Journal of Nuclear Materials 152 (1988) 35-40 35

North-Holland, Amsterdam

DEUTERIUM PERMEABILITY OF THE AUSTENITIC STAINLESS STEEL AISI 316

Othmar GAUTSCH and Guenter HODDAP

Materials Science Division, Joint Research Centre, Ispra Establishment, I-21027 Ispra, Italy

Received 5 August 1987; accepted 27 October 1987

The permeability in membranes of stainless steel AISI 316 was determined at different membrane temperatures and
upstream deuterium driving pressures. The results are in good agreement with others reported in the literature. Measurements
were also performed with membranes having an upstream surface covered with a layer of the liquid Pb-17Li alloy, to
represent the Pb-17Li/steel interface of a cooling tube in the blanket of a fusion reactor. In this case, the values for the
permeability decreased by a factor of 1 to 4 with respect to the absence of Pb-17Li, and also the activation energy of the
permeation decreased from 52000 to 34000 J/mol. This means that the solubility of deuterium in Pb-17Li is lower than in
AISI 316 and nearly independent of temperature, acting therefore as a permeation barrier for deuterium. Also the values of
the diffusivities have decreased by a factor of about 2 with respect to those obtained in absence of Pb-17Li. Since the
diffusivities have been derived from the same type of deuterium flux measurements than the permeabilities, it was concluded
that the Pb-17Li had acted as a diffusion barrier.

1. Introduction With:

The purpose of this work was to determine the F= F(H)(1/2)0’5, (2)

permeabilities of deuterium through membranes of AISI
one obtains:
316. The upstream surface of the membranes was
covered with a layer of the liquid eutectic alloy Pb-17Li F = 1.65. 1O-6
as shown in detail in the experimental section. Thus x exp( - 65800/RT) m3 STP/m s bar0.5. (3)
permeation of deuterium took place from an upstream
chamber filled with D, (g) through the system Pb-17Li
(l)/SS316 (s) into an evacuated downstream chamber.
This system was chosen to represent the tritium permea- 2.Experimental
tion from a Pb-17Li blanket of a fusion reactor through
the wall of the cooling tubes into the coolant. The In order to check on the performance of the experi-
experimental temperatures were between 518 and 665 K mental method and apparatus, some values for the
and the upstream deuterium pressures between 1 x 1O-3 permeabilities have first been determined in absence of
and 16 X 10e3 bar. Pb-17Li. Also the diffusivities of deuterium in AISI
From the deuterium permeabilities, F, those of tri- 316 have been obtained simultaneously by the same
tium, F(T), can be estimated to a good degree by measuring technique.
F(T)= F(2/3)0.5 [l] and the total amount of tritium The experimental method and apparatus were simi-
which permeates into the coolant can be calculated. lar to those described by Van Deventer 121. A schematic
The hydrogen permeation through austenitic SS is design is shown in fig. 1, where the permeation cell with
not very sensitive to the chemical composition. There- the cylinder shaped membrane is indicated by the num-
fore, in absence of Pb-17Li, the permeabilities can be ber 17. Details of the membrane are shown in fig. 2.
calculated with an accuracy of a factor of 1.5 either way The cylinder shaped membrane is welded to the
by [l], R = 8.314 J/mol . K: permeation cell at about half-height. The diameter of
F(H) = 2.33.1O-‘j the membrane is about 1.9 X 10m2 m and the length
4 X 10e2 m. The cylinder (i.e. membrane) is filled com-
xexp( -65800/RT) m3 STP/m s bar’.‘.
pletely with liquid Pb-17Li during the experiments, and
(1) not only the bottom of the cylinder but also the inner

0022-3115/88/$03.50 0 Elsevier Science Publishers B.V.

(North-Holland Physics Publishing Division)
36 0. Gautsch, G. Hoddap / Deuterium permeability ofstainlesssteel AISI 316

Fig. 1. Schematic design of the UHV installation.

walls are in contact (covered) with liquid Pb-17Li. tion feed through (20), and a palladium purifier for the
Permeation of deuterium takes place both through the deuterium (6) as well as some servo-controlled leak
bottom and the walls. The wall thickness of the mem- valves for maintaining a constant deuterium driving
brane was L = 0.05 X 10e2 m and the surface area pressure in the upstream compartment (10, 11).
22 X lop4 m2. The experimental installation shown in The up- and downstream compartments of the per-
fig. 1 consists of a deuterium source (l), auxiliary meation cell (17) were pumped on by separate pumping
vacuum pumps (2), manometer (3), liquid nitrogen trap units (16) each consisting of one rotary and two
(4), UHV-valves (5, 8, 13) deuterium storage tank (7), turbomolecular pumps. In some cases also a titanium-
Pirani gauges (9) UHV-gauges (12), window (19) mo- getter pump (14) was used. The deuterium permeation

248

Fig. 2. Permeation cell including the cylinder-shaped membrane.

