You are on page 1of 18

Journal of the Less-Common Metals, 49 (1976) 223 - 240 223

0 Elsevier Sequoia S.A., Lausanne - Printed in the Netherlands

ELECTROCHEMICAL METHODS FOR STUDYING DIFFUSION,


PERMEATION AND SOLUBILITY OF HYDROGEN IN METALS*

N. BOES and H. ZOCHNER

lnstitut fiir Physikalische Chemie der Universittit Miinster, Miinster (F.R.G.)

(Received December 8, 1975)

Summary

Electrochemical methods for the investigation of metal-hydrogen


systems are superior to other techniques, e.g., gas-volumetric, because of
their simple procedure and their flexibility towards variation of experimental
conditions. Moreover, these techniques allow measurements at very low
hydrogen equilibrium pressures which occur, for instance, in the V(a)-metal-
hydrogen systems in the concentration range of ideal dilute solutions at
normal temperatures.
The removal of surface layers, which retard the hydrogen passage from
the electrolyte into the metal and vice versa, is a necessary condition for
electrochemical methods to be applicable. Using ultra-high-vacuum tech-
niques, surface hindrances on base metals like V, Nb, Ta, Ti, and others,
can be eliminated. Thus, it is possible to study diffusion, permeation, and
solubility of hydrogen (isotopes), in these metals as well as in precious
metals such as palladium and its alloys.
Various electrochemical hydrogen-diffusion and permeation methods
are presented and compared. Their applicability and the limits of application
are demonstrated for pre-selected systems.
A recently developed electrochemical technique for measuring hydro-
gen (isotope) solubility in transition metals is described. By means of this
method, pressure-concentration equilibrium isotherms of V-, Nb-, and Ta-
H systems have been obtained recently, at ordinary temperatures, from
vanishingly low hydrogen concentrations, corresponding to extremely low
equilibrium pressures, up to atomic ratios H/Me of 0.5.

Introduction

Electrochemical methods, favoured by their simple procedures and


their flexibility with regard to experimental conditions, have recently gained

*Presented at the Meeting on “Hydrogen in Metals”, held at the University of


Birmingham, United Kingdom, on January 5 - 6, 1976, under the auspices of the Chemi-
cal Society (Faraday Division).
224

new and greater importance in the study of Me-H(D) systems. Their main
disadvantages so far, i.e., relatively poor reproducibility, and applicability
to a few metals only, could largely be removed by new techniques of sample
preparation [l] . In most cases, impedimenta to the interphase transfer at
the surface were the reason for poor reproducibility, and led sometimes to
gross misinterpretations of results obtained by electrochemical methods.
The new sample preparation technique, therefore, was developed with the
object of enabling and facilitating the interphase transfer of hydrogen from
the electrolyte solution into the metal, and vice versa.
When a metal sheet without any impediment of interphase hydrogen
transfer is immersed in a solution containing H- (or D-) ions, its electro-
chemical potential, as measured against a fixed reference electrode, can be
interpreted as the potential of a hydrogen electrode. At an “activated”
electrode coated with Pd or Pt-black [2, 31, the equilibrium:

is reproducibly established, and the electrode potential follows Nernst’s


equation: ’
RT a’
rl= (2)
r’“- 6
where a+ denotes the H(D) ion activity, and p, the hydrogen pressure.
If the metal dissolves hydrogen and the equilibrium:
H Me G== H’ + e- (3)
can be established, the H(D) concentration near the surface in the metal
determines the equilibrium hydrogen pressure (fugacity), p*, and thus the
electrochemical potential. This is the basis of the electrochemical techniques
employed here; a concentration change in the metal near the surface is
monitored by changes in the electrochemical potential of the surface, which
are measured against a reference electrode (saturated calomel electrode).
Beyond this, a constant penetration of hydrogen, j(mole H cmp2 s-l),
into the metal due to the electrolytical deposition can be measured by:
-j=i/FX q, (4)
i being the current (A), F the Faraday constant and q the sample surface
area. The time integral of the current yields the total amount of hydrogen
absorbed by the metal.
The equilibrium (3), of course, can only be established, and eqn. (4)
only be applied, if the metal surface is “open” for the penetration of hydro-
gen. This is true for Pd and for many of its alloys, but not for metals like V,
Nb, Ta or Ti, where corrosion layers, mostly oxides, prevent the equilibrium

