Helium Diffusion Through Glass

by FRANCIS J. NORTON
General Electric Research Laboratory, Schenectady, New York

Using the mass spectrometer as a n analytical tool

and as a means of rate measurement, the rates of
helium diffusion through nine types of glass were
measured. The temperature range covered was
from -Noto 600OC. T h e highest rates were for
fused silica and Vycor brand glass; the lowest
were for the lead glasses. The rate at 100C. decreased by a factor of one million as the percentage of glassformers (SiOz, B2O3, P205) decreased
in the glass composition. A few measurements
are also given for hydrogen, deuterium, and neon
rates through fused silica and estimates are made
of the upper limit for argon, oxygen, and nitrogen

NEUTRAL GAS

JMATING

.3SlTlVE IONS

ENLARGED SIDE VIEW OF

IONIZATION CHAMBER.

at 700c.

1.

Introduction

TIE great sensitivity of the mass spectrometer as a specific detector of small amounts of material, and the adaptability of the instrument as an analytical tool, led to
the expectation that it should be useful in measuring the diffusion rate of gases through glass. This prediction turned
out to be true, and the present study gives the results of this
particular application of the mass spectrometer to the diffusion of helium through glass.
For comparison, a few measurements on the diffusion
through glass of hydrogen, deuterium, and neon are given.
It was also determined that the rates for argon, oxygen, and
nitrogen are exceedingly low, and limits are set.

II.

FIELD
INTO PAPER

Fig. 1.

FORCE

Principles of operation of the mass spectrometer.

Operation of the Mass Spectrometer

mass is 4.0039, and deuterium, Dz, of mass 4.0294. These

two ions, both of nominal mass 4, can be distinguished readily
and measured separately in the presence of each other.
The mass spectrometer used was the sector type,l 60-degree single focusing with 6-inch radius of curvature, made by
the General Electric Company. The general operational details have been described.*

The mass spectrometer is essentially a luge vacuum tube,

through which gases are pumped, at a pressure of about
111111. The principles of operation are shown in Fig. 1. Neutral gas molecules enter a small chamber a t the head of the
tube. A beam of 70-volt electrons traverses the chamber,
and their impact with the neutral gas molecules makes some
positive ions. These are accelerated in the source slits by the
applied 2000 volts and travel down the tube at 4 X lo7cm. per
scc. or about one million miles an hour. Entering a strong,
uniform inagnetic ficld, the ions are bent from their course
and, since they constitutc a positive ion current, they are deflected by the magnet. The result of the electrostatic acceleration and clectroinagnetic deflection is to sort out the
I)enms a t the bottom of the tube, according to the mass of
each positive ion. The light ions are deflected most and the
heavy ions least, and their orderly array is known as the mass
spectrum. A particular ion beam of any one mass, e.g., mass
4, can be selected by a slit and collected on a collector plate.
Here elcctrons flow to neutralize the positive ions, and the
minute electron current of about 10-l4 amp. is amplified and
measured on a recorder. The fact that mass,4 ions are being
collected means that helium is being measured; in fact, the
tiny difference can be measured between He, whose exact

111. Glass Diffusion Cell

Pure helium was sealed a t a precisely known pressure (450
mm. at 25C.) into small spherical glass bulbs with thin walls.
They were on the average 0.45 inch in diameter. The surface
area was calculated from their diameter.
The wall thickness was examined in three ways:
(1) A narrow pencil of X rays from an X-ray photometer3
was passed through the thin-walled glass bulb. The X-ray
beam was 0.0312 inch in diameter, and its intensity variation
was measured a t various locations on the bulb. This nondestructive test gave a measurement of uniformity in wall
thickness, so that the most uniform bulbs could be picked out
and used.

A. 0. Nier, Mass Spectrometer for Routine IsotopeAbundance Measurements, Rev. Sci. Instruments, 11, 212-16
(1940).
F. J. Norton, Mass Spectrum of Monoisotopic BZ1@Hs,
J . A m . Chem. Soc., 71,3488-91 (1949).
H. A. Liebhafsky, H. M. Smith, H. E. Tanis, and E. H. Win-

Presctitcd at the Fall Mccting of the Glass Division, The

Amcrican Ccrarnic Society, Corning, N. Y., October 12, 1951.
IZeceivcd May 26, 1952; rcvised copy rcccived July 18, 1952.
The author is a research associate with the Physical Chemistry

slow, Chemical Analysis Based on X-Ray Absorption Measurements with a Multiplier Phototube, Anal. Chem., 19, 861-65
(1947).