 0. Gautsch, G. Hoddap / Deuterium permeability of stainless steel AISI 316 31

fluxes in the downstream compartment were determined

by means of quadrupol mass spectrometers (15) and
certified calibrated standard deuterium leaks (21).
The temperature of the membrane was kept constant
within f 3O C and measured with calibrated thermocou-
ples (18) which were introduced, together with an inter-
nal resistance heater (18) into the liquid Pb-17Li. A
resistance furnace was located externally around the
permeation cell (17).
The chemical composition of the membrane
materials, in weight S, was: 0.015 C, 0.56 Si, 1.71 Mn,
0.013 P, 0.006 S, 17.76 Cr, 11.82 Ni. That of the
Pb-17Li was: 99.31 Pb, 0.68 Li, 1.5 X lop3 Fe, 1.8 X
10m3 C, 1 x 10e4 N. During the heating up of the
membrane the temperature-time recording indicated
that the Pb-17Li melted at 509 K, only 1 K above the
reported melting temperature. It is therefore assumed
that the alloy did not contain significant oxide impuri-
ties.
Before the beginning of the experiment the mem-
brane was degreased and subject to ultrasonic cleaning.
Then the required amount of solid Pb-17Li which was
kept in a glass container under argon, was transferred
into the cylindrical membrane which, during the trans-
fer, was kept under a plastic foil in presence of a strong
argon flush.
Next the whole UHV-installation was evacuated to
better than lo-” bar and then outgassed for 5 d at 653 Fig. 3. Schematic design of the motion transducer system for
K, while the membrane was baked at 773 K. At this the determination of the absolute upstream pressure.
temperature the vapour pressure of Pb-17Li is 2 X 10e8
bar and the partial pressure of lithium is 4 X lo-” bar
[6]. It appears on the basis of qualitative compatibility motion transducer (nr. l-10) which in turn produces a
studies [9] that at 773 K interdiffusion, chemical and/or proportional voltage signal. By compensating this signal
physical reactions at the Pb-17Li/SS 316 interface to zero with an outside pressure (nr. 17) which activates
could concern only such a small depth of the membrane an absolute pressure gauge (nr. 15), the corresponding
that the dissolution of deuterium in the membrane absolute driving pressure could be read from the certi-
might possibly be affected but not the subsequent bulk fied manometer with an accuracy of + 0.3% of full scale
diffusion. After outgassing there was a vacuum of 1 X (90 mbar). The resolution was below 0.1 mbar.
lOpi2 bar at room temperature. The background vacuum When the driving pressure and the membrane tem-
in the downstream compartment during the experiments perature became stable, the partial ion currents of the
was always below 2 x 10-i* bar and the residual gas masses 2 (Z,+H,), 3 (Z,+HD), and 4 (Z,+D,)
was composed of H,O, N,, CO, and H,. were recorded in the downstream compartment with a
Measurements of the permeation flux Q. always quadrupol as a function of time. The total ion current
started by allowing the membrane to reach the pre- which corresponds to the deuterium permeation flux
selected temperature. Then the chosen deuterium driv- across the membrane is given by [3]:
ing pressure in the upstream compartment was estab-
z = z4 + 0.51, + 0.71(0.51, + I, - Z,), (4)
lished within 10 min above the membrane (or liquid
Pb-17Li). The absolute value of this pressure was de- where I, is the background ion current of mass 2. The
termined by means of a compensation method. The ratio of Q/Z is equal to the ratio of the deuterium flux
elongations of a metallic bellow (number 12 in fig. 3, from a certified calibrated standard leak Q, = 8.4 x
bakeable at 723 K) are proportional to the established lo-‘* (510%) m3 STP/s, to the corresponding mea-
driving pressure. These elongations are taken up by a sured ion current Z,, and Q = Q,Z/Z,.
38 0. Gauisch, G. Hoddap / Deuterium permeability of stainless steel AISI 316