*At small hydrogen concentrations Sieverts’ Law is valid:


n [mole H/mole Me] = G/K @a)
225

(3) from being established [4]. It is possible, however, to remove these im-
peding layers from the surface by annealing the foils of these metals under
ultra-high vacuum. In order to prevent the regeneration of such layers after
this treatment, the metal surface is covered with thin films of Pd evaporated
onto both sides of the foil, in the same ultra-high-vacuum system. With
samples prepared in this way, the equilibrium (3) is established very quickly
(similar to Pd) in a solution containing H(D) ions [l, 51.
In the following, a report on several techniques for the measurement
of diffusion and permeation will demonstrate the numerous possibilities
offered by electrochemical methods, and their usefulness wilI be emphasized
by a critical comparison of these methods. Furthermore, an electrochemical
technique for the measurement of H(D) solubility in base metals is presented
which permits measurements in pressure ranges inaccessible to gas-volumetric
methods [ 61.
Actually, electrochemical methods for obtaining solubility data have
been applied for a long time; an historical review is given by Lewis [‘7]. In
particular, Lewis et al. [8 - 111 and Flanagan et al. [ 121 have carried out
such measurements successfully. However, these methods have been restricted
only to metals with a high catalytic activity for equilibration with hydrogen
molecules such as Pd and its alloys.

Equipment for electrochemical measurements

Common to all techniques described here, is the following principle:


a membrane separates two solutions, thus it can be termed a bielectrode; at
one side of the specimen (entrance side) a change of hydrogen concentration
is established which, due to the mobility of hydrogen, penetrates through
the foil and can be observed and followed on the rear side (detection side)
by the time-variation of the potential. The time lag after which the concen-
tration change appears at the detection side is a measure of the diffusion
coefficient.
A measuring cell made of glass as shown in Fig. 1 is required for all
electrochemical methods described here. The specimen, of thickness s,
between the two halves of the cell, separates two electrolyte chambers from
one another. Each half is provided identically with a reference electrode
(calomel), a standard hydrogen electrode*, and a counter electrode made
from Pt wire; input and measuring sides of the specimen can therefore be
interchanged. One of the advantages of this symmetrical arrangement is the
possibility of studying isotope effects in the same specimen by using an H’
electrolyte on one side and a D’ electrolyte on the other, e.g., 1 N HzS04
and 1 N D2S04.

*The reference electrode can be calibrated against the standard hydrogen electrode
(p = 1 atm) containing the same electrolyte as the cell. Potential values reported here are
given with regard to this standard hydrogen electrode, i.e., 0 mV corresponds to a hydro-
gen pressure of 1 atm, and a 30 mV change at 30 “C corresponds to a pressure change of
a factor of ten.
226

thermostat

Fig. 1. Measuring cell.

By means of a double jacket, the cell can be thermostatted; measure-


ments have been carried out so far between -60 and 100 “C. A schematic
survey of the complete equipment used to study diffusion, permeation, and
solubility of H(D) by techniques developed so far is given in Fig. 2.
The potential difference between the sample and the reference electrode
is registered by a recorder, using a high-input impedance converter in order
to avoid current drains. To obtain the highest sensitivities, the zero point is
suppressed by connecting a potentiometer in series with the recorder. The
measured signal can be differentiated or integrated electronically.
The input side of the electrolyte cell can be connected to a current
source with timer as well as to a potentiostat, by which means a potential
difference between sample and reference electrode can be adjusted. The
constant current source can be used to create a current pulse (concentration
peak) or to establish a time-constant current.
The versatility of electrochemical techniques is based on the numerous
possibilities of varying, experimentally, the boundary conditions at both the
entrance and the detection sides of the sample. The entrance side, for instance,
can be polarized cathodically (hydrogen precipitation) by a constant current,
or by a short current pulse. At the detection side, on the other hand, the
time variation of the hydrogen potential can be followed directly, or it can
be followed by the time variation of a compensating (anodic) current that
is necessary to keep the potential constant. Instead of applying a constant
current at the input side one may adjust a constant potential, or even work
with periodically-changing potentials. There are many more possible com-
binations.
227

Fig. 2. Electrical circuit.

