux of A atoms through the cross - section at x  dx is
( J A ) x Kirkendall effect-
( J A ) x  dx  ( J A ) x   dx
x The occurrence of Kirkendall shift means that the entire crystal lattice is
 The rate at which the number of atoms inside the volume dx changes actually moving with respect to the observer during the diffusion process.
 n
( J A ) x  ( J A ) x  dx  [ D A A  n A v]  dx Let vB = bulk velocity of the lattice = velocity of the markers = vm
x x
The rate at which the number of A atoms per unit volum e changes : vD = velocity due to diffusion alone = velocity of atoms relative to markers
( J A ) x  ( J A ) x  dx n A  n Then we may write vtotal = vB+vD=vm+vD
  [ D A A  n A v]
dx t x x Flux of atom i can be written as Civ
N A  N A
  [DA  N A v] The total flux of atoms A with respect to an observer is CA(vm+(vD)A)
t x x
 N A  1 NB dC A
But C A (vD ) A   DA
N  N  ~ N A dx dC A
 A  [ N B D A  N A DB ] A  D  Fick' s second law For both components ( J A )T  C Avm  DA
t x x x x dx
~ dCB
 D  N B D A  N A DB - - second of Darken' s equation ( J B )T  CB vm  DB
dx
(JA)T=-(JB)T
dN A
vm  ( DA  DB ) NA=CA/Co
dx

dC A dC
( J A )T  ( DA N B  DB N A )  D A
dx dx
Fick’s 1st law
dCB dCB
( J B )T  ( DA N B  DB N A )  D
dx dx
D : mutual diffusion coefficient

DA, DB: intrinsic diffusion coefficient

C A  C  C
 [ N B DA  N A DB ] A  D A  Fick's second law
t x x x x

4
 Grube method
if the diffusivity D is assumed to be a constant,
N A  N A 2 N A
 Fick's second law:  D D
t x x x 2
two alloys of the elements A and B, one having the
composition N A1 and the other composition N A2 , ex. diffusion couple is formed by the welding two alloys of the elements A and B
N A1 : 40% A (the alloy to the right of the weld)
at the start of the diffusion process, N A2 : 50% A (the alloy to the left of the weld)
( N A2  N A1 ) x heated at temp. T1 and held 40 hr (144,000sec)
N A  N A1  [1  erf ] for    x   at distance : 2  10 -3 m to the right of the weld the compositio n N A is 42.5% A,
2 2 Dt (0.5  0.4) 0.002
 0.425  0.400  [1  erf ~ ]
2 y  y2 2 2 D(144,000)
where error function: erf ( y ) 
 0  e dy  erf ~
0.001
 0.5  ~
0.001
 0.477
D(144,000) D(144,000)
~ 11 2
 D  3.04  10 m / s

Grube method : the compositio n of the specimen at an arbitrary point 

the evaluation of the error function  argument y  the value of the diffusivit y

if the same condition as above; but we want to obtain the same compositio n
(42.5% A) at twice the distance from the weld interface :
x1 x2 x2 x2
 ~  ~ or 1  2
2 Dt 2 Dt t1 t2
1 2
 if x1  2 10 3 m
x2  4  10 3 m
t1  40hr  t 2  160hr

 x 
C  Cs  (Cs  C0 )erf  
 2 Dt 

5
 Fe3C

ex. the diffusivit y at N A  0.375 (C point)

Matano method x
: the reciprocal of the slope of the penetratio n curve at point C.
N A
( 610 m -1 )
integratio n from N A  0.375 to N A1  0 (4.66  10 -4 m)
~ 1 1
D(0.375)  ( )  (area from N A1  0 to N A  0.375)
N A  ~ N 2t slope at 0.375
Fick' s second law :  D( N A ) A if t  50 hr (180,000s)
t x x
~ 1 1
Matano - Boltzmann method of determinin g the diffusivit y  D(0.375)    4.66  10  4  2.1 10 12 m 2 / s
~ 2(180,000) 6.10
D uses graphical integratio n.
steps :1. plot a curve of concentration versus distance along Determine the diffusivity at any concentration not too close
the bar measured. (assumed the number of atoms per unit to the terminal compositions (NA=0 and NA=1).
volume (n A  n B ) is constant.
2. to determine that cross - section of the bar through which there
have been equal total fluxes of the two atomic forms (A and B)
 the Boltzmann solution of Fick' s equation is :
~ 1 x N A
D 
2t N A N A1
xdN A
t : time of diffusion
N A : concentration in atomic units at a distance x measured from the
Matano interface
N A1 : concentration of one side of the diffusion couple at a constant
compositio n point well removed from the interface

