Van Wieringen, A.

Warmoltz, N.
1956.

Physica X X I I
849-865

ON T H E PERMEATION OF HYDROGEN AND HELIUM IN

SINGLE CRYSTAL SILICON AND GERMANIUM AT
ELEVATED T E M P E R A T U R E S
by A. VAN W I E R I N G E N and N. WARMOLTZ
Philips Research Laboratories N.V. Philips' Gloeilampenfabrieken
Eindhoven-Netherlands

Synopsis
A mass spectrometer examination of the permeability of the elements silicon and
germanium to the gases hydrogen and helium has been carried out in the temperature
range 967-1207C for silicon and 766-930C for germanium.
Using certain crystal growing and cutting techniques two kinds of diffusion cells
were made by each of which it was possible to determine both diffusion coefficient and
solubility as well as the activation energies of diffusion and solution from non-steadystate permeation measurements. No permeation of the gases neon, argon and nitrogen
could be detected. I t seems t h a t hydrogen in silicon can occur both in elemental form
and as a compound. The elemental form diffuses easily and consists mainly of atoms
or protons.

1. Introduction. It is known that very small amounts of impurities, which

are incorporated in the lattice of a semiconductor, m a y greatly influence the
electric properties. The preparation of elemental semiconductor material,
such as silicon and germanium, is normally carried out by solidification
from the molten phase, the solid phase is exposed to the ambient gas at
high temperature for a considerable time. It therefore seems possible that
some of these gases will be taken into the semiconductor so that its electric
behaviour might be influenced.
In this article we report results concerning the diffusion and solubility as
a function of temperature and pressure of the gases hydrogen and helium
in single crystal silicon and germanium. For this purpose permeation
measurements were made at different temperatures and gas pressures.
The noble gases are very poorly soluble in metals. This m a y be the reason
that no spontaneous permeation has ever been reported in the literature.
Gtintherschulze
and W i n t e r 1) (1939) mention some evidence
concerning the mobility of noble gases in iron, brought into the metal by a
gas discharge and le C l a i r e and R o w e z) (1955) succeeded in studying the
diffusion process of argon in silver, the gas being introduced into the metal
also by means of a gas discharge.
Mfiller, P i k e and G r a h a m 3) 1926) reported that 1 cm3 germanium,
after melting and cooling in hydrogen, contained 2.7 X 1019 molecules of
-

Physica XXII

849

54

850

A. VAN W I E R I N G E N AND N. WARMOLTZ

hydrogen, whereas H a g e n and S i e v e r t z 4) (1930) found that the solubility

of hydrogen in solid or liquid germanium at resp. 900C and 1000C was
smaller than 1.4 101s molec. Hz per cm 3 Ge. I w a s e and F u k u s h i m a 5)
(1937) noted an increasing absorption of hydrogen by silicon at high temperature with increasing temperature.

2. The di//usion cell. By means of the Czochralski technique high ohmic

almost cylindrical single crystal ingots of silicon and germanium were
drawn from the melt 6). A cylindrical hole was then drilled through the axis
using a diamond cutter. The hollow cylinder was afterwards ground and
polished to the desired size and shape and after etching sealed at one end
with the same material also by the above mentioned pulling technique. The
end is given a wall thickness much greater than that of the cylindrical wall
A

Furnace

Semiconductor

. ..
~,

--

Thermocouple

Vacuum

To mass spectrometer

Fig. 1. "Open cell" of semiconductor material used for the gas permeation
measurements.
in order to minimize permeation in this part of the cell. It is important that
during the sealing the crystal orientation of the cylinder is continued into
the end, so that whole single crystal specimens are obtained in which crystal
boundary diffusion is avoided. The other end of the cylinder was sealed,
again by pulling, with a long growth of crystal through which a narrow
axial hole was drilled to the cylindrical chamber. A tube of suitable glass
was then sealed around the open end, thus making a gas tight joint to the
mass spectrometer. See fig. 1. This cell could be heated by a low heat
capacity furnace, consisting of molybdenum wire supported by aluminium
oxide tubes, which enclosed the semiconductor cylinder. The temperature
of the furnace was kept constant within lC by current stabilization,

