Measuring hydrogen in

aluminum alloys
by Paul D. Hess
A number of methods are available for meas uring H2 in AI alloys, but a combination of
methods looks like the best answer.

For control
purposes, the
H YDROGEN is the only gas appreciably
soluble in aluminum. Its solubility is
alloys before the metal is cast. This has long
been recognized as a problem and, since the
small related to the solubility of hydrogen early days of the aluminum industry, a va-
vacuum gas test
in many other metals. The problems en- riety of methods has been developed with
gives valuable varying degrees of sophistication. The meth-
countered with gas in aluminum are due to
information on ods may be divided into two general
the great difference in solubilities of hydro-
metal cleanli- classes-"indirect," which involves observa-
gen in liquid and solid metal at the freezing
ness. The Alcoa point. As shown in Fig. 1, the ratio of liquid tion of some physical characteristic but does
test extends its solubility to solid solubility at the freezing not represent an absolute determination of
sensitivity to point is approximately 20: 1, which results in hydrogen; and "direct," in which the actual
hydrogen. hydrogen coming out of solution to form hydrogen content of the metal is determined.
bubbles during solidification of the metal.
INDIRECT METHODS
This can cause porosity in castings and
ingots, blisters on sheet, and so called Straube·-Pfeiffer Test
"bright Bakes" in forgings. The Straube-Pfeiffer test,1,2,3,4 often called
For these reasons it is important to deter- the vacuum gas test, is convenient and rapid
mine hydrogen content of molten aluminum for control purposes. An iron crucible con-

TEMPERATURE,DEGREES F
1000 1100 1200 1300 1400 1500 1600
2 .2 2014 ALLoY 0 - 8EFOAE F ILTERING
o - AftER FILTERING
19
2 .0 III
20 8 !%J
1.8
III I!J
2.1
1.6 13
~ I!!l

0'"
9
1.4 ..~
,.:
t:
2.2

IliI III
..... 2 .3 El III
1. 2
5
Ii:
f0-
CI)
1.0
...
....

'" 2.'
(;;t~
.~
III
~"
E IliI m
0 .8 2~

0 .6
...~
26 (;)<il
0
0
0 .4 27 e (i) (;) <l>

0.2
2B
13
~ (;)

SOLID 0 01 02 03 O' 05 06 01 OB
~01..0""5~5:;0=60i::0==6:::::15~0---7-!-00""""'7..J5L-0---8..1.00-8..J5L-0.--J900 HYDROGEN CONTE T, mlll00;

TEMPERATURE ,DEGREES C
Fig. 2-The effect of filtering molten aluminum
Fig. 1-Solubility of hydrogen in aluminum at one alloys by the Alcoa 94 Process on Straube-Pfeiffer
atmosphere hydrogen pressure test results