Determination of the intrinsic diffusivities

The markers move in such a manner that the ratio of their displaceme nt squared  Generally agreed that diffusion in metallic
to the time of diffusion is a constant.
substitutional solid solutions is the result of the
x2
 k movement of vacancies, the factors that control
t
k : constant x : displaceme nt t : time of diffusion v : marker velocity jump rates into vacancies to the two different
x k x atomic forms are complex and not completely
v  
t 2 x 2t understood.
if x  0.0001m and t  50hr (180,000 s )
0.0001
v  2.78  10 10 m / s
2(180,000) Because:
if at the position of the markers N A  0.65; N B  0.35  1. The solution was not ideal.
~ N
DM  5.5  10 12 ; A  244m 1  2. The jumping rate of A or B atoms was dependent
x
Darken equations : of composition.
~
D  N B D A  N A DB
N A
v  ( D A  DB )
x
 5.5  10 12  0.35 D A  0.65 DB
2.78  10 -10  ( D A  DB )244
 D A  6.24  10 12 ; DB  5.10  10 12

Self-diffusion in pure metals

The diffusion of a solute consisting of a radioactiv e isotope in a solvent that
is a nonradioac tive isotope of the same metal.
(only small mass difference between the isotopes)
1 m 1
 ( )
* m* 
1  and 1  * are the jump - rate frequencies of normal and radioactiv e isotopes
( ,  * : the mean times of stay; m, m* : the mass of isotope)
because m  m *
~
 D  N B D A  N A DB  ( N A  N B ) D  D
~
 interdiffu sion coefficient ( D)  intrinsic diffusivit ies ( D  D A  DB )

Because self - diffusion in pure metals ocuurs in an ideal solution and with a
diffusivit y that is indepedent of concentration.
a2
for self - diffusion in a simple cubic system : D 
6
a2 a2
FCC metals : D  ; BCC metals : D 
12 8
 a2
in general, for any lattice : D 

 is a dimensionl ess constant depending on the structure.

6
  ( H  H ) / RT
in vacancies, ra  Ae m f
Because the diffusion coefficient is a function of compositio n and temperature.
ra : the number of jumps made per second by an atom ( 1  )
for the self - diffusion coefficient :
H m : the enthalpy change or energy barrier for overcome to move a mole ( S m  S f ) / Re  ( H m  H f ) / RT
of atoms into vacancies D  a 2e e
( S  S ) / R
H f : the enthalpy change or work to form a mole of vacancies suppose Do  a 2e m f and Q  H m  H f
 A may be replaced by Zν ( Do : frequency factor; Q : activation energy of diffusion)
 ( H  H ) / RT Q / RT
 r  Ze m f  D  Do e
( Z : the lattice coordination number;  : the lattice vibration frequency)  logarithm (2.3 log 10 ( x)  ln( x))
if consider the entropy changes associated with the formation and movement Q
log D    log Do
1  ( Gm  G f ) / RT 2.3RT
of vacancies :  Ze
 (ordinate intercept : log Do , and the slope : Q / 2.3R)
(Gm  H m  TS m ; G f  H f  TS f )
 D  a Ze 2  ( Gm  G f ) / RT
 a 2e m f  ( G  G ) / RT Ex.
(for BCC,   1/8, Z  8; FCC,   1/12, Z  12; so, Z  1)
The jump rate of atoms into vacancies (ra) varies directly as:
1. The no. of atoms (Z) that are next to a vacancy,
2. The frequency (v) that an atom moves toward a vacancy,
 H / RT
3. The probability that an atom will have sufficient energy to make a jump (e m )
 H f / RT
4. The concentration of vacancies in the lattice (e )

Chemical diffusion at low-solute concentration

If the solute concentration is very small,
 D  N B DA  N A DB
N A  1 ; N B  0  D  DB
(at very low solute concentrations, the chemical interdiffusion coefficient
approaches the intrinsic diffusivity of the solute)
( S Bm S Bf ) / R  ( H Bm  H Bf ) / RT
 DB  a 2 e e
 , S Bm , S Bf , H Bm , H Bf : for the solute atoms in the solvent lattice
( S Bm S Bf ) / Re
DBo  a 2 e ; QB  H Bm  H Bf
Q ex. the diffusion of several different solutes in Ni, (diffusion couples consisting
m  8000
2.3R of a plate of pure Ni welded to another composed of an alloy of Ni-1 atoms %
 Q  2.3(8.314)8000  153,000 J / mole
indicated element)
The ordinate intercept of the experiment al curve has the value - 3.3
 Do  10 b  10 3.3  5  10  4 m 2 / s
 D  Do e Q RT  5  10  4 e 153,000 / RT m 2 s 1

