Gas Fluxing of Molten Aluminum: An Overview and Update

Geoffrey K. Sigworth

Abstract species. From the law of mass action the pressure of gas in
equilibrium with a melt follows the relationship:
The control of dissolved gas in liquid metal is absolutely
H
vital for the production of high quality castings, and for P  (2)
this reason there has been a continual evolution in S C
degassing equipment over the years. A brief review and where P is the pressure of hydrogen vapor in atmospheres
theoretical analysis is given of the chemical and kinetic and H is the amount of gas dissolved in the melt, usually
factors which control metal quality after gas fluxing, and given in standard cc/100 g.1 The solubility, S, is the gas
the evolution of degassing technology is summarized content in pure aluminum at a hydrogen gas pressure of
briefly. An electrochemical probe has been developed one atmosphere. It is given by the equation [5]:
recently, which measures hydrogen in situ. With this log S = -2692/T + 2.726 (3)
equipment it is possible for the first time to measure
where T is the thermodynamic temperature (in degrees
hydrogen continuously during degassing. These
Kelvin). At 1000 K (727 C) the hydrogen solubility is
measurements show that our understanding of the process
1.081 cc/100g.
is incomplete. It is necessary to account for the gas pickup
from moisture in the atmosphere. A simple model is
The solubility of hydrogen in solid and liquid aluminum is
proposed as a first step to ‘plug the gap’ in our
plotted in Figure 1.
understanding. The results suggest that future
improvements in degassing will require shielding of the Temperature (degrees C)
melt. 600 700 800 900
10
Introduction
Aluminum is an extremely reactive metal. When it comes Liquid
in contact with moist air the water vapor decomposes and
releases hydrogen into the melt. It has been well
documented that this dissolved gas has a detrimental effect
Hydrogen Solubility

1
on the mechanical properties of aluminum castings [1, 2].
In order to produce good quality castings, one must degas
(cc/100g)

before casting.

At first glance, the scientific basis for gas removal appears

to be well established [3-5]. Any degassing process is 0.1
subject to two fundamental laws. The first is the
thermodynamic relationship between gas dissolved in Solid
metal and the pressure of hydrogen in purge gas. The
second concerns the rate of hydrogen diffusion into
ascending gas bubbles. Both must be considered.
0.01
Thermodynamic Considerations 1100 1300 1500 1700
Temperature (degrees F)
Experiments have shown that only hydrogen gas dissolves Fig.1 Hydrogen solubility in pure aluminum
in molten aluminum. The gas solubility has been measured
carefully in a number of studies. These results were An alloy correction factor, C, is employed to account for
reviewed and summarized in [5]. The solubility follows the change in gas solubility in any particular alloy.
Sievert's law. That is,
1/2 H2 (v) = H (in Al) (1) 1
The gas content is also occasionally reported in parts per
The diatomic hydrogen molecule decomposes to form two million by weight (ppm). It is easily shown that one cc /100
atoms, which dissolve in molten metal as a monatomic grams is equal to 0.892 ppm, or 8.92 x 10-5 weight percent.

1
Equations are given in [5] which may be used to calculate In other words, 514 alloy will require almost five times as
correction factors for any alloy composition. Table 1 lists much inert gas to obtain the same gas removal as in A356.
values for the correction factor for some common casting
alloys. 500