46-JOURNAL OF METALS, OCTOBER 1973

taining a molten sample of the aluminum
alloy is placed in a vacuum chamber. Pres-
sure is reduced rapidly to a selected value,
usually 50 torr for foundry alloys or 2-5 torr
for wrought alloys and held during solidifi-
cation. The sample may be evaluated by:
a) careful observation for bubble evolution
during solidification, b) determination of its
density and comparison with standard den-
sity for the alloy, and c) sectioning the
sample and examining cut surfaces for poros-
ity. The behavior of the sample is affected by
presence or absence of inclusions, which act
as nucleating agents to facilitate formation
of hydrogen bubbles during solidification.
This test does not provide an absolute mea-
sure of the hydrogen content but rather re-
flects the combined effects of hydrogen and
inclusions.6 A useful control test it will
give valuable information as to cleanliness
of the metal with respect to inclusions and
potential effects of hydrogen.
With the development of modern methods
of metal treatment, such as filtration
and combination filtration-fluxing processes,
7,8,9,10,11 it became necessary to develop
more sensitive tests. Figure 2 shows the
effect of filtering molten aluminum alloys by
the Alcoa 94 Process on Straube-Pfeiffer test
results. Although hydrogen is not removed
by filtration alone, increased densities and
less evolution of bubbles are always ob-
served in the vacuum gas test samples. It
was found that the limit of sensitivity on
inclusion free metal is about 0.3 ml hydrogen
per 100 grams. This is far too high for critical
products and rapid test was developed that
would be sensitive to lower hydrogen con-
tents.
2.5 r-----,.......----.-------~
>-
>-
in
z 0 ,6
w o
o
...o
>
faN 2 .6
;:;
in
zw
'"
§.
2 ,7
l;
2 .8 L-..-_ _ _ ___'~ _ _ __ _ l_ _ _ ____'
o 0 ,1 0 ,2
HYDROGEN COtlTENT. 11\111000
Fig. 3-Relation of hydrogen content by Telegas to
5 torr Alcoa sensitized vacuum gas test densities
(2014 alloy)
Alcoa Sensitized Vacuum Density Test
In development of this test, a number of
methods were tried to improve sensitivity of
the vacuum gas test to hydrogen in the ab-
sence of inclusions. These include lower
pressures, maintaining the sample marten for
longer periods of time, seeding the sample
with aluminum oxide, and controlled vibra-
tion of the sample during solidification. It
was found that controlled vibration was the
most effective and consistent of these meth-
ods. It is necessary to control both the
amplitude and frequency of vibration. At
the optimum conditions of vibration, a sen-
sitivity to hydrogen content of 0.1 ml/lOO g
is obtained. To obtain maximum benefit from
the sensitized test in determining the level
of hydrogen per se, the metal should not
The advantage
contain inclusions. Hence, a static vacuum
gas test sample should be taken and deter-
of the initial
mined to be sound before the Alcoa Sensi- bubble test is
tized Vacuum Density Test is applied. This the rapidity and
test is semiquantitative in nature and the convenience
density of a properly treated sample of in- with which it
clusion-free metal is related to the hydrogen can be
content of the metal (Fig. 3). performed.
Initial Bubble Test
The initial bubble test was first proposed
by Dardel in 1947. 12 In Russia, it is known as
the Gudchenko test. In this method, the
crucible containing the sample is placed in
an electrically heated vacuum chamber so
that the sample remains molten during the
test. Pressure is reduced until the first gas
bubble is observed at the surface of the
metal. The pressure and temperature at
which the first bubble appears are recorded.
A nomograph relating pressure, temperature,
and hydrogen solubility of the alloy being
tested is used to indicate hydrogen content.
0,7 r--,--.---,----r---r--,.----.
...
",,,,
0.5
~8
~~ 0.4
m:i
",w
CD 8 0.3
..J'"
",0
t=~
~ 0 .2
0.1
0 .3 0.1 0.2 0.3 0 .4 0.5 0.6 0.7
HYOROGEN, mlliOOg
TELEGAS
Fig. 4-Comparison of Telegas and initial bubble
test results
OCTOBER 1973, JOURNAL OF METALS-47