After the first experiment the membrane was out- in SS 316 [6]. The solubilities (S) range from 0.17-0.25
gassed until its deuterium concentration decreased to a m3 STP/m3 bar0-5 and the diffusivities were between
value near the detection limit. Then higher values of 3.1 X lo-‘* and 60 x lo-l2 m*/s. The calculated per-
temperatures and/or pressures were established, and meabilities, F = DS, amounted to 0.52 x lo-‘*-15 x
this procedure was repeated, until the experiment at the lo-‘* m3 STP/m s bar0-5. All these values agree fairly
highest selected temperature and pressure was com- well with those in table 1.
pleted. The results obtained in presence of Pb-17Li are
At steady-state (d I/d t = 0) the measured permea- shown in table 2. The standard deviation on D corre-
tion flux Q, is directly related to the permeability sponds to 15-22%, that for F is below *20%;.
F = Q,L/AP~.‘, where P, is the upstream driving pres- Table 3 contains a comparison of the results, ob-
sure and A the surface area of the membrane. From the tained from the respective Arrhenius plots, in absence
transient rise time period (dZ/dt > 0) and decline time and presence of Pb-17Li. It can be seen that in pres-
period (dZ/dt < 0 for P, = 0) the diffusivity D of de- ence of Pb-1Li:
uterium in SS 316 was calculated by means of Q, Q, (1) the value of F does not vary with temperature as in
and the appropriate relations [4,5]. The solubility of absence of Pb-17Li;
deuterium in SS 316 was estimated from S = F/D. (2) the values of F and D are lower by a factor of
about 1-4 for F, and about 2 for D;
(3) the value of E(F) = 34000 J/mol is lower, than in
3. Results and discussion absence of Pb-17Li where E(F) = 52000 J/mol/
(4) the value of E(D) = 42000 J/mol has remained
About 30 single determination of F and D have essentially that in absence, i.e. E(D) = 39500
been made at each given membrane temperature and J/mol;
deuterium driving pressure. The standard deviations of (5) the value of the solubility of deuterium calculated
F, in absence of Pb-17Li, were below +lO%, those of from F and D remains nearly constant allowing for
D correspond to 16-268. There is, in the case of D, a an error of about f30%, whereas in absence it
satisfactory agreement with the values of ref. [3], as increases with increasing temperature as expected.
shown in table 1. The same holds for F and ref. [l] as Regarding point (l), the following explanation can
well as for S and ref. [7]. From the relation log F - l/T be offered: the first step in the permeation through the
an activation energy of permeation of E(F) = 52000 membrane is the dissolution of deuterium in the
J/mol was calculated. The activation energy for the Pb-17Li. Its solubility 3, is reported to be very slightly
diffusion results to be E(D) = 39500 J/mol, which [6] or even non dependent on temperature [8]. There-
agrees quite well with the 428000 J/mol reported [3]. fore, the steady-state concentration in Pb-17Li C, =
Thus the activation energy for the dissolution of deu- sPi.5 will have the same temperature dependence. This
terium in SS 316 would be E(S) = E(F) - E(D) = concentration is assumed to be homogeneously distrib-
12500 J/mol, about 2 times higher than the reported uted in the Pb-17Li and to have the same value also at
8000 J/mol [6] and 5900 J/mol [7]. In the temperature the interface Pb-17Li/SS 316 membrane, i.e. the con-
range of 528-680 K direct measurements have been centration of deuterium in the upstream surface of the
performed of the solubility and diffusivity of deuterium membrane is C = C,. This assumption appears justified

Table 1
Deuterium diffusivity in SS AK1 316: Comparison of the results obtained in absence of Pb-17Li with values reported in literature

T+3(K) D (lo-‘* m*/s) F (lo-‘* m3 STP/m s bar’.‘) S (m3 STP/m’ bar’.j)

This work Ref. [3] This work Ref. [l] This work Ref. [7]
540 2.3kO.6 2.7 0.27 * 0.01 0.84 0.12 f 30% 0.35
547 4.6kO.8 3.2 1.0*0.04 1.0 0.22*221!& 0.36
600 6.8 f 1 7.0 3.0fO.l 3.4 0.44+18% 0.41
632 6.9+1 11 4.8 f 0.6 6.2 0.75 + 18% 0.42
652 10 *2 14 6.9*1 10 0.69 f 38% 0.44
680 18 *3 20 9.0*1 15 0.50 + 28% 0.47
 0. Gauisch, G. Hoddap / Deuterium permeability of stainless steel AISI 316 39