Table 1. Correction Factors for Common Alloys

Gas Removal Ratio

400
Alloy 206 319 332 354 355 C355
C 0.73 0.55 0.59 0.69 0.76 0.80
Alloy A356 A357 359 A360 380 413
C 0.80 0.83 0.80 0.65 0.45 0.54 300
Alloy 444 514 712 100 150 170
C 0.73 1.74 0.87 0.89 0.95 0.98
200
It should be noted that the solubility in equation (3) is an
exponential function of temperature. The gas solubility (S)
doubles for each 110C (200F) increase in temperature. 100
This is why gas problems become much worse at higher
furnace temperatures.
0
Another important conclusion may be drawn from equation 0 0.1 0.2 0.3 0.4
(2). Our ability to degas can be no better than Gas Content (cc/100g)
thermodynamic laws permit. We admit purge gas bubbles Fig. 2 Gas Removal Ratio in Aluminum at 750 C
into the melt. They collect hydrogen as they float towards
the surface. The best possible situation is when bubbles are
saturated with hydrogen gas as they leave the surface of the Kinetic Considerations
melt. In this case, the degassing process efficiency is 100%
from the thermodynamic point of view. We now consider the kinetics of hydrogen removal in the
melt. This subject was studied in detail by Sigworth and
However, as the gas content drops so does the pressure of Engh [3]. They considered bubbles floating up towards the
hydrogen in the bubbles. This means more purge gas is surface and developed equations for gas removal. A
required to remove a specific volume of hydrogen from the 'dimensionless hydrogen concentration' was used to
metal. This is best seen by defining a gas removal ratio, R, characterize the approach to thermodynamic equilibrium.
which is the volume of inert purge gas needed to remove The practical implications of their result are best seen by
one liter of hydrogen from the metal. At one atmosphere an example calculation, where the degassing efficiency in a
pressure the removal ratio is given by the relation: 250 kg (550 lb.) crucible furnace is shown as a function of
2
1 P  S C  bubble diameter. For this case the bubbles ascend through
R   1 (4) approximately 0.7 meter (27”) of metal. The results are
P  H 
shown in Figure 2. Note that a high degassing efficiency
The gas removal ratio, R, is shown in Figure 2 for pure (more than 80 %) occurs when the bubble size is less than
aluminum at 750C. It can be seen that as the gas content 2 to 3 millimeters in diameter.2 In other words, very small
drops below 0.1 cc/100g, the ratio R becomes more than bubbles will contain close to 100% of the hydrogen gas
200. This behavior limits our ability to degas to extremely that thermodynamic laws permit.
low levels.
When one halves the bubble size, the surface area of
As noted above, the solubility increases exponentially with bubbles increases by a factor of four. This increases the
temperature, but (S)2 appears in equation 4. In other words, rate of hydrogen delivery to the bubble, since it is
a temperature increase of 110C doubles the solubility, S, proportional to surface area. But the rise velocity of the
but the gas removal ratio is increased by four times. Thus, bubble also decreases. The smaller bubbles take longer to
keeping metal temperature low is even more important for
degassing efficiency. 2
A similar calculation was made in reference [4]. Unfortunately,
the earlier calculations were not correct. The mathematics is
The square of the alloy correction factor, C, also appears in
rather complex, and the mistake was not discovered until
equation (4). If we compare the high magnesium alloy 514 recently, when the relevant equations were placed into an Excel
with A356, we find the ratio of solubilities for the two program. This program also contains the more recent degassing
alloys is 1.74/0.80 or 2.18. Squared this ratio becomes 4.8. models discussed in this paper, and is available upon request
from the author.

2
 float out, and are in contact with liquid metal a longer time. splashing, vortexing, or a gas/metal emulsion are produced.
So, they have more time to reach thermodynamic Various design solutions have been placed into service,
equilibrium. The two effects are additive, and process either singly or in combination:
efficiency increases sharply as bubbles become smaller. -using baffle plates
-placing the rotor off center in the ladle/crucible
100 -using 'pumping' heads, to direct metal flow
-using rotor designs, having a high localized shear
where gas enters the melt
Degassing Efficiency (%)

80
-reverse rotation of the rotor assembly
60
0.1 cc/100 gm In the last method the rotation of the rotor shaft is reversed
periodically at 5-10 second intervals. In this way the
rotational momentum introduced into the melt during one
40
cycle is canceled by the following reverse rotation.
20
0.3 cc/100 gm From Figure 3 we understand that any process which
produces small bubbles will yield a high degassing
0 efficiency. Unfortunately, not all equipment using rotating
0 1 2 3 4 5 6 7 8 9 10
Bubble Diameter (mm)
impeller heads give equivalent results. Some companies
have focused more on the cost of machining the head, and
Fig. 3 Degassing Efficiency vs. Bubble Size have unknowingly sacrificed bubble size, and process
(Results are shown for two different gas contents) efficiency.