Recently, there has been a resurgence of
interest in this test as a result of the market-
ing of an apparatus which enables close con-
trol and provides convenience in both testing
and calculation. I3 The advantage of the ini-
tial bubble test is the rapidity and conveni-
ence with which it can be performed. The
greatest disadvantage is the frequ ent occur-
rence of erratic results. Samples are affected
by presence or absence of inclusions, in the
same manner as in the Straube-Pfeiffer test.
In investigations of the initial bubble test at
Alcoa Research Laboratories, a bubble did
not appear in 18% of the samples, even when
pressure was reduced to < 1 torr. In these
PAUL D. HESS cases, there was still appreciable hydrogen
Paul D. Hess received present as determined by both Telegas and
the Bachelor of Science
in Chemical Engineering solid extraction methods. Figure 4 shows the
degree from Linsly
Institute of Technology. relationship between initial bubble test re-
In 1939, he joined Alcoa sults and Telegas hydrogen determinations.
in the Metallurgical
Dept. of the Vernon The relationship between initial bubble test
Works, Calif., where his
activities included ingot, results and solid extraction determinations
rivet and aluminum and
magnesium foundry is shown in Figure 5. Duplicate results by
work, and solution of
air pollution problems. the initial bubble method are compared in
In 1955, he transferred Figure 6. It will be noted that several times
to Alcoa Research
Laboratories. where his there were no bubbles in one test and
work has been primarily
on molten aluminum bubbles in a duplicate test, as well as no
operations. He is the
holder or co-holder of bubbles in either test.
seven patents and has
authored numerous
papers on subjects such
as gas in aluminum The initial bubble test is theoretically at-
alloys, methods of
molten metal treatment,
safety, air pollution
tractive and under certain conditions may
control, and refractories give results that check with other methods.
for molten aluminum
operations. However, as with the Straube-Pfeiffer test,
modem methods of metal treatment provide
metal of such freedom from inclusions as to
preclude formation of a bubble, even when
an undesirable concentration of hydrogen is
present. It has been stated that vibration
assists the bubble formation, but as far as
we know, only the effects of adventitious
48~JOURNAL OF METALS, OCTOBER 1973
Fig. 5-Comparison of solid extraction and initial
bubble test results
0.7 ,----.------,------.-----,--. - - - - . - - --"
0 .6
...
V> .,.
0.5
",0
... 0
wS
.J e
0.4
III
III -
;:>Z
IlI~
.Jo 0 .3
...~:I:,..
<'"
_0
0 .2
0 .1
0.1 0 .2 0 .3 0.4 0 .5 0 .6 0 .7
HY DROGEN , mlll009
SOLID EXTRIICTION
vibration from the vacuum pump or from
plant machinery have been observed, with
no attempt to use controlled vibration.
DIRECT METHODS
Vacuum Fusion and Solid Extraction
Direct methods, which provide absolute de-
termination of hydrogen in aluminum alloys,
include vacuum fusion, solid extraction and
Telegas, or slight modifications of these
methods. The first two are essentially labora-
tory methods, not suitable for use on a plant
floor. Solid samples are used in both vacuum
fusion and solid extraction methods although
the solid extraction method has been modi-
fied experimentally at Alcoa Research Lab-
boratories to permit direct insertion of a
molten sample into the apparatus. In the
modified test, the sample is allowed to
solidify, and extraction is begun just below
the freezing point. This procedure had the
advantage of eliminating sample prepara-
tion but was suitable only for laboratory use
and required the apparatus to be adjacent to
the source of molten metal. H ence, the use
was discontinued. Vacuum fusion and solid
extraction have b een described by Brandt
and Cochran.14
If a sample cast from a melt is used, it
must be carefully taken in a mold which
permits rapid solidification coupled with
good feeding to prevent porosity. A copper
mold, as designed by Ransley,I5 is satisfac-
tory for this purpose. After machining, the
sample must be treated to remove surface
gas and moisture. Although benzene wash-
ing has been used, treatment with Alcoa-5