Techniques for measurement of diffusion

The time-dependent equalization of hydrogen concentration in the


metal is described by Fick’s second Law:
ac a2c *
-=
at Da,cz * (5)

In order to evaluate the results obtained electrochemically, one has to find


solutions c(x, t) of eqn. (5) which satisfy the boundary conditions established
by experiment.
For certain techniques it is necessary to know the variation of the
anodic current, i( t)x =s. This is obtained from Fick’s first Law by differen-
tiation of the concentration c,(t):

i(t),=, =-D qF (6)

Technique 1 (non-steady-state galvanostatic time-lag method)


In this technique, a constant hydrogen flux, starting at t = 0, is esta-
blished at the entrance side by a constant cathodic current, whilst at the

*The concentration range covered by one individual measurement is always small


enough to permit the use of a constant diffusion coefficient.
228

Fig. 3. Principle of the non-steady-state galvanostatic time-lag method.

detection side at t > 0 the anodic current, which is required to maintain


c = 0 at the surface, is recorded. In Fig. 3, the concentration profiles at
successive times, and the typical variation of the anodic current, are shown.
Assuming the initial and boundary conditions to be:
t=O: c=O for O<x<s

jo=oc"
= const.
t> 0: s
1 CT9= 0
the solution of the problem is, [13] :

c(x , q
Jo(s-x) -- 8jos m t-1)” 2n +1
sin ~ n(s - x)
D Dn2 5 (2n + l)2 2s

(7)

Combining eqns. (6) and (7) one obtains


4 (-1)” (2n + 1)2n2Dt
i,(t) = i, 1 - -77 E0 -2n + 1 exp 1).
(8)
4s2
This corresponds to an s-shaped curve for i,(t) as schematically indicated in
Fig. 3.
The time, ti, of the inflection point becomes.
3 In 3 s2 *
ti=--- - (9)
2 n2D ’
The intersection of the tangent at the inflection point with the initial level,
i, = 0, yields the break-through time, tb
S2
t,, = 0.76 - (10)
.rr2D ’

*In calculating the inflection point time, ti, and the break-through time tb, only the
first two terms of the series need to be considered. The error made by neglecting higher
terms was shown to be less than one percent by a more exact computer calculation.
229

Fig. 4. Principle of the step-method.

by which the diffusion coefficient can be calculated.


Following the time integral of current, i.e., the total quantity of hydro-
gen emerging from the detection side, the slope of the curve becomes cons-
tant when a stationary linear concentration gradient has been established in
the foil (i, = const), as shown in Fig. 3. The intercept on the t-axis of the
extrapolation of the straight line gives the so-called time lag, tL.
1 ,2

tL==, (11)

wherefrom the diffusion coefficient can be obtained.

Technique 2 (step-method) [ 14, 151


Starting with a homogeneous hydrogen content in the sample, the
hydrogen concentration is raised by a step, AC, at time t = 0, at the entrance
side of the foil and is kept constant thereafter (potentiostatic connection).
At the detection side, the time-dependent change of concentration, i.e., of
potential, is followed by a compensating recorder (Fig. 4). There is no flow of
anodic current in this case, and, because of the large capacity of the metal
for dissolved hydrogen as compared with that of the electrolyte, the hydro-
gen does not escape out of the.foil in substantial amounts. The boundary
condition (&z/ax), =s.O = 0 is fulfilled therefore to a good approximation.
The initial and boundary conditions are:
t = 0: c = cl for 0< x < s

The change of concentration satisfying these conditions may be described


by [16] :
4 1 (2n + 1)nx
c(x, t)- cl = AC 1 - - c” ~ sin
71 0 2n+l 2s

(2n+ 1)27722t
exp -
( 4ss -) 1)
(12)

and at x = s by:
230

10-6 10e3 atm 1


P-

Fig. 5. Plots of diffusion coefficients of H in Pd/Ag SO/20 and 60140 us. electrochemical
potential (f logarithm of Hz pressure) at 30 “C.