This fact should be communicated to your operators in the It is beyond the scope of this paper to offer a history of
cast shop, and procedures established to inspect the bubble rotor head design, or to present a detailed comparison of
pattern on a routine basis. If large bubbles are observed at the different heads which have been used. Nor are we
the surface, actions need to be taken to restore degassing prepared to explain the hydrodynamic theory related to the
efficiency. (Checking the inert gas flow rates, replacing subject. For the reader who wishes to pursue this area of
worn impeller heads, etc.) technology in more depth, there are excellent publications
which present a great deal of information [7-11].
The importance of having small bubbles cannot be
overemphasized, since it has been the essential impetus In spite of the fact that many degassing systems do not
behind the commercial development of improved operate at highest efficiency, it will be convenient to
degassing processes during the last fifty years. If one consider the case where the degassing occurs close to
places a lance into a melt, the bubbles produced are several thermodynamic equilibrium [5]. This simplifies greatly the
centimeters in diameter and give a thermodynamic equations that describe gas removal.
efficiency of only 10 or 20%. The use of a porous plug
improves the situation greatly compared to a simple lance, For a batch degassing process, the hydrogen content in the
but even in this case the efficiency is typically only about melt as a function of time is given by the relation:
30 or 40%. The next significant improvement came sixty
1 1 V (5)
years ago, when Szekely used mechanical stirring in the   t
SNIF (Spinning Nozzle Inert Flotation) process [6]. Not H H o 10 M CS 2
only were fine bubbles produced by this means, but they where the inert gas flow ( V ) is given in cc/sec, the mass of
were dispersed throughout the melt, instead of rising in a metal in the crucible or furnace (M) is given in kg, the
plume (as with a porous plug). The original SNIF process solubility (S) is given in cc/100g, and the time (t) is given
worked extremely well, but the impeller was complex and in seconds. The instantaneous hydrogen concentration (H)
costly, being a modified flotation cell machined out of and the starting original hydrogen content (Ho) are also
graphite and immersed in liquid metal. In subsequent years, given in cc/100g.
many companies have developed more simple means of
producing the desired fine bubbles. Note the factor of S2 in equation (5). This means that the
rate of gas removal will be sensitive to metal temperature.
Other things being equal, the degassing performance of a It will also be dependent on alloy composition, from the
rotor head improves when the stirring energy is increased. factor C2.
Of course, as the stirring intensity increases, one has to
deal somehow with the increased amount of rotational
energy placed into the metal. Otherwise a great deal of

3
It will be useful to present an example calculation using The first iteration made was to consider the effect of
equation (5) for a 2,000 pound (900 kg) melt of A357 bubble size. Purge gas bubbles are often (perhaps usually)
alloy. The inert gas flow rate is assumed to be 20 SCFH larger than the 2-3 millimeters which produce a high gas
(8.4 l/min). The calculations were made for temperatures removal efficiency. Equations given in [3], used to
of 1450F (788C) and 1350F (732C). The gas removal plots calculate the curves shown in Figure 3, were placed in an
are shown in Figure 4. It can be seen that the 100F (56C) Excel program, using subroutines created with Visual
increase in temperature doubles the time needed for Basic for Applications (VBA). An example of the results
degassing. If a final gas content of 0.1 cc/100 grams is obtained is shown in Figure 5. This shows the predicted
desired, then the degassing times at the two temperatures degassing of a crucible of 354 alloy held at a temperature
are approximately four and eight minutes (dashed lines in of 1350F. The blue curve is for very small bubbles, or
the figure). The importance of keeping metal temperature 100% degassing efficiency. The red curve is for the case
low is easily seen from this example. where the average bubble size is 5 mm. This result
suggests that even with somewhat larger bubbles, a
degassing time of 20-30 minutes should produce the
desired low gas content (of 0.1 cc/100g).

0.3

0.25

Gas Content (cc/100 gm)

0.2

0.15

0.1

0.05

0
Fig. 4 Gas removal in A357 alloy at two temperatures 0 5 10 15 20 25
Degas Time (min)

Theory--100% Efficiency Actual

Gaps in Understanding
Fig. 5 Calculated Gas removal in 354 alloy for two
The information presented above is important for an bubble sizes
understanding the degassing process. Unfortunately, the
picture is not complete. This gap in our understanding did
not become apparent until better methods of gas analysis However, gas measurements after a 30 minute treatment in
became available [12]. When one is using the reduced recent trials were found to be 0.13 and 0.15 cc/100g. After
pressure test, it is easy to compare the ‘before’ and ‘after’ some reflection it appears that in many cases one reaches a
gas samples with the curves shown in Figure 4, and steady state, where gas is removed as fast as it enters from
conclude that our understanding is complete. That is what I the atmosphere. In other words, an additional iteration is
did for almost thirty years. required in our model of the degassing process. It is
necessary to account for the reaction with water vapor in
However, more recent gas measurements show that the gas the atmosphere.
contents predicted by Equation (5) are seldom obtained; or
if they are, much longer degassing times are required. Reaction with Atmosphere