Bright Dip is superior.
In the vacuum fusion test, the machined,
surface cleaned, and weighed sample is
melted under vacuum. Hydrogen is ex-
Fig. 6-Comparison of duplicate hydrogen deter-
minations by the initial bubble test
0.7 r--,--.,...---,---,-----.-----r---"
0.6
0.5
C!'
o o
o ~::::
~~ 0.4
WE
.... ~ 0%
OZ
~~ 0 .3
uO
we>:
<1)0
,..
l: 0 .2
0 .1
0.1 0.2 0.3 0.4
HYDROGEN. mi/iOOg
FIRST TEST
tracted and separated from background gas
by a palladium tube. Volume of hydrogen
evolved is determined by measuring pres-
sure increase in a system of known volume.
In the solid extraction method, the sample
is heated under vacuum to a temperature
near but below the melting point. The hy-
drogen evolved is separated and measured
in a manner similar to that used for vacuum
fusion . Since much longer time is required
for hydrogen extraction from the solid sam-
ple, this method is much slower than vacuum
fusion. However, it is greatly preferred for
alloys containing elements such as magne-
sium and zinc which will distill out of the
sample under vacuum and condense in the
apparatus.
When properly performed, the solid ex-
traction method is accurate and is used
frequently to calibrate other methods of
hydrogen determination.
Telegas
The Telegas, developed by Ransley, Talbot,
and Barlow 16, 17 is the only method for abso-
lute determination of hydrogen content of
molten aluminum that can be used on the
plant floor under production conditions. In
this instrument, a small volume of nitrogen
is recirculated through molten aluminum
until the hydrogen, which diffuses into the
nitrogen bubbles, is in equilibrium with that
in the metal. The partial pressure of hydro-
gen in the nitrogen is then equal to that in
the metal and is determined by measuring
the thermal conductivity of the gas mixture.
Actual hydrogen content is found from a
calibration chart which relates meter read-
ing, hydrogen solubility as affected by
temperature, and hydrogen content in
Fig. 7-Comparison of duplicate Telegas hydrogen
determinations made with the same instrument
0 .7 ,...-- , - -. , - - - , - - - - - . - - - , . - - -......,...---,.
0 .6
0 .5
C!'
o
....WE
~
0.4
ZW
88
We>:
0.3
,..
"'0
~
0.2
0 .1
0 .1 0 .2 0 .3 0.4 0.5 0.6 0 .7
0.5 0.6 0 .7
HYDROGEN. mlliOOg
FIRST TEST
milliliters of hydrogen per 100 grams of Telegas can be
aluminum. The chart, supplied with the in- used on the
strument, is based on pure aluminum. Be- plant Hoor under
cause partial pressure of gas in a liquid is operating
a function of solubility, which varies with conditions.
alloy, it is necessary to apply a correction
factor to the results when the instrument is
used for alloys. This correction factor is the
ratio of solubility of hydrogen in the alloy to
that in pure aluminum.
Telegas is a very useful instrument, al-
though, like any precise apparatus, it must
be handled with care. It is capable of great
precision. Duplicate results with the same
instrument are compared in Figure 7. Pre-
cision between different Telegas instruments
is also very good, as shown in Figure 8. The
satisfactory agreement between Telegas and
the solid extraction and submolten extraction
methods is shown in Figures 9 and 10.
Standard deviations for duplicate samples
with the same Telegas instrument, with
different Telegas instruments, and between
Telegas and solid and submolten extraction
determinations are listed in the table.
DISCUSSION
All the methods described are useful, de-
pending on circumstances and the type of
data required. The Straube-Pfeiffer test
provides an indication of metal cleanliness,
and with clean metal, has a sensitivity to
hydrogen of about 0.3 mlj100 grams. The
Alcoa Sensitized Vacuum Density Test,
again with clean metal, extends this sensi-
tivity to 0.1 m1/100 grams. Both methods are
suitable for routine plant control use. The
initial bubble test can give good results un-
der many circumstances, but is subject to
the same conditions of insensitivity as the
OCTOBER 1973, JOURNAL OF METALS-49
 Fig. 8-Comparison of duplicate Telegas hydrogen
determinations made with different instruments
0.7 r-- , - --,-----,--- - , - - r - - , - -71
Comparison of various methods of
0 .6 hydrogen determination