All diffusion coefficients tend to follow an empirical activation law :

for self - diffusion : D *  Do*e Q / RT
~ ~ ~
chemical diffusion : D  Do e Q / RT ( D  N B D A  N A DB )
and D A  D Ao e Q A / RT ; DB  DBo e QB / RT

Chemical diffusion using radioactive tracers

 ln  A
D A  D *A (1  N A )
N A
*  ln  B
DB  DB (1  N B )
N B
D A , DB : intrinsic diffusion coefficients
D *A , D *A : tracer diffusion coefficients
 A ,  B : activity coefficients of two components
N A , N B : atom fractions of two components
Gibbs - Duhem equation : N A  ln  A   N B  ln  B
(N A  N B )
N  ln  A  N B  ln  B N  ln  B
 1 A  1  1 B
N A  N B N B
( thermodynamic factor)
N  ln  A
If in an ideal solution, a A  N A   A  1  1  A  1 0  1
N A
 D *A  D A and DB*  DB
* Express the chemical diffusivit y in terms of the tracer - diffusion coefficients
~
Darken' s equation : D  N B D A  N A DB
~  ln  A  ln  B
 D  N B D *A (1  N A )  N A DB* (1  N B )
N A N B
~ * *  ln  A
orD  ( N B D A  N A DB )(1  N A )
N A

Diffusion along grain boundaries and free surfaces

Atom movements in solids are not only the interiors of crystals but also the
surfaces and boundaries of crystals.
the surface and grain - boundary forms of diffusion also obey activation
(Arhennius - type laws)
 Ds  Dso e Qs / RT and Db  Dbo e Qb / RT
( Ds and Db are the surface and grain - boundary diffusivit ies.
in general : Ds  Db  Dl

 Surface diffusion plays an

important role in metallurgical.
 The grain-boundary diffusion
is more immediate concern,
the GB area is many times
larger than the surface area.
 Polycrystalline samples:
combined effect of volume
and GB diffusion.

7
  GB component small the apparent diffusivity equals the
 At very high temp. diffusion through the lattice tends to
volume diffusivity.
overpower the GB component.
 GB component large the apparent diffusivity diverges
 At low temp. diffusion at the GB becomes more important.
from the lattice diffusivity.
 Ex. For grain size 35 microns polycrystalline silver
 GB diffusion is a function of temperature.
specimens: below 973K  GB component is important.
Above 973K  volume-diffusion component is important.

General implications:
 (a) diffusivities determined with polycrystalline specimens
are more liable to be representative of lattice diffusion if
they are measured at high temp.
 (b) the reliability of the data can be increased by
controlling the grain size of the specimens. The larger the
grain size, the smaller the GB contribution to the
diffusivity.
 For accurate measurements of lattice diffusivities using
T  the rate of diffusion through the lattice increases more rapidly th an polycrystalline specimens, high temp. and large-grained
specimens should be used.
the rate of diffusion along the boundaries .
T  the rate of diffusion along the boundaries decreases less rapidly.

Diffusion in Non-isomorphic alloy systems

If this gradient is missing at any position, the flux of A or B
atoms past this position w ill stop.
 for the B component :Δ GB  GB  GBo  RT ln a B
(GB : the partial molal free energy of B, GBo : the free energy
per mole of pure B, a B : the activity of B component)

•Continuous diffusion across the couple can only

occur if the partial-molar free energy decrease
continuously with distance.
•The only way that a gradient in the partial-molar
free energy can be obtained in an interval where
the partial-molar free energy wants to remain
constant is for the thickness of the interval to
 Single-phase fields appears with a finite width, but in vanish.
which the two-phase fields are represents by surfaces. •In the diffusion couple the two-phase regions
 The reason why the two-phase fields appear as surfaces have to appear as surfaces (thickness = 0). If this
in a diffusion couple: 1. For diffusion to occur, a did not occur, there could be no net diffusion flux
concentration gradient must exist across the couple. (or across the couple.
activity gradient)

* The thickness of a diffusion layer is determined by the

relative velocities with which its two boundaries move.
* These boundary movements are controlled by diffusion.
* The growth velocities of the two interfaces of the b phase
depend on the concentrations of the phases at the boundaries,
the diffusion coefficients and the concentration gradients.