Creating models for metallurgical processes is often a As far as I am aware, the only published study of the
stepwise, iterative process. One creates a model and problem at hand is the 1984 paper by Engh and Pederson
compares its predictions to actual practice. When a [13]. Unfortunately, the complete model was not presented
difference is noted, the model is then revised. in the paper, and the experimental data was insufficient to
extract the needed rate constants. In other words, this paper

4
does not allow us to model the surface reaction. It does, aluminum oxide is much more stable than hydrogen oxide
however, point to the importance of the reaction at the (water). At some distance from the metal surface, the water
surface. Hence, a simple model is proposed here as a first vapor pressure will be that given by the temperature and
step to further our understanding. relative humidity in air near the furnace. This pressure is
represented by the symbol PH2O.
Two reactions occur at the gas/metal interface. Hydrogen is
removed by the reverse of reaction (1); and hydrogen gas It is now possible to consider the case of ‘natural’
may be added by: degassing. The flux of hydrogen into the molten metal may
1/2 H2O(v) = H (in Al) + 1/6 Al2O3 (6) be represented by the equation:
Both processes must be considered. J H  2 J H 2O  2k H 2O AS PH 2O  0  (7)
where
At first glance, this seems counter intuitive, but I remember
JH is the flux of hydrogen (moles/sec)
reading old papers many years ago, and finding anecdotal
references to ‘natural’ degassing. That is, when a melt is JH2O is the flux of water vapor towards the metal
held a sufficiently long time (perhaps overnight) in an (moles/sec)
electric furnace, one may obtain a reasonably low gas level kH2O is the mass transfer coefficient for water vapor
in dry air. (m/sec-m2-atm)
2
AS is the area of the metal in contact with the air (m )
The situation may be considered as shown in Figure 6. PH2O is the prevailing pressure of water vapor in the
air
gas pressure
In a similar fashion, the flux of hydrogen gas out of the
metal is given by the equation:
liquid gas above metal J H  2 J H 2  2k H 2 AS PH 2  0 (8)
metal
where
JH is the flux of hydrogen (moles/sec)
JH2 is the flux of water vapor towards the metal
(moles/sec)
water vapor kH2 is the mass transfer coefficient for water vapor
(m/sec-m2-atm)
2
AS is the area of the metal in contact with the air (m )
PH2 is the prevailing pressure of hydrogen gas at the
hydrogen gas surface of the melt

At equilibrium the hydrogen will leave the melt at the same

rate as hydrogen is picked up from water vapor in the air.
In other words:
distance from metal surface
2kH2AS(PH2) = 2kH2OAS(PH2O) (9)
Fig. 6 Gas pressures above a melt of aluminum The factor of 2AS cancels out, which means:
PH 2O k (10)
At the surface of the metal, a hydrogen gas pressure arises  H 2  Rk
according to the chemical equilibrium given in equations PH 2 k H 2O
(2) and (3). This pressure is represented by the symbol PH2. For temperatures between 40F and 115F, the pressure of
This gas diffuses through a boundary layer in the gas phase water vapor in air at 100% relative humidity is represented
near the metal surface. At some distance from the metal,
reasonably well by the equation:
the hydrogen pressure becomes zero (the value found
s
normally in air). The variation in hydrogen pressure is log (P H2O) = -2.645 + 0.0141 T (11)
shown schematically in Figure 6 by the red curve.
The prevailing pressure of water vapor in the air (PH2O)
The situation is the reverse for the pressure of water vapor will be the saturated value times the relative humidity.
in the air. At the surface of the metal, the water vapor
pressure will be very small, close to zero. This is because