Table 2 ment one obtains for the deuterium flux at steady-state

Deuterium diffusivity in SS AISI 316: Results obtained in in presence of Pb-17Li:
presence of Pb-17Li

T*2 D F S
(K) (lo-r2 m2/s) (lo-l2 m3 STP/ (m3 STP/
m s bar o.5) m3 STP/m3 bar0-5) where Csp is a slight function of temperature and (Q,/
2C,), #f(T), indicating that the permeation flux Qsr,
518 0.68 f 0.11 0.39 f 0.07 0.57 f 34%
in presence of Pb-17Li is only slightly temperature
548 1.7 Yko.3 0.54 *0.05 0.32i271
570 1.9 *0.3 0.74+ 0.06 0.39 f 27% dependent. This explains why Fp = f( Q,,) varies less in
584 2.1 *0.3 1.1 *0.1 0.52 i 24% presence of Pb-17Li than in absence (table 3).
586 2.3 kO.3 0.7 *0.1 0.30 f 24% As to point (2) it can be seen directly that since
625 5.7 Yco.9 1.3 *0.2 0.23*27x Csp ++z C,, (table 3), Qsr, <<< Q,, and also Fp < F,. In
626 5.0 f0.6 1.1 *0.2 0.22 rfr27% presence of Pb-17Li F is not only less sensible to
629 4.6 +l.O 1.2 *0.2 0.26 + 27% temperature changes, but its values are also lower. The
665 8.5 k1.4 2.0 *0.4 0.24 f 36% relative decrease (table 3) is by a factor of about 1.7 at
526 K, to a factor of about 4 at 667 K. This unusual
variation of F = DS is probably due to the interconnec-
tion of D =f(T) and S #f(T). The explanation why D
because the diffusivity of deuterium in Pb-17Li (4 X
is lower in presence of Pb-17Li is the following.
10-10-l x 10e9 m2/s, 523-673 K) is much higher than
To calculate the diffusivity from the rise time period,
that in the SS 316 membrane (6 X 10-‘3-4 x lo-”
the formula valid for sufficiently long times has been
m’/s, 473-673 K) [6]. Since there is no deuterium
used [5]:
concentration gradient in the liquid Pb-17Li alloy at
steady-state, the alloy can be considered as “decoupled” D = cln ’ - ;t/Qs
.
from the membrane, and as a constant source of de- n2t
uterium for the membrane. Thus a deuterium con- The time t at which a given value of Q/Q, is obtained
centration gradient exists only in the membrane, i.e. is always longer in presence of Pb-17Li, leading to
CJL. The deuterium concentration in the downstream lower D-values (table 3). Also the time necessary to
surface of the membrane is C = 0, therefore Q, = DC,/ reach the values of the steady-state permeation fluxes is
L. If the letter “p” indicates the presence on the mem- longer. It can thus be seen that the measurement of the
brane of Pb-17Li during the experiment, and the letter permeation flux in presence of Pb-17Li was not a
“a” the absence, one has correct method for the determination of D.
As far as point (3) is concerned: since E(F) = 34000
(5) and E(D) = 39000 J/mol (the error may be up to
& 50%) it can be assumed that E(S) must be very
or, with Da/D, = 2 (table 3) small, and thus possibly equal to the activation energy
for the dissolution of deuterium in Pb-17Li which is
($i,= (g+f(T). (6) E(g) = 1350 J/mol [6] as compared to the previously
found E(S) = 12500 J/mol for the dissolution in SS
(Q, is in m2 STP/s, C, in m3 STP/m3). By rearrange- 316.