PERMEATION OF

H2 AND

H e IN SINGLE CRYSTAL Si AND Ge

851

As the glass joint must be kept at a lower temperature thero always

exists a temperature gradient along the cell. B y an appropriate construction
of the furnace it was possible to keep the temperature of the thin walled part
,~i the cell uniform within 15C. The temperature of this effective area wa~
defined b y averaging the values measured along its length.
In order to minimize diffusion through the long thick end of the cell at
high temperatures, the temperature gradient along the cell was made large
at the transition region between the end and the thin cylindrical wall of the
cell.
The gas under investigation is introduced into chamber A and diffuses
through the thin wall into chamber B from which it runs into the mass
spectrometer. This kind of cell will be called the "open cell".

Wofor
ZZI....~

~'tlaCO

Fig. 2. "Capsule" of semiconductor material for permeation measurements.

With this cell diffusion coefficients and solubilities can be found separately
b y analysing the curves, showing the permeation as a function of time during
a non-steady state which obtains when the steady state concentration pattern
in the wall is disturbed b y a sudden change of pressure or temperature.
Generally pressure changes can be established more rapidly than temperature
changes but when the non-steady state after a temperature change lasts
long in comparison to the time needed to obtain this change it is preferable
to use temperature changes as a means of disturbing the concentration
pattern and initiating the non-steady state permeation. With the cell
dimensions used in this experiment the temperature change method requires
a diffusion coefficient smaller than about 5 10-5 cm 2 sec -1.

852

A. VAN WIERINGEN AND N. WARMOLTZ

If the diffusion coefficient is small enough to be determined by the

temperature change method a semiconductor capsule, which is much
simpler to prepare, m a y be used. This capsule is a hollow cylinder sealed
at both ends in such a way that during sealing the crystal pulling apparatus
is flooded with the gas or a mixture of the gases to be investigated. We thus
obtained a whole single crystal specimen enclosing a quantity of gas or
gases. See figure 2.
The enclosed gas was allowed to permeate through the wall of the capsule
at high temperatures. To do this the capsule was placed in a fused silica
chamber A which was connected to the mass spectrometer; this chamber
in turn was placed in an evacuated chamber C to prevent helium diffusion
from the atmosphere through the silica wall.

..

Diff. c e i l - -

Manometer
6as supply

Calibration

"~l f

To vacuum pump

I
L~ . . . . . . .

) B Ii,4__Oven
L

||
Fig. 3. General assembly including mass spectrometer.

The capsule technique is free from the disadvantage of seals and temperature gradients to other materials and thus the whole specimen can be kept
at uniform temperature. The pressure of the gas inside the capsule can be
determined at the end of the experiments by opening it in a vessel of known
volume. The pressure decrease during the experiments can be entirely
neglected because the number of permeating molecules is very small in
comparison to the amount present in the capsule.

3. Measuring apparatus. For measurements of gas permeation m a n y

instruments can be used provided that their sensitivity is adapted to the
problem and correction can be made for the background originating from
the desorption of impurity gases. In the present experiments the pressure