4 (-1)”
c,(t)-cl =Ac l--c”----- (13)
77 02n+l
The expression in parentheses at the right hand side of eqn. (13) is
equal to that in eqn. (8). This means that the time dependence of c,(t) - cl
is the same as that for the current i,(t) in the technique described first. Thus,
the experimental values ti and t, are related to the diffusion coefficient by
eqns. (9) and (lo), respectively.
This method is appropriate for measuring the concentration dependence
of the diffusion coefficient by raising the potential (concentration) stepwise.
That means that the final concentration, after an individual measurement,
is used as a starting value for a new run. Applying small potential steps, the
diffusion coefficient stays constant at each individual measurement. Values
of diffusion coefficients of H in Pd/Ag 80/20 and 60/40, obtained in this
way, are shown in Fig. 5 as a function of potential and hydrogen pressure,
starting from very small pressures and going up to one atmosphere [ 151.
By making use of known p-n-isotherms, the relationship between diffusion
coefficient and hydrogen content can easily be obtained.

Technique 3 (non-steady-state potentiostatic time-lag method) [17]


In this method, first developed by Devanathan and Stachurski [17],
the boundary conditions of Technique 2 at the entrance side are combined
with those of Technique 1 at the detection side. The concentration is ini-
tially uniform throughout the membrane. At time, t = 0, a hydrogen con-
centration, cl, is established at the entrance side and kept constant there-
after; at the detection side the current and/or the time integral of current
are registered while c, = 0 is maintained (Fig. 6).
231

Fig. 6. F’rinciple of the non-steady-state potentiostatic time-lag method.

cs

L
Fig. 7. Principle of the pulse-method.
/
'b

ti t

A solution satisfying the boundary conditions:


t=O: c=O for 0GxG.s
t>O: ce=cr and c,=O
is [16] :
x 2 -c
c(x, t) = Cl - Cl -+-zlsin (14)
S 711 n

and according to eqn. (6):

i,(t) = i, 1 + 2 7 (-1)” exp (- Dn;;zt)). (15)


i
The relationship for ti is obtained finally as:
In 16 s2
ti=----
(16)
3 n2D ’
and for tb:
s2
tb = 0.5 - (17)*
n2D ’
The time-lag, tL, defined analogously to Technique 1, is given by [18,19] :

(18)
The diffusion coefficient can be determined by eqn. (17) as well as by eqn.
(18).
*The relationship given in ref. 17 is incorrect.
232

Fig. 8. Typical recorder chart using the pulse-method. The time-scale must be multiplied
by 1.2 s.

Technique 4 (pulse-method) [5,15, 201


The pulse-method differs from the step technique in the way of
charging the foil with hydrogen. Starting again with a uniform concentra-
tion, the foil is subjected to a short cathodic current pulse by which a hydro-
gen peak is forced into the sample (cross-hatched area in Fig. 7). This ini-
tially small peak smooths out by diffusion through the sample, and the con-
centration at the detection side increases as soon as the first amount of
hydrogen occurs there. The curve of concentration with time is followed by
recording the potential.
If the initial and boundary conditions are:
t=‘O: cg=f(x’)=g(x’)+c,

t>o:
(a”1
ax x=o,s
=o

the solution of eqn. (5) is [16] :


nnx’
C(X, t) =tJf(x’)dx’ + F cos mr Fexp cos - dx’.
s

From this, the curve of concentration with time at x = s is obtained as:


n2n2Dt
c,(t)-ccl=Ac 1+2? (--1)“exp
S2 )I-

The expression in parentheses is identical with that in eqn. (15). Thus, the
diffusion coefficients must be calculated from the measured values of t,, by
means of eqn. (17).
233

TABLE 1

Diffusion coefficients of H and D in various metals at 25 “C


using the pulse technique (n + 0)

Metal 2 -1
ref. D(H)(cm s ) D(D)(cm
2 -1
s )