Standard
1-", 0.5 Deviation
%0
Wo No. ml/IOO g,
:::. - of Tests STP
::> ::::
a::E 0.4
I- Duplicate detenninations, same
II>z Telegas instrument 44 0.014
~w
Duplicate detenninations, diHerent
00'"
zc::
00
0.3 Telegas instruments
Solid extraction versus Telegas
43
21
0.013
0.028
U)-
WJ: Submolten extraction versus Telegas 11 0.019
II>
0.2

0. 1 Straube-Pfeiffer test. With inclusion-free

metal, difficulty is encountered in getting
0 .1 0 .2 0 .3 0.4 0.5 0 .6 0.7
bubbles to form at moderate to low hydro-
HYDROGEN. mI/IOO; gen contents. Vacuum fusion and solid
FIRST INSTRUMENT extraction methods are limited to laboratory
use and are most useful for determining hy-
Fig. 9-Comparison of Telegas and solid extrac- drogen in samples from solid metal, such as
tion results castings or wrought products, and for cali-
brating other methods of hydrogen deter-
0.7 r---,-- -r---,---r---.----,,--.,,
mination. The long time required at reduced
pressure precludes their use for control
0
0 '" 0.6
purposes. Telegas can be used on plant Hoor
"
E
1--
0 .5
and provides a quantitative determination
%
W
I-
of the hydrogen content of a melt which Is
%
0
U 0.4
not affected by the relative cleanliness of
%
W
the metal with respect to inclusions. It pro-
'"a::
0
0 .3 vides a means of following changes in gas
0
)-
J:
content of a melt as determinations can be
made in less than five minutes.
'"'" 0.2
'"--'w
W
Extremely useful in metallurgical inves-
I- 0. 1
tigations, it can be used as a control instru-
ment when the highest quality material is
0.7 being produced. Telegas is rather fragile,
HYDROGEN CONTENT .ml/IOOg however, and must be handled with the care
SOLID EXTRAC TION
usually afforded a precise instrument. 0
Fig. 10-Comparison of Telegas and SUb-molten
extraction results
0. 7r--.---.--.-----,------r---.--~
REFERENCES
M. Bosshard. H. Hug and E. Hansler, Aluminium,
0'"
1

0
0 .6 (November, 1937), Vol. 19, pp. 699-704

"E...... 0 .5
2 B. R. Deoras and V. Kondies, Foundry Trade Journal,
(May 24, 1956), Vol. 100, pp. 361-364, 366
3 L. Moreau, Bull. Assoc. Tech. Fonderie, (1936), Vol.
z 10 (12), pp. 446-447
w
I- • H. Rosenthal and S. Lipson, Transactions of the Ameri-
% can Foundrymen's Society, (1955), Vol. 63, pp. 301-305
0
u 0.4 • Rev. Aluminium, (1948), Vol. 150, p. 374
z • K. J. Brondyke and P. D. Hess, Transactions, Metallurgi-
W cal SOciety of AIME, December, 1964, p. 1542
'"0c:: 0 .3
7 K. J. Brondyke and P. D . Hess, Transactions, Metal-
lurgical Society of AIME, December, 1964, p. 1553
0
)- • U.S. Patent No. 2,863,558
J: 9 U.S. Patent No. 3,281,238
II> 0 .2 10 U.S. Patent No. 3,039,864

'"'"--'
11 U.S. Patent No. 3,172,757
W 12 Y. Dardel, AIME, Inst. of Metals Div., Metals Tech-
W nology 15, No.8, Tech. Pub. No. 2484, (1948)
I- 0 .1 ,. U.S. Patent No. 3,517,543
.. J. L. Brandt and C. N. Cochran, JOURNAL OF
METALS, December, 1956, Vol. 8, No. 12, pp. 1672-1674
15 C. E. Ransley and D. E. J. Talbot, Journal of the In-
0 .1 0.2 0 .3 0.4 0 .5 0.6 0.7 stitute of Metals, 1955-56, 84, p. 445
18 C. E. Ransley, D. E. J. Talbot, and H. C. Barlow,
HYDROGEN CONTENT. mIIlOO; Journal of the Institute of Metals, 1957-58, 86, pp. 212-219
SUB - MOLTEN EXTRACTION 17 U.S. Patent 2,861,450

SO-JOURNAL OF METALS, OCTOBER 1973