P E R M E A T I O N OF

H2

AND

He

IN S I N G L E CRYSTAL

Si

AND

Ge

853

of the impurity gases was one or two orders of magnitude higher than the
pressure of the gases under investigation and thus the mass spectrometer,
which at the same time acts as a selective detector, appears to be the only
suitable instrument for gas flow measurements.
The mass spectrometer used was of the 60 degree sector type with a
radius of curvature of 5 cm, made of copper and stainless steel. See figure 3.
It was connected to a mercury diffusion pump with a liquid air trap, by
means of a metal stop cock with a teflon seal. All joints were made by
copper gaskets between stainless steel flanges. The whole system could
therefore be baked out at about 300C. 7).
The mass spectrometer and its power pack have been described by
P e 1c h o w i t z s). An oscillograph displays a large range of the mass spectrum
and also a single mass peak at set intervals, for example every 5 seconds.
In this way an idea can be obtained immediately of the composition of the
gas stream.
For the permeation measurements onlY the mass peak of the gas under
investigation was shown on the screen and the height was measured. The
ionizing electron current was 400#A, the electron energy 84 eV.
For calibration of the mass spectrometer an ionization gauge of the
tetrode type was mounted near the ion source.
The whole system including the diffusion cell was first baked out, the cell
being disconnected from the spectrometer and the gas evacuated by the
backing pump. When a sufficiently low pressure had been reached stop
cock C was closed, and B was opened, the baking out being continued. After
one to four days the baking o u t w a s stopped and the ion source of the spectrometer was excited. The room temperature pressure reached after the baking
out was less than 10-7 mm Hg. The whole vacuum system was free from
adsorbed gases to such an extent t h a t the stop cock A between the spectrometer and the pump could be closed for m a n y minutes without a considerable
pressure rise. This was a necessary condition in the helium experiments
because helium permeability through silicon and germanium is so small
t h a t only by collecting the helium over a certain period could reasonably
accurate readings be made. In such interval no measurable amount of helium
was absorbed by the apparatus; this was demonstrated by the fact that the
peak heights did not change after closing cock B.
The hydrogen permeability was so great that it was measurable with
cock A opened, i.e. even with the mass spectrometer connected to the pump.
Under these conditions the absorption in the system was negligible.
In order to calibrate the mass spectrometer an ionization gauge of the
tetrode type was mounted near the ion source. For the helium experiments
the spectrometer was calibrated by filling it with some helium from the
calibration gas system and then comparing the peak heights on the screen
with the pressure readings on the ion gauge, under exactly the same electrical

854

A. VAN WIERINGEN AND N. WARMOLTZ

conditions as during the experiments. The sensitivity was 1 m m peak height

on the screen for 1.90 10-9 m m Hg helium pressure.
The calibration for hydrogen on the other hand was carried out with the
gas flowing. Coming from a container, which was connected to a McLeod
gauge the hydrogen passed via a capillary leak through the spectrometer
to the pump. The rate of flow was thus known from pressure measurements
in the container and the peak height on the screen was determined as a
function of the gas flow. The result was 1 m m peak height for a flow of
1.20 1011 molecules of hydrogen per sec.
4. Analysis o/ the diHusion process. The assumption was made t h a t in
all measurements permeation of the gas through the wall was determined

"X~.\\\\\\\\\\~

of cy!ina,t"

J~l

It

i
Fig. 4. Concentration p a t t e r n during the s t e a d y state p e r m e a t i o n in a cylindrical wall.

only by the bulk diffusion process, this being the slowest of the successive
processes which make up the whole permeation process. The validity of this
assumption is borne out by our experiments.
The experiments gave the permeation F in molec, sec -1 through the cyhnder surface as a function of the diffusion coefficient and the concentration
gradient in the wall at r = a where the gas leaves it, i.e. the vacuum side.
See fig. 4.
The permeation measured is:
F = - - A . D . (~c/~r)~=~ Inolec. sec -1.

(1)

in which A is the cell area in cm z, D the diffusion coefficient in cm 2. sec -1

and (Oc/0r)r=~ the conc. gradient in molecules per cm 3 per cm at r = a.
In the open cell technique the inner side of the cylinder was exposed to

P E R M E A T I O N OF

H~

A N D H e IN SINGLE CRYSTAL

Si A N D

Ge

855

the v a c u u m in the spectrometer and the other side to a certain gas pressure
or to v a c u u m . To know the concentration gradient at r = a we re~luire a
solution of Fick's diffusion equation for the hollow cylinder with the given
conditions. T h e y are:
1 r = a

c=0

for a l l t

2 r = b

c = c , = b for a l l t

3 a < r < b

c=0

for t =

The solution is of the form c = [(t, r) with D and C~=b as parameters and is
given b y C a r s l a w and J a e g e r 9). In terms of k = b/a a n d l = b -- a from
this solution the following expression can be derived:

(Oc)_~,=a = C,=b k--1E

l

Ink

1+

2kl/'lnk

.~-~(--1)n.exp

k-- 1

( - - -~2n2
'7
~Dt)A(2

which for k values between 1 a n d about 1.5 gives a very good approximation.
Values of k exceeding 1.5 involve the use of the exact expressions with
Bessel functions as given in C a r s l a w and J a e g e r .
If c is finite at r = a a n d zero at r = b, for t = 0, t h e n the gas flows in
opposite direction t h r o u g h the wall. In this case the concentration gradient
at r = b is represented b y :

ac

c~_~ k - - 1
l "klnk

2kl/~lnk
k- 1

x~'=:

Dtll

(3)

The equations (2) and (3) were derived for the condition t h a t at t = 0, c = 0
for all r. T h e y can likewise be used for the description of the permeation
as afunction of time when a s t e a d y state is disturbed b y a change of the
gas concentration in the wall at the entrance side and the system proceeds
to a new s t e a d y state. In this case for the gas concentration in the wall at
the entrance side the difference of the entrance side concentrations in b o t h
s t e a d y states has to be taken. The time is defined to be zero when the sudden
pressure change takes place.
For values of k smaller t h a n about 1.5, which were always used, the
factor: k 1/2 in k~ k - 1 deviates less t h a n one percent from u n i t y which
deviation is negligible. The factor between [ ] in the Eqs. (2) and (3) now
reads:
1 + 2 ~ - ~ o ( _ 1)n.exp (--(~n~/12)Ot) which, p u t t i n g
(4)
exp (--

(~2/12)Ot) = q,

(5)

reduces to the infinite series:

1 -

2q + 2q4 -

2q9 + 2q16 . . . .

(6)

which rapidly converges for not too small values of t. For t = oo the perm-

8~6

A. VAN W l E R I N G E N AND N. WARMOLTZ

ation is Fs, that of the steady state, and according to (1), (2) and (3) is given
by:
C~=b k - - 1
c~=a k - - 1
Fs =
l
ln----k D.A, or F8 -- ~ - . k ln----kD.Ao
(7)
where A, is the inside surface and A0 the outside surface. From Eqs. (1),
(2) and (7) we conclude"
F/Fs = 1 -- 2q + 2q4 -- 2q9 + . . . .

(8)

which holds for both directions of flow through the wall.

Putting both D and l equal to unity the curve of log F / F , versus log z~2t
is drawn, using Eqs. (5) and (8). This curve is considered as the standard
curve and serves for the rapid determination of the diffusion coefficient. For
this purpose the non-steady-state measurements are plotted as log F / F s
versus log ~9't and the curves thus obtained are compared to the standard
curve b y superposition. The curves must be identical and the ratio of the
time scales thus obtained gives the value D/l 2 in which l is the known
wall thickness.
Once D is known from Eq. (7), C,=b can now be calculated. This represents
the equilibrium concentration in the wall at the entrance side during exposure to the applied gas pressure. Assuming the validity of Henry's law
the equilibrium concentration of the gas in a piece of semiconductor material
placed in one atmosphere of that gas can now be obtained. This concentration is defined as the solubility S and is given in molecules, cm-3. It is
to be noticed that for hydrogen, which turns out to dissociate on dissolving,
the square root of the gas pressure in Henry's law has to be employed.
The permeability P is defined by:
P ---- D S molecules, sec -1 . cm -1.

(9)

It represents the amount of gas in molecules per sec. which passes through
1 cmg of a flat wall of thickness ! cm with one atmosphere pressure on the
entrance side and vacuum on the other.
The temperature dependence of the quantities in (9) is expressed by:
P = P0. exp (-- Qp/RT) = Do. exp(-- QD/RT). So exp (-- H s / R T ) =
= DoSo. exp {-- (QD + Hs)/RT}.

(10)

Thus the activation energy of permeation Qp is the sum of the activation

energy of diffusion QD and the heat of solution Hs:
QP = QD + Ms

(11)

5. Results o/experiments. H y d r o g e n t h r o u g h s i l i c o n .
The open cell was used, the inner and outer diameters of the thin-walled
part being 1.80 and 2.24 cm respectively and the length 5.55 cm.

