HYDROGEN DIFFUSION

Lecture 7
 INTRODUCTION

The study of the interaction of hydrogen with

metals and alloys is widespread in the world.
Such interaction is considered in two directions.
On the one hand, in connection with the
development of hydrogen energy, it is necessary
to study new hydrogen storage materials. On the
other hand, hydrogen has a negative effect on the
crystal lattice of metals, thereby leading to
embrittlement of materials. Therefore, it is
necessary to develop various protective coatings,
make surface modifications, etc.
 DIFFUSION IN SOLIDS

The movement of atoms in solids is carried out

mainly by vacancies, that is, nodes of the crystal
lattice free of intrinsic atoms. If another atom of
the crystal lattice occupies a nearby free (vacant)
place, then the vacancy will move along the
crystal. For atomic diffusion in a solid, an uneven
distribution of nV vacancies in the solid is
necessary, i.e. vacancy density gradient −nV. In
this case, the flux of diffusing vacancies will be
nonzero if the atoms adjacent to the vacancy
have enough vibrational energy to jump from the
site of the crystal lattice to the vacancy.
 DIFFUSION MECHANISMS

Vacancy Interstitial
The position of the
The position of an
atom before diffusion
atom after diffusion

Terms of implementation: Terms of implementation:

• availability of vacancies • presence of foreign atoms of
small dimensions
 On that picture the movement
of atoms over vacancies in a
solid is presented. A job moves
in the opposite direction to
atomic diffusion.
In this case, the diffusion coefficient
is determined by the average
atomic lifetime в in the crystal
lattice site  and the squared atomic
displacement d2 during hopping,
where d is approximately equal to
the interatomic distance in the
diffusion direction
The movement
of atoms through d2
vacancies in a D= .
solid 6
Here 1/6 takes into account that the atom can move in six
independent directions. The lifetime of an atom in the node
of the crystal lattice is the longer, the greater is the energy
Ea necessary for its hopping to a neighboring vacant place.
The probability due to thermal fluctuations to acquire
energy Ea, according to the Boltzmann distribution, is
proportional to the equation:
 Ea 
exp  − 
 kT 
the frequency of atomic diffusion jumps in the lattice
exponentially increases with temperature
Diffusion coefficient also changes:

where D0 is a constant determined by the

properties of the substance

diffusion flow in solids obeys Fick's law:

the diffusion coefficient in solids is immeasurably less than

in gases. So, for gold it is equal to 10−35 m^2/sec, and for
oxygen in the atmosphere it is approximately 10−5
m^2/sec.
The diffusion equation is an equation that describes the
diffusion process of concentration equalization in a system
that is weakly nonequilibrium along the x axis - the one-
dimensional diffusion equation has the form:

in the three-dimensional case, the resulting equations must

take into account changes in the corresponding parameters
not only along the x axis, but also in the direction of the y
and z axes:

Where:
SOLUTION OF DIFFUSION EQUATIONS BY
THE FOURIER METHOD
One of the common methods for solving the diffusion
equation is the method of separation of variables - the
Fourier method.
Consider the equation of non-stationary diffusion in a flat
unbounded plate: n( x)  2 n( x )
= D(t )
x x 2
It is necessary to find a solution (for a finite solid) in which
the variables x and t are separated, i.e. function of the
form:
nm ( x, t ) = am X m ( x)Tm (t ).
We substitute the relation for nm(x,t) into the diffusion
equation
1 T m(t ) 2
1  X m ( x)
= = −  2

D(t )Tm t X m( x) x 2
m

Since Xm(x) is an even function due to the symmetry of the

problem Xm(-x) =Xm(+x), then
X m ( x) =  m cos  m x.
From the boundary conditions, we obtain the equation for
determining the eigenvalues χm
 md  md
tg =
d
 1N +  2 N1 .
2 2 2lD
Border condition:

 n( x )  N1 
= j H ( , t ) = − D n( , t ) 1 N +  2 N1  − 
d d
 D(t )
x x = d 2  2  
2
For a sample of thickness d~10-3-3 sm, we have the estimate
 md  md
tg =
d
 1N +  2 N1   10−8см 10-17 см21015 см-2  5 102 >> 1.
2 2 2lD 10 см
An approximate solution to this equation will be the
eigenvalues 2m + 1
m = ( )
d
These eigenvalues correspond to zero concentrations at the
plate boundaries. If the diffusion flux is irreversibly directed
to the surface of the plate with subsequent desorption of
hydrogen molecules into vacuum, then the reverse flow of
atoms from the surface of the evacuated plate into the
volume of the plate can be neglected: d
n ( , t ) = 0,
2
These conditions do not apply to the boundaries of the
sample, closer than to where, although the atomic flux is
directed only to the surface, the concentration is not zero
d 1 d − lD
n (  , t ) = n(  , t )  0.
2 2 2
Using the found eigenvalues, we obtain
  2m + 1  2 t 
Tm (t ) = Tm (0) exp −     D( )d 
  d  0 
The solution with given initial conditions has the form
 
 2m + 1    2m + 1  2 t 
n ( x, t ) =  am X m ( x)Tm (t ) =  am  cos

x   exp − 

   D( )d 
m=0 m=0 d   d  0 
The coefficients аm are determined by the initial
distribution of the concentration of hydrogen in the sample
n0(x) : +
d

2 2
2m + 1
am =
d dn0 ( x) cos d xdx
−
2

With a uniform initial distribution of hydrogen over the

thickness of the sample n(x,t=0)=n0 , -d/2 ≤x≤+d/2
4n0 (−1) m
am = ,
 (2m + 1)
4 (−1) m 

2m + 1    2m + 1  2 t 
n( x, t ) = n0   cos x   exp −     D( )d .
 m = 0 2m + 1 d    d  0 
The flow of hydrogen from each side of the plate is given by
Fick's law:
n( x) n( x) 4    2m + 1  2 t 
J H = − D(t ) = D(t ) = n0 D(t )  exp −     D( )d 
x x=−
d x x=+
d d m=0   d  0 
2 2
In the general case, using the equation for the eigenvalues,
 d
we obtain 4n
sec
2
m

am = 0
 ,
md 2lD lD2
1+ +  m2
d  1 N +  2 N1   1 N +  2 N1 2

  2t 
n( x, t ) =  am (cos  m x )  exp −  m  D( )d ,
m=0  0 
 2t 
exp −  m  D( )d 
n( x) 4 
 
J H =  D(t ) == n0 D(t )  0
.
x d d 2l l 2
x= m=0
1+ D
+ D
 m2
d  1 N +  2 N1   1 N +  2 N1 
2 2

For thin samples, at times determined by the inequality

t

   −2 2
D ( ) d  4,5  10 d
can be limited in the0 expansion by the first member of the
series (m = 0):  
 
2 t
4
JH = n0 D(t ) exp −  
d   d 
 D( )d 
0 
So for samples with a thickness of d<0.1 mm in materials
with a diffusion coefficient of the order of 10-5 sm2/sec, this
corresponds to t>45 s. Therefore, the choice of thin samples
(d <0.1 - 0.05 mm) is convenient for conducting
experiments on the study of hydrogen diffusion in metals
and quantifying the parameters of diffusion processes.
If the processes of adsorption and recombination of
hydrogen atoms on a surface into molecules do not limit
the diffusion release of hydrogen from a metal into a
vacuum, then the obtained relation allows us to determine
the relationship between the time of maximum thermogas
evolution in linear heating and the value of the activation
energy of hydrogen diffusion.