Pd 29 6.3 x lo-’ 6.8 x 1o-7


Pd/Ag, 90/10 29,5 5.8 x lo-’ 3.4 x 1o-7 *
Pd/Ag, 80120 29 4.6 x 1O--7 5.2 x 1O-7
Pd/Ag, 70130 29,5 2.0 x 1o-7 1.8 x 1O-7 *
Pd/Ag, 60140 15,5 2.3 x 1O-8 * 4.7 x 1o-8 *
Pd/Ag, 50150 15,5 1.9 x 1o-g * 3.7 x 1o-g *
Pd/Ag, 40160 15,5 3.5 x lo-lo * 4.3 x lo-lo *
V 1,5 3.7 x 1o-5 1.9 x 1o-5
Nb 1,5 5.2 x 1O-6 4.5 x 1o-6
Ta 1,5 2.0 x lo+ 1.0 x lo+
Ti 28 2.3 x 1o-7 -

*Samples not annealed.

In Fig. 8, a copy of a recorder chart from diffusion of H and D through


a V sample of 700 hrn thickness demonstrates a typical run of concentration
(potential) and of the differentiated signal at the detection side. From the
values obtained for the break-through time t,,(H) = 7.6 s and t,,(D) = 18.6 s,
the diffusion coefficients D(H) = 3.3 X lop5 cm2 s-l and D(D) = 1.3 X lop5
cm2 s-l can be calculated.
Values for diffusion coefficients, which so far have been obtained for
various metals using the pulse technique, are listed in Table 1.

Technique 5 (enforced oscillation method) (Fig. 9)*


This technique represents an interesting variant of the step method.
The entrance side of the foil is connected to a potentiostat supplied, in this
case, with a sine wave generator. This arrangement allows the concentration
at the entrance side to oscillate around a preselected value, c, with an ampli-
tude, A, and a frequency, w . After a few periods, the concentration at the
detection side builds up stationary oscillations also. However, the amplitudes
are damped and the phase is shifted with reference to the primary oscilla-
tions at the entrance side. Both quantities, amplitude damping as well as
phase shift, can be used to obtain the diffusion coefficient.
In this case the boundary conditions are:
c(x=s, t)-Z=Ae’“’ (x = s means the entrance side)

=0 (x = 0 means the detection side).

*This technique has been developed by U. Westerboer using a gas-volumetric


method [ 211.
234

Fig. 9. Principle of the oscillation methods.

-
The stationary oscillations in the membrane can be described by:

cash F(1 + i)
c(x, t) --c=A
cash $ (1 + i)
e iwt, with 1 =
V20
-
0
. (21)

From this equation, the ratio of the amplitudes at x = s and x = 0 is obtained


as:
4x=0 2 S
-=
(22)
Ax=, cos 20 + cash 2a withO=I*

The phase shift results in :


AT
-a=--- 271 = arc tan (tanh o-tan u). (23)
7
If u 9 1, then u = -@; thus the simple relationship:
20 (-@)2
0= (24)
S2

is received, or, after introduction of the period r = 271/w:


S2
r= (25)
471(AT/~)~ D -
In Fig. 10, the phase shift is plotted uersus the square root of frequency.
The experimental points obtained by measurements on a Pd/Ag 70/30 alloy
(s = 100 pm) at 22 “C show sufficient agreement, with a straight line passing
through the origin according to eqn. (24). The diffusion coefficient has
been calculated from the slope to be D = 0.75 X lo-’ cm2 s-l.

Technique 6 (self-excited oscillation method) [22]


In this technique, the potentiostat acts in an unusual way: it is con-
nected, firstly with the counter electrode at the entrance side, but, contrary
to the step method, secondly with the reference electrode at the detec-
tion side. If the potential difference between the sample and the reference
electrode at the detection side differs from the preselected value adjusted
by the potentiostat, a current will flow at the entrance side, in order to
equalize the voltage differences. But the hydrogen must first pass through
235

.7 -
I .6 - s=100/um 70130
Pd/Ag . l /

a=22 oc
.5-
@” ml” -112 /
.L-

.3- ./

n -H- 2P

.I .2 .3

Fig. 10. Plots of frequency us. phase shift.