Section, General Electric Research Laboratory, Schenectady.

90

Xelium Diffusion Through Glass

AZarch 1953
Table 1.

Summary of Bulb Dimensions

----- ---

Bulb surface

7-

Wall thickness

KO

~.

Glass*

S O ? (C)
N o . 1720 (Cr
No 65O(L)
Yo 0080 ( C )
KO 7740(C)

I
2
:1
4

X-ray (P)

(i

91

(Mils)

(Mm.)

5.7
4.9
2 0
6 0
6 5
8.0
8.0
5.6
5.0
6.0

0.145
.126
051
152
165
,203
,203
.143
,127
.152

Diam.

Cm.

Area

In.

(cm.2)

1.135 0.447
1.171
.461
428
1 087
475
1 207
1 069
421
1.143 ,450
,371
0.942
.445
1.130
,430
1.092
1.143 ,450

4.0
4.3
3 7
4 6
3 6
4.1
2.8
4.0
3.7
4.1

AUX.
VACUUM

MASS
SPECTROMETER

7 Ib borate (GI
8
\.ycorbrakd(C)
9
Phosphate (H)
10 No. 7740(C)
___
________* Lctters in parcnthcses refer t o source of glass (see Section
I\-).

Fig. 2 .

( 2 ) Wall thickness was measured by an electrical method,

w1ierel)y the bulb was filled with a conducting water solution
arid the electrical capacitance was measured as the bulb was
immersed in water. Knowing the dielectric constant for that
glass and the surface area, the average thickness could be computed.
(3) Finally the bulbs were broken open and the thickness
was measured by a ball micrometer. A summary of bulbs
used is given in Table I. It is estimated that the thickness is
known to *lo%.

V.

Technique of Permeation Rate Measurements

The helium a t a known pressure of 450 mm. was sealed in

the thin-walled glass diffusion cell described. This cell was
sealed into a container of thick-walled glass, usually type No.
1720 (C). The loss of helium from this outer container was
negligible, owing to its low permeability, to its thickness, and
to the low helium pressure in it.
Because of the great sensitivity of the mass spectrometer as
a gas-flow measuring device, the loss of helium from inside the
glass cell during the measurement periods was entirely negligible. It amounted, in general, to a change well under 1%
of total pressure.
The outer container, as shown in Pig. 2, was attached to a
vacuum collection system, equipped with McLeod and Thermistor vacuum gauges.4 It could be evacuated into a subsidiary vacuum system, or directly into the mass spectrometer.
Then the helium diffused directly into the mass spectrometer
through the thin glass wall. The outer container was surrounded by a furnace or by a low-temperature bath, depending on the temperature range covered.
Three different overlapping methods were used to measure
permeation rate, as follows:
(1) The gas was diffused into the vacuum gas collection
system of known volume. The pressure was measured by a

IV. Types of Glass Used

The types of glass used, arranged in general order of decreasing rates of helium permeation, are given below with
their sourcc. The representative percentage compositions
arc given in Table 11. The glasses were as follows:
(1) Fused SOz,General Electric Company (G).
(2) I7ycorbrand, Corning Glass Works (C).
( 3 ) Chemical Pyrex brand No. 7740, Corning Glass Works

(3.
Rorosilicate No. 650, Owens-Illinois Glass Company

(4)

Apparatus used in permeation rate measurements.

(I,).
( 5 ) Phosphate glass, Haverford Glass Company (H). Only
traccs of SO, are present in this hydrofluoric acid-resistant
glass.
(ti) Soda-lime No. 0080, Corning Glass Works (C).
(i) Combustion or ignition tubing No. 1720, Corning
Glass Norks (C),
(8) X-ray shield glass, Pittsburgh Plate Glass Company
(P).
(9) Lead borate (78.27, PbO, molecular ratio PbO/B203
= 1.12), General Electric Company (G).

(4; G. L. Pearson, Thermistors: Their characteristics and

Uses, Bell Labs. Record, 19, 106-111 (Dec. 1940).
( b ) K. P. Dowell, Thermistors as Components, Elec. Mfg.,
84-91, 190-216 (Aug. 1948).

Table II. Glass Compositions

Glass No.