PERMEATION OF

H2 AND He IN SINGLE CRYSTAL S i AND Ge

857

The cell was first connected to the backing pump and heated for two days
at 800C before the measurements were started, as in the beginning of the
heating large quantities of gas were developed. These gases were mainly
water vapour, carbon monoxide, hydrogen and the lower hydro-carbons;
their presence in the semiconductor is a problem in itself.
u

II.B
11.6
t%1

I. 606 rnrn Hg

11.4
11.2

2.37~.

,,

3. 179

....

~.. 83

....

,,

e
71.0

"6
E

I0.8

.Q

I0.~

10.2
IO.C
9A

6.6

6.6

6.8

7.0

7.2

Z$

Z6

7.8

8.0

VT x 10$

Fig. 5. Hydrogen-silicon. Permeation as a function of temperature at several pressures.

The steady state permeation F s in molecules of hydrogen per sec. was
measured as a function of temperature for several gas pressures; figure 5
shows these measurements as converted for 1 cm 2 of a plane wall of 1 cm
thickness. All lines are straight and nearly parallel; t h e y represent an
activation energy for the permeation Qp of 54 4- 4 kcal. mole -1. The relation
between the permeation Fs of the hydrogen and its pressure p on the entrance
side of a wall can be represented by:
Fs ~ p~

(12)

at constant temperature.
Here the value o f x should be 0.5 if the hydrogen diffuses only as atoms or
protons and if diffusion is the ratedetermining step in the process of permeation. As a matter of fact from the data in figure 5 it turns out t h a t x
increases from 0,50 at 1050C to 0,56 at 1200C. This small increase at
higher temperatures will be discussed in section 7.
The non-steady-state of permeation was caused by a sudden replacement
of the vacuum in the outer chamber A by a certain hydrogen pressure.

858

A. VAN W I E R I N G E N AND N. WARMOLTZ

Immediately after this sudden change the increasing gas stream was
measured as a function of the time elapsed since the change. Curves of the
type of figure 6 are obtained in these measurements.

T
rime

Fig. 6. P e r m e a t i o n as a f u n c t i o n of t i m e d u r i n g a n o n - s t e a d y s t a t e a f t e r a s u d d e n
p r e s s u r e rise.

In section 5 a procedure has been given to derive the diffusion coefficient

D from these curves by means of comparison with a standard curve. "When
D is known the corresponding value of the solubility S can be calculated.
Table I gives data for D at different temperatures together with the correTABLE I
Hydrogen through silicon. Diffusion coefficient D, solubility at one
atmosphere S and permeability P at different temperatures.

oC

O x 10 ~'

SxIO-I/"

in m o l e c ,

PxlO-1 I

molec, c m -

cm2.sec - I cm_3.ot lot~ ~ec-l.at latin

1092

1.70

3./,

5.7

1115

1.80

/*.1

7.4

8.1.

1127

1.80

/*.7

1132

1.76

5./~

9.4

1195
1199

2,23
2.30

9.8
10,0

21.9
33.0

1200

2'.I?

11.2

2/+.3

sponding solubilities and permeabilities; t h e y are plotted as a function of

1/T in figure 7. From the slopes one obtains for the activation energy of
diffusion QD = 11 kcal. mole -1 and for the heat of solution I-Is = 43 kcal.
mole -I. These values are correct to within about 10%. The diffusion coef-

PERMEATION OF H2 AND He IN SINGLE CRYSTAL Si AND Ge

859

ficient and the solubility are given by:

D = 9.4 10-3.exp (-- 11000/RT) cm 9"sec -1.

43000/RT) molecules cm-3 at

S = 2.4 10al. exp (--

1 atm gas pressure.

Helium through silicon.

The same equipment as for hydrogen was used. The flow density of the
gas from the cell was now too small to be measured with continuous
evacuation oi the spectrometer. The stop cock A was therefore closed for
intervals of 3 minutes during which the pressure inside the spectrometer
increased to a measurable value.

11
Io
9

\
"i

t~

5~
I

\
3
J

2.5

>c 1.5
l

6.7

6.S

6.9

ZO

?.I

Z2
?.3
~ I/T X 104

7.4

7.5

Fig. 7. Hydrogen-silicon. Diffusion coefficient D and solubility S as

a function of temperature.