Table 3
Deuterium diffusivity in SS AISI 316: Comparison of the results obtained in absence and presence of Pb-17Li

T(K) In absence of Pb-17Li In presence of Pb-17Li

D x lOI F x lo’* s D x lo’* F x 1012 S
526 2.4 0.68 0.28 1.2 0.41 0.34
556 3.7 1.2 0.32 2.0 0.6 0.30
589 6.0 2.3 0.38 3.2 0.9 0.28
626 9.6 4.2 0.44 5.0 1.4 0.28
667 16 7.8 0.49 8.5 2.0 0.24
40 0. Gautsch, G. Hoddap / Deuterium permeability of stainless steel AISI 316

With regard to point (4) it turned out that E(D), = 52000 J/mol in absence of the alloy. These differences
E(D),. This means that the presence of Pb-17Li on the can be explained by the fact that the solubility of
steel membrane surface does not alter the activation deuterium in Pb-17Li is much lower than in the steel
energy for diffusion E(D). Therefore it can be assumed membrane and that it is nearly independent of tempera-
that also the frequency factor of the Arrhenius relation, ture. The Pb-17Li acts as a barrier for the deuterium
Do, for the bulk diffusion is not altered by the presence permeation flux.
of Pb-17Li on the membrane surface. Therefore also The diffusivity was by a factor of about 2 lower than
the bulk diffusion coefficients are in reality the same, in absence of the alloy. The activation energies for the
both, in presence and absence of Pb-17Li, i.e. Dp = Da. diffusion were, however, nearly the same in absence and
The apparent decrease of a factor of 2 (table 3 and presence of Pb-17Li. Since also the frequency factor D,,
explanation regarding point (2)) is due so the use of of the Arrhenius relation must be the same it is con-
permeation flux measurements which yield different cluded that also the diffusivities are the same. The
results in presence and absence of Pb-17Li. apparent decrease of a factor of 2 in presence of the
Concerning point (5): in presence of Pb-17Li F is alloy is due to the fact that the diffusivity is calculated
only slightly, whereas D is “ normally” (as in absence of from the ratio of two permeation fluxes and that it is
Pb-17Li) dependent on temperature. Therefore S = inversely proportional to the time, D =f(l/t), at which
F/D resulted to be nearly constant (table 3) with a given value of this ratio is reached. In presence of
increasing temperature (with a small tendency to de- Pb-17Li it takes about two times longer to reach the
crease). same value.
There are several sources of errors. That on F, which Although the values of permeability, diffusivity, and
is obtained by measuring an ion current (proportional solubility are helpful in estimating the permeation and
to a pressure), may reach f15W [4]. The error on D inventory of tritium in a breeding blanket, one ought to
(the ratio of two ion currents is involved) is about consider that both, permeation and inventory will be
_t 30%, and that on S (= F/D) about +45%. These strongly influenced, especially at lower temperature, by
values roughly correspond to the standard deviations in the tritium production rate in the Pb-17Li and the
tables 1 and 2. A further error of _tlO% derives from tritium extraction rate from the alloy. The presence of
the calibrated deuterium leak, from the deuterium driv- Pb-17Li in the blanket will not constitute a permeation
ing pressure (up to +lW) and from the temperature of barrier for tritium, because tritium is produced by
the membrane (+ 1% to + 2%). breeding within the alloy.

4. Conclusion References

The permeabilities and diffusivities of deuterium in [l] A.D. Le Claire, J. Nucl. Mater. 122 & 123 (1984) 1558.
SS AISI 316 membranes have been determined by means [2] E.H. Van Deventer, J. Nucl. Mater. 64 (1977) 325.
of permeation flux measurements as a function of tem- [3] W.G. Perkins and T. Noda, J. Nucl. Mater. 71 (1978) 349.
perature and deuterium driving pressure. The results [4] R. Frauenfelder, J. Chem. Phys. 48 (1968) 3955, 3966.
were reproducible and, within experimental error, in [5] R. Broudeur, J. Fidelle, P. Tison, C. Roux and M. Rapin,
Rapport CEA-R-4701 (1976).
good agreement with those reported in literature.
[6] F. Reiter and M. Caorlin, Nucl. Fuel Cycle Div., lspra
When the upstream surface of the membrane was
Establishment, private communication.
covered with a layer of the liquid eutectic alloy Pb-17Li, [7] J.L. Cecchi, J. Vat. Sci. Technol. 16 (1979) 58.
the permeability appeared much less temperature de- [8] C.H. Wu, NET-Group, MPI fir Plasmaphysik. Garching,
pendent as in absence of the alloy, and it was also by a private communication.
factor of 1 to 4 lower. The activation energy of permea- [9] E. Rued1 and V. Coen, Sci. Div., lspra Establishment,
tion was found to be only 34000 J/mol as compared to private communication.