__

1
Fused
SiOr ( G )

2
Vycor
brand ( C )

3
No. 7740
(C)

100

96
3

81
13)

No, 650

Phosphaie

0
5
77
11

(L)

W)

6
No. 0080

7
No. 1720
(C)

8
X-ray
(P)

9
Pb
borate ( G )

72

62
5

31

0
22

18
15
8
61

78

67

31

22

(0

~~

SiOl

10

7
-

Sun1 of yo of glassformers SiOz

BzO~
f Pzo6

100

99

4
94

90

17

82

72

Journal of The American Ceramic Society-Norton

92

-4

Table 111.

Vol. 36, No. 3

Gas Permeation Rate Units

al

5-5

--5z
LL

Author

Barrer

-6

(a)

cp,

-7

P -8

rn
-I

I
-14

Positive

Ion

I
-13
Log,,

Fig. 3.

Amount

Current or Peak Heiaht

I

-I 2

-I I

Current

1
-10

(omp.)

Calibration curve showing the extended linearity between gasflow rate and peak height.

Smith and Taylor ( c )

Newkirk and Tooley ( d )
Rayleigh ( e )
Keesom (f)
Dushman (g)
Dushman (g)
Van Amerongen ( h )
Wilson, et al. (i)
Fast ( j )

Time

Sec.
Hr.
Sec.
Gm.
Sec.
M m 3 Day
Sec.
Sec.
pl.*
Hr.
Cm.3 Sec.
Cm.3 Sec.
Cm.3
Gm.

Cm.3

Hr.

Area

Cm2
Cm2
Cm2
Cm2
Cm.2
Cm.2

Cm2
Dm.2

Thickness

Pressure

Mm.
Mm.

Cm.
Atm.
Atm.
Atm.
Atm.

Mm.

Mm.
Mm.
Mm.
Mm.
Mm.
Cm.
Cm.
Mm.

diff.

Cm.
Atm.
Atm.
Atm.
Atm.

( a ) See p. 133, footnote 5.

( b ) See p. 116, footnote 5.
( c ) P. L. Smith and
W. Taylor, Diffusion of Helium
Through Several Glasses, J. Am. Ceram. Soc., 23 [5] 13946

,?.

(1940).

calibrated McLeod gauge; from period of collection, temperature, and pressure, the amount diffusing through the glass
wall could be computed. This was used for average and relatively high rates. A small portion of this gas was then analyzed by the mass spectrometer to make certain that i t was
100% helium.
( 2 ) The gas was led directly into the mass spectrometer.
The spectrometer was set so that the positive ion current,
or peak height due to helium of mass 4, was read continuously.
This stayed steady for hours at a time as steady flow state
prevailed. From method (l), where the absolute amount of
gas flowing was measured, the peak height could be plotted
against rate, and this is a linear relationship for the range of
rates used here. A calibration curve was therefore obtained
that could bc cxtended down to very low rates of gas flow,
since the mass spectrometer is so extremely sensitive. This
amounted to a continuous quantitative and qualitative analysis, with the mass spectrometer simultaneously acting as a
recording flowmeter. The time of establishment of equilibrium rate of flow could be determined readily and steadystate flow conditions were thus insured.
A calibration curve showing the extended linearity between
gas-flow rate and peak height is given in Fig. 3.
(3) A further extension of sensitivity into regions of very
low rates of gas flow is afforded by the accumulation method.
The outer container forms a small known collection volume,
about 25 c m 3 This has two stopcocks (greased with Apiezon
L), one leading to the mass spectrometer, the other to the
auxiliary vacuum system. Gas diffusing out of the thinwalled bulb into this space was collected for a measured length
of time, and then the stopcock opened quickly to the mass
spectrometer, which was focused and held for helium on peak
4. There resulted on the recorder a pulse indicating the helium. The height of this pulse with constant pumping conditions was found to be proportional to the amount of gas in the
container space. Separate calibrations without the glass
diffusion cell in the container space gave the absolute amount,
measured by the McLeod gauge in the known volume. I n
this way the accumulation or pulse method Fave a good
calibrating means in the range of very low diffusion rates, and
it overlapped the other calibrating methods.
It is essential, in this pulse method, that the mass spectrometer be focused accurately on the mass peak maximum.
This is insured by bleeding in through a small leak an extremely small amount of the gas being studied. Focusing can
then be done at leisure and the stopcock opened when all is
steady and at maximum focus. The amount of helium bled
in for the pulse method was a negligible fraction of the peak
pulse height (less than 0.1%). By this means, with collection
times of many hours, rates of flow down to one atom per microsecond could be measured.
It is also worth noting again that using the mass spectrometer as the measuring device, one can measure the diffusion