At a constant pressure and varying temperatures the permeability was

determined in the range 967 - 1207C. In figure 8 the log P data are
plotted versus 1/T; the slope of this curve gives an activation energy for the
permeability of 40 kcal. mole -1 with a possible error of 10%.
B y means of sudden pressure changes the diffusion coefficient was
determined at different temperatures in the same way as was done for the
hydrogen diffusion. Figure 9 gives some values of D together with the
corresponding S, measured by this method. The points are too few to be
able to draw straight lines through them without further data. There is,
however, a method of ascertaining the slope of these lines by a direct
measurement of the activation energies QD and H s by means of fast temperature changes. When the diffusion is not too fast i.e. when D is smaller (see
section 2) t h a n about 5 10-5 cm 2 . sec -1, as is the case here, the concen-

860

A. VAN W I E R I N G E N A N D N. WARMOLTZ

tration pattern in the wall corresponding to the previous steady state is

maintained for a while after this temperature change whereas D immediately
8.4
8.,1
e.o

"".

7.~

"r
7.6

Z4

"x,

Z~

7.O
6.0
6.6
6.6

6.8

ZO

7.2

Zg

7.6

7.8

8.0

8.2

I/T X 104`
F i g . 8. H e l i u m - s i l i c o n . P e r m e a b i l i t y P a s a f u n c t i o n of t e m p e r a t u r e .

assumes a value corresponding to the new temperature. From the values for
F before and just after the temperature change the activation energy {)D
can be derived; the heat of solution Hs is derived by comparing the value
of the permeation just after the temperature change to that in the new
steady state, which is established after a lapse of time. Table II shows these
TABLE I I
Helium through silicon. Act. energy of diff. QD, heat of solution H 8 and act. energy of permeation
Qp as determined by temperature jumps, together with the mean values.

QD (kcol./
rnol.)
2,$
30

Hs (kcoZ./ Qp(kcol./
tool.)
tool.)
f5

39

39

33
26

11

37

32

4.1

29

11

4.0

energies measured at different temperatures and their mean values; the

latter are used to determine the slopes of' the straight lines to be drawn
through the points in figure 9.
The scattering of the points in fig. 9 m a y be partly explained by unstability
of the mass spectrometer in the course of one D measurement for which

PERMEATION OF Hg. AND

He IN SINGLE CRYSTAL Si AND Ge

861

about 100 minutes arerequired. The diffusion coefficient and the solt~bility
are given by:
D = 0.11. exp (--

29000/RT) cm~. sec -1.

S = 6.5 1014. exp (-- 11000/RT) molec, cm -3 at 1 atm.

Hydrogen through germanium.
As no open cell was available the capsule method was used here. Inner
and outer diameters of the cylindrical wall of the capsule were respectively
1.73 and 2.16 cm and the length was 4.25 cm. The sealed ends were about
5 times thicker and 6 times smaller in surface t h a n the cylindrical wall and
-'2.
2.2

2.0 r
1.8 '~
1.6 t3
1.4.
1.2
1.0
0.8 i
t

e4
E
to,

6.7

6.8

6.9

I/T X l04
Fig. 9. Helium-silicon. Diffusion coefficient D a n d s o l u b i l i t y S as a
function of temperature.

therefore the permeation through the ends amounted to only about 30/o of
the total permeation. For the determination of D no correction with regard
to this leak was necessary; the measured solubility, however, had to be
reduced by 3~/o. The pressure inside the capsule varies with temperature,
and this has to be accounted for in the capsule measurements.
The P versus 1/T plot gave again a straight line in the measured interval
780 -- 930C. The slope of this line gave an activation energy Qp = 47 -1- 4
kcal mole -1. See fig. I0.
P = 2.3 1019 exp (--

47000/RT)

The non-steady-state of permeation obtained by means of a temperature

change could not be maintained long enough to make possible any measurements of P as a function of time. From this it is concluded that the diffusion