PdlAg 80120; SE .115mm i a=30 OC

Fig. 11. Typical recorder chart using the self-excited oscillation method.

the foil, because the membrane is interposed into the potentiostat circuit.
The signal to the potentiostat therefore is delayed by the time-lag of diffu-
sion through the foil. Moreover, this time-lag causes the potential to over-
shoot at the entrance side. Feedback thus gives rise to stationary oscillations
of the potential at the detection side.
In order to derive a simple relationship between the period of the oscil-
lation and the diffusion coefficient, a further boundary condition (addi-
tional to those of Technique 5) that describes the mode of action of the
potentiostat, has to be considered:

c(x = 0, t) - c = -a
(2ax1 * X=S

The negative sign corresponds to the phase shift of 7~between the potential
at the detection side and the current at the entrance side. This additional
boundary condition leads to the result:
tanh~+tan~=O. (26)
236

Taking into account only the solution for the oscillation with the smallest
damping, one obtains:

7r=O.,,$ (27)

As 71 is greater by a factor of ten compared with the break-through time,


tb, of the pulse-technique, this oscillation method is well adapted for mea-
suring higher diffusion coefficients.
A copy of a recorder chart (Fig. 11) shows the typical course of con-
centration with time at the detection side of a Pd/Ag 80/20 sample of 115
pm thickness. The period 71 = 400 s in this case yields a diffusion coefficient
D = 1.85 X lo-’ cm2 s-l [22].

Discussion of diffusion measurement techniques

If single-layer specimens are used which have ideal permeability for


H(D), all the techniques which have been discussed are generally well appli-
cable. That means, the boundary conditions considered for the mathemati-
cal solution can easily be established experimentally, or, in other words,
the relaxation time necessary to reach the desired boundary condition is
short compared with the measuring time, tb or tL.
At the surfaces of almost all metals, however, impeding layers have
to be considered which more or less retard the permeation of hydrogen.
The single layer problem becomes then - in the simplest case - a triple
layer problem: impeding layer-pure metal-impeding layer.
The impeding layers increase the relaxation time necessary to establish
the boundary conditions for non-steady-state techniques like the potentio-
static and the galvanostatic time-lag method (Techniques 1 and 3) or the
step method (Technique 2). In such a case, the relaxation time can easily
reach the values of the measured time intervals, t, or tL, or even exceed
these appreciably. The metal then is subjected to. boundary conditions that
differ from those of the mathematical solution which gives the relationship
between the diffusion coefficient and the measured time interval. In most
cases, neither the thickness of the impeding layers, nor the diffusivity, or
the solubility of hydrogen in these layers are known, moreover, the existence
of the layers is ignored completely; hence, the measured values of tb, and
in particular, of tL, yield more or less incorrect values for the diffusion
coefficient.
A smaller diffusivity of hydrogen in the impeding layer as compared
with the pure metal will always be perceptible as an additional diffusion
resistance, and thus will increase the measured values of tb and tL*. If, also,
the solubility is smaller, the layer will even impede permeation, the term
“impeding permeation” includes “impeding diffusion”.

*Here, and in the following text it is always presupposed that the thickness of the
impeding layer is included in the total thickness, s, of the specimen.
231