( d ) T. F. Newkirk and F. V. Tooley, Study of Effect of

Cation Concentration and Size on Helium Permeability of AlkaliSilica Glasses, ibid., 32 [9] 272-78 (1949).
( e ) Lord Rayleigh, Attempt t o Detect Passage of Helium
Through a Crystal Lattice at High Temperatures, Proc. Roy.
SOC.(London),A163, 376-80 (1937).
(f) W. H. Keesom, Helium, p. 131. Elsevier Book Company,
Inc., New York, 1942. 494 pp.
(g) Saul Dushman, Scientific Foundations of Vacuum Technique, p. 532. John Wiley & Sons, Inc., New York, 1949. 882
pp.; Ceram. Abstracts, 1950, May, p. 113f.
( h ) See footnote 6.
(i) L. H. Wilson, W. L. Sibbitt, and M. Jakob, Flow of
Gases in Porous Media, J . AppZ. Phys., 22 [8] 1027-30 (1951);
Ceram. Abstracts. 1952. Tan.. D. 13b.
( j ) J. D. Fast, Gas Peimeability of Metals, Philips Tech.
Rev., 6, 365-71; 63540 (1941); 7, 74-82 (1942).
* Microliters.

rate of the particular gas in which one is interested, to the

exclusion of other gases that may be diffusing. It is also relatively easy to measure a mixture of several diffusing gases
and to determine the specific rate for each one individually in
the presence of the others.

VI.

Permeation and Diffusion

A distinction is drawn between these two terms, especially

the permeation constant and diffusion constant.
Permeation is the over-all steady-state flow process from the gas
phase on one side of the membrane or wall to the gas phase on
the other side, Strictly speaking, the term diffusion in a solid
applies to the internal process by which an atom is handed on
or changed from one lattice position to another. In the case
of gases going through solids, diffusion as a separate process is
difficult to measure. The over-all process of permeation is
much easier to study.
Permeation involves several steps, as follows :
(I) Impact of the gas atoms or molecules on the surface.
(2) Adsorption.
(3) Possible dissociation upon adsorption.
(4) Solution of the gas in the wall material a t the incoming
surface to some equilibrium solubility value.
( 5 ) Movement of the gas atoms from this saturated surface layer through the interior as atoms or ions, under a concentration gradient, to the outgoing surface. This constitutes
diffusion proper. The concentration gradient is the driving
force for diffusion and the process is given quantitative formulation in Ficks law.6
(6) Transfer of the dissolved gas to the outgoing surface
layer with possible recombination.

6 R. M. Barrer, Diffusion in and through Solids.

Cambridge
Univ. Press, London; Macmillan Co., New York, 1941. 464 pp.;
Ceram. Abstracts, 21 [8] 179 (1942).

M a r c h 1953
Table IV.

Helium Diffusion Through Glass

Helium Diffusion Through Glasses; Experimental

Values of Permeation Constant K"
'Temp.
(OC.

Permeation
constant K

Vycor brand (C)

2
x 10-12
0
6 . 6 x 10-11
28
1.6 X
SiO2 glass (clear fused silica) (G)
- 78
1 . 3 x 10-12
- 23
2.15 X
0
5 . 3 x 10-11
25
1.14 X
100
5.85 x 10-10
700
2 . 1 x 10-8
Chcmical Pyrex brand No. 7740 (C)
4 . 9 x 10-14
- 78
1.6 X
-23
4 . 0 x 10-12
0
9.1 x 1 0 - 1 2
25
1 . 5 x 10-11
25
8 . 4 x 10-1'
100
1 . 0 x 10-10
107
1 . 4 X 10-l0
110
3 5 x 10-10
160
7.1 X
205
1 1 x 10-9
230
1 . 9 x 10-9
260
1 . 7 x 10-9
265
Rorosilicate No. 650 (L)
9 . 4 x 10-14
-23
9 . 2 x 10-13
+25
7 . 0 X lo-"
178
Phosphate (H)
1 . 9 x 10-14
28
9.1 x 10-13
120
7 . 6 x 10-12
200
1 2 x 10-11
215
2.i X 10-11
233
1 . 2 x 10-10
340
Soda-lime No. 0080 (C)
7 . 5 x 10-16
25
2 . 5 x 10-13
100
2 . 6 X lo-"
255
1 . 5 x 10-10
405
1 . 0 x 10-9
560
1 . 3 x 10-9
580
Combustion tubing No. 1720 (C)
119
3 . 0 x 10-14
185
3 . 1 x 10-13
236
1 . 2 x 10-12
310
8 . 3 x 10-12
1 . 0 x 10-11
329
380
1 . 6 x 10-11
476
5 . 1 X lo-"
X-ray shield glass ( P )
84
1.4 X
84
1.8 X
277
6 . 8 X lopla
888
5 . 9 x 10-12
Lead borate (G)
158
1.9 x 10-15
158
2 . 8 x 10-16
218
1 . 6 x 10-14
223
7 . 3 x 10-16
3 . 0 x 10-14
256
294
5 . 1 x 10-14
330
6 . 3 x 10-14