869~

A. VAN WIERINGEN AND N. WARMOLTZ

coefficient D at a n y t e m p e r a t u r e in the m e n t i o n e d range is greater t h a n

5 10-5 cm 9. . sec-1. B y m e a n s of Eq. (9) the solubility S is t h u s f o u n d to be
smaller t h a n 4.6 1014 molecules Ha. cm -8.
11.4

11.3
17.0
,J.
J

10.8 ~

10.6

10.l,
10.2

10.0
9.8

\,o

9.8

9.2

8.2

8.4 8.6 0.8

9.0

9.2 9./*
I/T X 104

9.6

9.8

Fig. 10. H y d r o g e n - g e r m a n i u m . P e r m e a b i l i t y P as a f u n c t i o n of t e m p e r a t u r e .

Helium through germanium.

The capsule used h a d inner and oute~ d i a m e t e r of respectively 1.78 and
2.20 cm and length 4.95 cm.
7s~c

~C

16.3kcol

~. 7kcaZ
TkcoZ

7#90c
!
I

3"o ~o

~o 8o

~bo t~o

Minutes

Fig. 11. Helium-germanium. Permeation as a function of time applying sudden

temperature changes.

PERMEATION

OF

Hg. A N D He

IN S I N G L E C R Y S T A L

Si A N D

Ge

863

The measurements were carried out in the temperature range 795 -~ 872C.
The activation energy Qp as derived from the straight log Fs versus lIT
plot was 32 4- 4 kcal mole -1.
The activation energy for the diffusion QD is derived from the step change,
and the heat of solution from the subsequent gradual change of the permeation time curve in the non-steady state. See fig. 11. QD was found to be
16 4- 2 kcal mole -1 and Hs 13 2 kcal mole -I, both independent of the
temperature. The non-steady-state parts A -- B and C -- D in fig. 11 were
used to determine graphically the diffusion coefficient at different temperatures by comparing with the standard curve. These data together with those
for the solubility are gathered in figure 12; the slopes of the two straight

?
o

?
(n

--.<.

,4

8.8

9.0

9.~

9.4

~ ~r x lO4

Fig. 12. Helium-germanium. Diffusion coefficient D and solubility S as a

function of temperature.

lines drawn through them are made to correspond with the previously
derived activation energies for D and S. The diffusion coefficient and the
solubility are given by:
D = 6.1 x 10-a exp (-S = 1.4 1015 exp (--

16000/RT) cm 2 . sec -1
13000/RT) mole. cm -3

at 1 atm.

864

A. VAN W I E R I N G E N AND N. WARMOLTZ

TABLE III

Survey of measured activation energies and heats of solution, and values of diffusion coeffcient,
solubility and permeability given for corresponding temperatures of silicon and germanium.

Gasmater'iaZ
Couple

D in
cm2.sec-1

S in molec. Pin molec.

Hs Oo+Hs
crn- 1.sec- 1, Oo
cm-3.at I
k cal / kcal /
kcal l
arm at lotto.
gnat. 9nat
gnat
1,0xi015
2.3x1011
11
.3") 5.')

H2

Z3xIO -4

(1200 C) He

5.6x10 -6

1.6xi013

8.8xi0 7

29

11

60

> 5 x i 0 -5

<4.6xfO 16

2.3x101o

67 x)

5.2x10 - 6

4.5x10 f2

3.3xi07

16

13

29

s~

Se

(e6oc)
x)

He

these values include half the dissociation energy for one gr. real. H 2 (52 kcol).