The break-through time, t,, roughly speaking, is the time need& by


the first hydrogen particles to arrive at the detection side of the sample,
after the boundary conditions at the input side have been changed suddenly.
This time interval, therefore, is essentially independent of the amount of
hydrogen flow through the sample, i.e., the quantity, tb, is mainly deter-
mined by the diffusion resistance. The time-lag, ti,, however, is essentially
the time required to obtain a steady-state flow through the sample after
the sudden change of boundary conditions. The amount of hydrogen neces-
sary to establish the stationary concentration profile has to pass through
the impeding layer with low permeability. This retards the adjustment of
the steady-state gradients, and may increase the tl-time substantially.
A foil of 100 pm Ta, covered on one side by a 100 ,& film of Pd, can
be imagined as an idealized model of a two-layer sample that exhibits such
behavior. The Pd film acts as an impeding layer, because of its lower hydro-
gen solubility (by a factor 10M4 as compared with Ta). If Barrer’s relation-
ship for time-lag in multilayer samples [23] is applied to this example, a
t, value results that is larger by a factor of 2.7 than that of the pure Ta spe-
cimen. The diffusion resistance of the Pd layer, however, is so small that it
does not show up in the break-through time. In real cases, both diffusivities
and solubilities of such layers (for instance, oxide layers for V(a) metals),
are much smaller than this ideal “impeding Pd-layer”; thus, the permeation
resistance of these layers may even exceed the resistance of the pure metallic
layer in the sample to be studied. In such a case the time-lag, tL, is largely
determined by the impeding layer, and is not qualified for obtaining diffu-
sion coefficients of the pure metal.
A number of published D values of H(D) in metals, in particular for
V(a)-metals and Ti, have been derived from such tL measurements [ 24 - 261.
Since the influence of impeding layers which can change tL appreciably
despite their small thickness was not considered, the resulting diffusion
coefficients are likely to be erroneous. The only values that can be obtained
by such measurements are the diffusion coefficients in the impeding layers,
when their thickness is introduced instead of that of the pure metal. On the
grounds mentioned above, measurements of the quantity t,, are generally
to be preferred over the quantity, t,.

Electrochemical measurement of permeation

In Technique 3 for diffusion measurements, the time-lag, tr,, is the time


required to reach the steady-state concentration-gradient in the sample. As
soon as the steady state is established, the cathodic current at the input side
equals the anodic current at the detection side (Fig. 6).
If the concentration at the input side is changed to a new constant
level, the concentration at the detection side remaining zero, a new steady
state, i.e., a new constant current, i, , is attained after a certain time. If the
concentration is low enough to permit application of Sieve&’ Law (eqn. (3a)),
a plot of i, us. <p according to:
238

i, =PqF*
s ’
resulting from eqns. (3a) and (6), yields the permeation coefficient, P. This
is related to the diffusion coefficient, D, to Sieverts’ constant, K, and to the
molar volume, V, of the metal by:
D
p=-
KV’
These permeation measurements are influenced by impeding layers in a
similar way as tL measurements. The drop of chemical potential in the im-
peding layer reduces the permeation flow through the specimen, as compared
with the case in which the pure metal is subjected to the same difference of
chemical potential, i.e., of hydrogen pressure. If the existence of the im-
peding layers is ignored and the permeability is attributed to the pure metal,
erroneous permeation coefficients are obtained. If almost the total drop of
the chemical potential occurs in the impeding layers, no conclusions can be
drawn regarding the permeation coefficient of the pure metal.

Solubility measurements [6]

So far as the diffusion coefficients are known, measurements of per-


meation can yield Sieverts’ constant, the inverse of the solubility (eqn. (29)).
This indirect method is applicable for small concentrations, but it suffers
from the strong influence of impeding layers on permeation measurements.
The principle of the direct electrochemical technique to measure solu-
bilities, first being mentioned by Frumkin et al. [27], has been applied to
base metals (V(a)metals) [6]. For this method, the same apparatus is used
as for the measurement of diffusion and permeation. In this case, a small
cathodic current is adjusted at the entrance side whilst at the rear side a
potentiometer-type voltage recorder follows, without current, the variation
of the potential with time, whilst the sample fills up with hydrogen from
the input side. When the current is kept constant, the hydrogen concentra-
tion- in the sample increases linearly with time (Faraday’s Law, eqn. (4)),
whereas the potential, according to Nernst’s Law, eqn. (2), increases propor-
tionally to the logarithm of the equilibrium pressure. Thus a log G/n iso-
therm is recorded which can easily be converted to ap/n isotherm.
Conditions necessary for application of this method are (l), all the
hydrogen precipitated cathodically must enter the sample and (2), the con-
centration must be constant in space within the sample at any time, i.e., the
sample must be filled up uniformly over the total thickness.
The first condition can be satisfied by applying sufficiently small
current densities, for instance, i/q S 10 mA cmv2 in the case of properly
prepared foils of V(a)metals. The second condition limits the maximum
thickness of the sample for a given value of diffusion coefficient; a typical
239

1o-2

lc?