93

800 400

200
I

Temp. ("C.)
100 +25 0 - 2 5
I
I l l

-78

- 78

* Cin.3 gas (N.T.P.) per scc. pcr cm.2 area per mm. thickness
per unit diffcrcncc in partial pressure, measured in cm. of Hg.

(7) Desorption of this gas and release on the low-pressure

side of the membrane.
The surface process is an important distinction between (a)
hydrogen diffusing through iron and (b) its passage through
glass or polymers. In case (a), the hydrogen is adsorbed as a
molecule, but the competing attraction of the iron surface ren-

I'

2
IOOO/T

(OK.)

Fig. 4. Permeation velocity K of helium diffusing through various glasses.

The permeation velocity, laglo K, is platted against the reciprocal of the
absolute temperature, lOOO/T. The permeation velocity K i s in units of
gas (N.T.P.)/sec./cm.Z area/mm. thickness/cm. Hg (gas pressure
difference). Curve ( A ) lead borote glass; ( B ) X-ray shield glass; (C)
combustion tubing No. 1720; (D)soda-lime glass No. 0080; (1 phosphate glass; ( F ) borosilicate glass No. 650; (G)chemical Pyrex brand
glass No. 7740; ( H ) fused silica; and ( I ) Vycor brand glass.

Table V.

Activation Energy of the Various Glasses

Glass

Vvcor brand ( C )
F&d silica (G)'
Chemical Pyrex brand No. 7740 (C)
Borosilicate No. 650 (L)
Phosphate (H)
Soda-lime No. 0080 (C)
Combustion tubing No: 1720 (C)
X-ray shield (P)
Lead borate (G)

Activation energy
(cal./rzm. atom)

4.900
4; 900
6,400
7,600
11,000
11.000

12;ooo
12,500
9,300

ders its H-H bond weaker. Dissociation results, as shown

by the dependence of the rate of hydrogen permeation through
iron on the square root of the pressure. The hydrogen then
goes through as isolated atoms or protons and recombines on
the other side. For glass and polymers, the rate of permeation varies directly as the pressure, indicating no surface dissociation. The hydrogen penetrates these materials as the
molecule.
For the case of nondissociation, and the rare gases, the total
amount of material, 4,permeating a membrane is givena by

K = permeation velocity constant.

A = area of membrane exposed.
t = time.
= gas pressure on high side.
p 2 = gas pressure on low side.
d = thickness of membrane.

Temperature is assumed to remain uniform.

Two other constants are involved in the constant K: D,
the diffusion coefficient for the internal process, and S, the gas
solubility.

* G. J. van Amerongen, "Permeability of Rubbers t o Gases,"

J . APPl. Phys., 17, 972-85 (1946).

Journal of The American Ceramic Society-Norton

94

The units for expressing K vary widely in the literature.

Those used by Barrer5 will be eniployed here. K is in units
of cubic centimeters of gas a t normal temperature and pressure (0 and 760 mni.) per second per square centimeter area
per inillimeter thickness per centimeter of mercury (gas pressure difference). Table I11 gives the units employed by several authors for permeation velocity K . Important contributions to the knowledge of gas diffusion through glass have
been made by Taylor and others.7

VII.

al

As in so many physical and chemical processes, a graph of

ratc of pcmcation against the reciprocal of the absolute temperature generally gives a straight line.
The equation relating the permeation constant K to the
temperature is
I<

A =
Q =
K =
1 =

Ae-Q/R?

a colistatlt.
activation energy per gram atom.
the gas constant.
absolute temperature (OK. or C.

a
$ 4

..0

al
0

(1)
SiO,
Fig. 5.