6. Discussion o~ results. The graphical analysis of the measured nonsteady-state curves showed that within experimental errors all the curves
are identical in shape to the standard curve which was based on the concept
of a simple bulk process. From this we m a y conclude that in our measurements no slow surface processes were present. This is supported b y the fact
that the hydrogen permeation at least at the lower temperatures is proportional to the square root of pressure, which after C h a n g and B e n n e t t 10)
proves the absence of slow surface processes.
There is a small deviation from the square root law at the higher temperatures for the hydrogen permeation through silicon. This could be explained
b y assuming that the diffusion at high temperatures occurs partly in the
form of hydrogen molecules.The effect is too small however for this explanation to be conclusive.
All curves ill which log Fs was plotted versus 1/T are straight lines
within the error of the measurements. It is therefore unlikely that transition
points occur in the silicon and germanium crystal lattices within the temperature intervals covered b y the experiments. This is in agreement with the
findings of E. A. W o o d 11).
The interstitial sites in the silicon and germanium lattices are sufficiently
large to be able to easily contain the hydrogen and helium atoms. From the
fact that the diffusivities are extremely high it m a y further be concluded
that the diffusion of hydrogen and helium in silicon and germanium is very
likely to be interstitial; compare F u l l e r , S t r u t h e r s , D i t z e n b e r g e r
and W o 1f s t i r n 1~). The development of HzO. from the silicon and germanium
specimens gives rise to the suggestion that at the investigated temperatures
hydrogen can exist in the lattice in at least two different ways. Firstly there
is occurence as atoms or protons of which the behaviour has been studied
and secondly it is present in a form in which it is more or less bound to

PERMEATION OF

H2 AND

H e IN SINGLE CRYSTAL Si AND G e

865

oxygen, because the escape of water continues even after total extraction
of any unbound hydrogen.
There is no evidence of any permanent influence of dissolved hydrogen on
the electrical properties of silicon and germanium. We must therefore
conclude that hydrogen at least in silicon is not ionized at room temperature,
as in view of the relatively high solubility any reasonable ionization would
be detected as a change in resistivity. R e i s s 18) pointed out that dissolved
hydrogen will behave differently from the bigger elements Li and Cu which
ionize to a certain degree.
The gases neon, argon and nitrogen showed no measurable permeability
in silicon and germanium. Their permeability constants P are estimated to
be smaller than 5 108 sec -1. cm -1 at 1200C for Si and 900C for Ge.

Acknowledgement. The authors are much indebted to Dr. F. V a n d e r

M a e s e n of this laboratory for helpful discussions and his aid in solving the
mathematical problems. Prof. J. D. F a s t , Dr. J. L. M e i j e r i n g , It. F. H.
S t i e l t j e s and Dr. H. J. V i n k of this laboratory gave many a good
suggestion in the course of this investigation. Thanks are due to Messrs.
G. T. B r e u r i n g and A. M. J. G. V a n R u n for the growing and sealing
of the crystals.
Received 26-6-56.
REFERENCES
1)
2)
3)
4)
5)

G f i n t h e r s e h u l z e , A., W i n t e r , A., Z. Physik 112, 648-53 (1939).

Le Claire, A. D., Rowe, A. H., Rev. M6tallurgie 52, 94-104 (1955).
Muller, J. H., Pike, E. F., G r a h a m , A. K., Proe. Amer. philos. Soe. 65, 15-32 (1926).
H a g e n , H., S i e v e r t z , A., Z. Anorg. AUg. Chemie 185, 225-38 (1930).
I w a s e , K., F u k u s h i m a , M., Nippon Kinzoku Gakukai-Si 1, 202-13 (1937); C.A. 32, 5750

6)
7)
8)
9)

Teal, O. K., L i t t l e , J. B., Phys. Rev. 78, 647 (1950).

W a r m o l t z , N. Applied Mass Spectrometry, Institute of Petroleum, London, 1954.
P e l c h o w i t z , I., Philips Res. Reports 9, 1-80 (1954).
C a r s l a w , H. S., J a e g e r , J. C. Conduction of heat in solids p. 178 Oxford, Clarendon press
1947.
C h a n g , P. L., B e n n e t t , W. D. G., J. Iron Steel Inst. 205-13 (1952).
Wood, E l i s a b e t h A., J. phys. Chem. 60, 508-9 (1956).
F u l l e r , C. S., S t r u t h e r s , J. D., D i t z e n b e r g e r , J. A., W o l f s t i r n , K. B., Phys. Rev. 93,
1182-9 (1954).
Reiss, H., Bull. Am. phys. Soe. 30, 22 (1955).

(1938).

10)
11)
12)
13)

Physica XXII

55