1o-L

10-5

lo+

lo7

10-a

II .?L .06 .qe .l - n[H/Nb


.l .2 .3 .L .?I
- nlH&Tal

Fig. 12. p-n isotherms of hydrogen in V(a)metals.

case is: s d 100 pm for D = low6 cm2 s-l. Samples of V, Nb, and Ta have
been investigated successfully with this method after they were “permeation
activated” as described above. Figure 12 shows isotherms obtained in this
way for the V(a)-metals.
A great advantage of the electrochemical method for measuring solu-
bilities is the possibility of measuring, at normal temperatures, directly from
very small concentrations in the range of ideal dilution up to atmospheric
pressure. In particular, in V(a)-metals, the range of ideal dilution can only
be studied by these methods since the equilibrium pressures in question,
X 10M1’ Torr, are too low to be accessible to gas-volumetric methods.

Acknowledgement

The authors thank Prof. E. Wicke for numerous suggestions and dis-
cussions.

References

1 N. Bees and H. Ziichner, Phys. Status Solidi A, 17 (1973) K 111.


2 A. Kiissner and E. Wicke, Z. Phys. Chem., (N.F.), 24 (1960) 152.
240

3 A. Kiissner, Z. Elektrochem. Ber. Bunsenges. Phys. Chem., 66 (1962) 675.


4 H. Ziichner and N. Boes, Z. Phys. Chem., (N.F.), 93 (1974) 65.
5 N. Boes, Dissertation, Miinster, 1974.
6 N. Boes and H. Ziichner, Ber. Bunsenges. Phys. Chem., 80 (1976) 22.
7 F. A. Lewis, The Palladium-Hydrogen System, Academic Press, New York, 1967.
8 F. A. Lewis, Platinum Met. Rev., 4 (1960) 132.
9 F. A. Lewis, Platinum Met. Rev., 5 (1961) 21.
10 F. A. Lewis, Platinum Met. Rev., 6 (1962) 22.
11 J. C. Barton and F. A. Lewis, Talanta, 10 (1963) 237.
12 J. W. Simons and T. B. Flanagan, J. Phys. Chem., 69 (1965) 3581.
13 H. S. Carslaw and J. C. Jaeger, Conduction of Heat in Solids, Oxford Univ. Press,
London, 1959.
14 A. Kiissner, Z. Naturforsch., Teil A, 21 (1966) 515.
15 H. Ziichner, Dissertation, Miinster, 1969; H. Ziichner’and N. Boes, Ber. Bunsenges.
Phys. Chem., 76 (1972) 783.
16 J. Crank, The Mathematics of Diffusion, Clarendon Press, London, 1975.
17 M. A. V. Devanathan and Z. Stachurski, Proc. R. Sot. London, Ser. A, 270 (1962)
90.
18 H. Daynes, Proc. R. Sot. London, Ser. A, 97 (1920) 286.
19 R. M. Barrer, Trans. Faraday Sot., 35 (1939) 628; 36 (1940) 1235.
20 H. Ziichner, Z. Naturforsch., Teil A, 25 (1970) 1490.
21 U. Westerboer, Dissertation, Miinster, 1976.
22 N. Bees, U. Westerboer and H. Ziichner, Ber. Bunsenges. Phys. Chem., 77 (1973)
708.
23 R. Ash, R. M. Barrer and D. G. Palmer, Brit. J. Appl. Phys., 16 (1965) 873.
24 A. C. Makrides, M. Wright and R. McNeill, Final Report on Contract DA-49-186-
AMC-136(D); Harry Diamond Laboratories, 1965.
25 G. Holleck, J. Phys. Chem., 74 (1970) 1957.
26 E. Brauer and E. Nann, Werkst. Korros., 25 (1974) 409.
27 A. N. Frumkin and N. A. Aladzhalova, Acta Physicochim. URSS, 19 (1944) 1.
28 E. Brauer, R. D&r and H. Ziichner, Z. Phys. Chem., (N.F.), (1976) to be published.
29 N. Boes and H. Zi.ichner, to be published.

You might also like