+ 273).

Experimental Results

The values oi permeation velocity K obtained for helium

diffusing through different glasses are given in Table IV, and
the activation energy in Table V. The data are represented
in Fig. 4 as a plot of loglo R,the permeation velocity, against
the reciprocal of the absolute temperature, lOOO/T.

IX.

-~

(wt. %)

Heat of permeation of helium through various glasses.

-9
h

c
.c
-10

-al
0

-11

0
.-

-12

0)

-13

Activation Energy

From the slopes of the permeation velocity-temperature

curves, as explained before, the activation energy, Q, for each
of the glasses used can be calculated. This value determines
the temperature coefficient. Q can be calculated from two
values o permeation velocity R at two temperatures. From
equation (1) it follows that

+B,O, +P,05

(A) lead borate glass; (6) X-ray shield glass; [C) combustion
tubing No. 1720; ( D ) soda-lime glass No. 0080; (1 phosphate
glass; ( F ) borosilicate glass No. 650; (G)chemical Pyrex brand
glass No. 7740; [ H ) Vycor brand gloss; and ( I ) fused silica.

Q differs from heat of diffusion by the energy involved in

the surface and solution processes with the solid.
From this equation it is evident that if Q does not vary appreciably with temperature, and if logarithms are taken of
each side, log K against 1/T will yield a straight line, the slope
of which is proportional to the activation energy.

VIII.

Temperature Dependence of Permeation

Vol. 36, No. 3

x -14
0
-I

-15
-16
0

20

40
SiO, + B,O,

60

80

100

+ P205 (wt. %I

Permeation velocity K of helium diffusing through various glasses

The permeation velocity K is in units of C I ~ gas
. ~ (N.T.P.)/sec./
cm.* area/mm. thickness/cm. Hg (gas pressure difference). ( A ) Lead
borate glass (no SiOz); ( 6 ) X-ray shield glass; ( C ) combustion tubing
No. 1720; ( D ) soda-lime glass No. 0080; (1 phosphate glass (no SiOz);
( F ) borosilicate glass No. 650; (GIchemical Pyrex brand glass No. 7740;
( H ) fused silica; ( I ) Vycor brand glass.
Fig. 6.

4.57

2.3 X K O ;ROis the universal gas constant (cal. per mole).

The values of Q in order of decreasing permeation rates for

any given temperature are given in Table V and in Fig. 5 .
These energies in general increase with increasing additions
of nonglassformers. The one exception is the chemical compound lead borate, indicating that a different mechanism of
diffusion may operate here. As Fajans has indicated in his
papers, this material is unusual in many respects.

7 ( a ) N. W. Taylor arid William Rast, Diffusion of Helium

nud Hydrogen Through Pyrex Brand Chemically Resistant
Glass, J . Chewz. Yhys., 6 [lo] 612-19 (1938); Ceram. Abstracts, 20

.I61. 141 (1941).

( p ) W. D. Urry, Further Studies in Rare Gases: I, Permeability of Various Glasses t o Helium, J . Am. Chem. Soc., 54
[lo] 3887-3901 (1932).
( c ) W. D. Urry, Diffusion of Hydrogen and Helium Through
Silica Glass and Other Glasses,,,ibid., 44, 2160-67 (1922).
( d ) C. C. Van Voorhis, Diffusion of Helium Throueh
Several Widely DifferingGlasses, Phys. Rev., A23 [5] 557 (192;).
(e) I,. S. TSai and,,T. R. Hogncss, Diffusion of Gases
Through Fused Quartz, J. Phyr. Chem., 36 [lo] 2595-2600
(1932); Cernm. Abstracts, 12 [2]54 (1933).
(f)G. A. Williams and J. B. Ferguson, Solubility of Helium
and Hydrogen in Heated Silica Glass and the Relation of This to
the Permeability, J . Am. Chem. SOC.,46,635-39 (1924).

at 100C.

X. Relation to Glass Constitution

The general conclusion is that the lower the percentage of
glassformers, the lower is the permeation velocity. The glassforming oxides8are SOz,Bz03,and PzOs. These are summed, in
Table 11, and Fig. 6 shows the relationship for permeation at
100C. The permeation may be reduced by a factor of ten
million compared with silica glass, by introducing nonglassformers into the glass composition.
The passage oi helium through glass also is related to its
being an unorganized structure instead of an ordered crystal.
Lord Rayleigh attempted to measure the permeation through

* J. E. Stanworth, Physical Properties of Glass. Oxford Univ.

Press, London, 1950. 219 pp.; Ceram. Abstracts, 1950, Oct.,
p. 199f.

* Reference (e), Table 111.

Helium Difusian Through Glass

March 1953

4
0

-15

95

X-RAY SHIELD

d.488-

2.0

2.2

2.4

2.6 f

LEAD BORATE
d.5751

2.8

3.0

Density ( g r n . / ~ r n . ~ )
Fig. 7. Glass density vs. helium permeation velocity K a t 100C. ( A )
Vycor brand glass; ( B ) fused silica; ( C ) chemical Pyrex brand glass No.
7740; ( D ) borasilicate glass No. 650; ( E ) soda-lime glass No. 0080; (F)
phosphate glass; (G)combustion tubing No. 1720; (H)crystalline quartz
(Raleigh). X-ray shield glass and lead borate glass a r e off the graph.

Fig. 9.

(A)

Fig. 8. ( A ) Type of atomic arrangement in crystalline material possessing

symmetry and periodicity. ( B ) Form of atomic arrangement in a glass.

mica, a crystal, a t 41SC., and found it less than 7 X 10+ cu.

niiii. per day per 1 mm. thick per
area per atmosphere
helium pressure difference. I n the units of this paper, this
means that the permeation constant K is under 1 X 10-14cm.3
per sec. per em.* area per nim. thickness per cm. gas pressure
for mica ai 415C., which is roughly equivalent to a value of
K (at 100C.) under 1 X I0-lG. This latter is in the region
of the extrapolated value for K on Fig. 6, for no glassformers
prcscnt.
Rayleigh was also unable to find any diffusion of helium
through crystalline quartz, so the limit there is similar to that
for mica. Thiq mcans that in ordered crystalline silica, the
helium diffusion is many millions of times slower than through
the glassy form of the same composition, SiOz.
The contrast in closeness of packing in fused silica and crystalline quartz is rcflccted in the densities: SiOa glass = 2.20
gin. per
quartz = 2.65 gm. per
The densities of the glasses studied plotted against helium
permeation velocity constant K a t 100show a trend, illustrated in Fig. 7, where the wide differences in rate for the two
forms of silica are indicated. The two lead glasses, of high
density because of their lead content, lie outside the plot.

Atomic arrangement in a soda glass.

A picture of the atomic arrangements involved in these

various materials is afforded by the discussions of Zachariasen,9 Warren,lo and Stanworth
Figure 8 ( A ) indicates in two dimensions the type of atomic
arrangement in crystalline material possessing symmetry and
periodicity. The form of atomic arrangement in a glass is
given diagrammatically in Fig. 8 ( B ) .
Note that in forming an irregular lattice of glass, there are
many larger holes formed than occur in the regular crystal
structure. It is possible that these larger random openings in
the glassy structure of fused silica account for the increased
helium permeation. I n Fig. 9 is indicated the insertion of Na
ions, as in soda glass formation, into these lacunae. These
added nonglassformers plug the openings, reducing the helium
permeation rate below that of pure fused silica.
This picture may be extended by measurements of permeation of silica glass, for example, by gases whose molecular
sizes vary.
I t was found, in brief, that helium, hydrogen, deuterium,
and neon permeation rates could be measured, but on going to
argon, oxygen, or nitrogen, the permeation rates a t the same
temperature were lower by a factor of one hundred thousand,
or more. They could not be measured with this very sensitive mass spectrometer method; only an upper limit can be
given.
Rayleigh found an immeasurably low value for helium diffusion through crystalline silica and mica, lower than the rate
through silica glass by a factor of ten million or more.
This gives the picture of the helium atom being too large to
fit into the vacancies or holes in the quartz lattice; hence diffusion is lacking. But when the lattice opens up and loses
order as in forming the glass, there are openings large enough

9 W. H. Zachariasen, Atomic Arrangement in Glass, J . Am.

Chenz. Soc., 54 [lo] 3841-51 (1932).
10 B. E. Warren. X-Rav Determination of Structure of
Liquids and Glass, J . A&Z. Phys., 8 [lo] 645-55 (1937);
Ceram. Abstracts, 17 [2] 87 (1938).

Journal of The American Ceramic Society-Norton

96

Table VI.

Vol. 36, No. 3

Permeability Constant K Through Fused Silica

700C.

Helium
Hydrogen
Deuterium
Neon
Argon
Oxygen
Nitrogen

Table VII.

2 . 1 x lo-*
2 . 1 x 10-9

1 . 7 X 10-o
4.2 X 10-l0
Under 10-16
I'

' I

"

,'

600'C

1.25 X
2.8

10-10

Hydrogen Permeation Constant K

668OC

No. 1720 glass (C)

Mullite

Fig. 10. Diagram showing size of helium and argon atoms relative
to crystal lattice of quartz (helium radius = 1 a.u.)

between constituent atoms for the diffusion of helium to occur

readily.
An attempt is made to picture the situation in two dimensions for the arrangement in crystalline quartz, where the
Si-0-Si-0 lattice may take the form of a hexagon in certain
planes. The size of the atomic radii are represented as Si4+
= 0.41 a.u., 0--= 1.4 a.u. These are the values from Pauling" for the ionic form. Superimposed in Fig. 10 is the
helium atom of radius 1 8.u. It is just a little too large to fit
into the crystalline hole, but when a glass forms, the opening
becomes big enough, as indicated in Fig. 8(B),and the helium
can diffuse. The argon atom is so much larger, however, that
diffusion does not occur even when the lattice is opened up by
glass formation.
The values which the author measured for various gases
diffusing through fused silica are given in Table VI.
The atomic diameters from the gas kinetic theory, which
relies niainly on viscosity measurements, are given below, a t
700".
Atomic
diameters (a ii ) *

Helium
Neon
Hydrogen
Ikuterium
Oxygen
Argon

Nitrogen

1.95
2.4
2.5
2.55
3.15
3.2
3.4

The jump in size between deuterium and oxygen makes the

larger molecules just too large to fit into even the expanded
glass structure.

4.5 x 10-12
1 . 8 x 10-11

It is also evident that size is not the only factor. The neon
atom is smaller than the hydrogen molecule, but diffuses more
slowly by a factor of five. It is probable that surface and
solubility effects are related to the greater permeation velocity
for hydrogen.
It should again be emphasized that these measurements
made with the mass spectrometer are for the specific gas involved only, and this is established by continuous analysis.
The conclusions on openings may be summarized as follows :
(1) Crystalline quartz: helium diffusion very slow because of compact structure.
(2) Silica glass: helium diffusion rapid because of openings.
(3) Substituent glasses: helium diffusion slow again because substituent atoms of Na, Pb, etc., block the openings.

There are two other measurements of hydrogen permeation that are of interest because very low rates were obtained
in these materials which also gave low rates for helium. The
materials were (1) Corning combustion or ignition tubing No.
1720 and (2) fired aluminum silicate ceramic, mullite (3A1203.2Si02),of English-make T-1605 Morgan Battersea triangle
H5. The tubing was dark brown, 3/16 in. 0.D. with 1-mm.
wall. It could be sealed to No. 1720 glass. Cells for diffusion
were made from these two materials and the hydrogen permeation was determined (see Table VII).
The permeation of helium through glass No. 1720 (C) a t
the same temperature gave the extrapolated value of K =
3 x 10-10.
In the usual methods of measuring permeation velocities,
the gas is collected and its volunie and pressure are measured.
These quantities are so small that generally in the past no
analysis of this diffused gas has been made or has been even
possible. Hence, deceptively high results could have been obtained since other gases out of the glass and from the surface
were added to the diffusing gas, and the total was counted as
diffusing gas. The great sensitivity and specificity of the mass
spectrometer in analyzing and measuring the actual gas diffusing through the glass as shown here, should make future
measurements much more certain and simple.

.___
l1 Lirius Pauling, Nature of Chemical Bond and Structure of
Molecules and Crystals. Cornell Univ. Press, Ithaca, N. Y . ,
1940. 429 pp.; Ceram. Abstracts, 19 [l] 28 (1940).
* From Landolt-Bornstein Tabellen, I, 325 (1950).

Acknowledgments

The author wishes t o thank Louis Navias and Alexis G. Piiicus

OF this laboratory for helpful advice and discussion.