5
 It is now necessary to consider the ratio of the two mass holding times will be required. Also, when one melts with
transfer coefficients, Rk. The conditions in the boundary gas-fired furnaces, water vapor in the combustion gases
layer will be the same for both gases. However, the will gas the metal.
hydrogen molecule is lighter and smaller, so it diffuses
about 2.8 times faster in air. From theoretical Are the above results science, or science fiction?
considerations [14], the ratio Rk should be approximately Unfortunately, we cannot know the answer without
equal to the square root of the ratio of the diffusivities, or conducting experiments. In our simple model we ignored
1.7. the role of the oxide film on the melt. For example, in
small laboratory experiments, Sr and Ca additions have
Adding everything up and combining equations, we find been shown to increase the rate of gas pickup from
that the equilibrium gas content during ‘natural’ degassing moisture in the surrounding atmosphere [15]. From my
is: foundry experience, I know that Mg has the same effect.
These factors cannot be predicted from the simple model
PH 2O (12) proposed above, which accounts only for mass transfer in
H  SC
Rk the gas phase boundary layer above the melt.
Figure 7 shows the results of calculations made with eq.
Alcoa researchers conducted studies of their 622 degassing
(12) for pure aluminum.3 The results are plotted for two
process, and in the process reported equilibrium constants
different atmospheric conditions as a function of metal
for the reaction shown in equation (6). Four different alloys
holding temperature (Rk = 1.7).
were studied (2024, 3004, 5182 and 7050), but only one
0.4 temperature studied (1350F). Unfortunately the raw data
were not preserved and presented in the report, so a
detailed analysis is not possible. However, the limited
Gas Content (cc/100g)

results suggest that the ratio Rk may be less than one.

0.3
Experimental trials would be needed to settle the question.
90 F, 80% RH
Gas pickup while degassing
0.2
The effects of the surface reactions have been estimated by
placing the equations for the surface fluxes (equations 7
0.1 and 8) into an Excel program. The equations were solved
using an iterative integration, so the solution is only
approximate. However, considering the other possible
60 F, 50% RH
errors and our limited knowledge, the result is satisfactory.
0
1200 1400 1600 Figure 8 shows a second calculation for the 354 alloy melt
shown in Figure 5. The bottom two curves are for the
Metal Temperature (F) theoretical 100% degassing efficiency, and for the lesser
Fig. 7 Calculated gas contents after ‘natural’ performance produced by bubbles having a 5mm
degassing diameter.; and are the same as shown in Figure 5. The top
(red) curve is for the case where the purge gas bubbles are
Surprisingly low gas contents may be obtained with low 5mm in size, and when there is also gas pickup from
metal holding temperatures, even in the presence of a fair moisture in the atmosphere. The air was assumed to be at
amount of atmospheric water vapor. This occurs because 85F and 50% relative humidity, and the mass transfer
the hydrogen pressure in the melt is much larger at low coefficient kH2O was assumed to be equal to 0.3. This value
temperatures (for a given gas content). And degassing will produces a curve roughly in agreement with (limited)
occur until the hydrogen pressure comes close to the experimental measurements.
pressure of water vapor in the atmosphere. (Eq. 9) These
considerations suggest that natural degassing would be
possible, since the melting point of A356 alloy is about
1150F (620C). It may not be practical, however, since long

3
For alloys, multiply the values for gas content by
correction factor, C.

6
 0.3

0.25

0.2

0.15

0.1

0.05

0
0 5 10 15 20 25 30 35 40 45

Fig. 9 Launder delivery of molten metal

Fig. 8 Calculated Gas removal in 354 alloy for two
bubble sizes; and for the surface reactions.

Discussion

Many years ago I worked on a new method of gas analysis,

which we called a direct pressure measurement (DPM)
technique [17]. Working with engineers from Stahl
Specialty Company, a piece of equipment was built to
measure gas continuously in the dipout well of one of their
furnaces. A similar machine was placed in another foundry,
where an insulated and heated launder provided level-pour
delivery from the melting furnace to the casting furnace
(Fig. 9). The distance of metal transfer was rather long in
this foundry; more than 100 feet (30 m); so the launder was
covered with insulated and heated tops. Perhaps of more Fig. 10 Rotary degasser in launder delivery
importance for this discussion was the use of dry nitrogen,
to blanket and cover the metal in the launder. There was A great deal of work needs to be done to flesh out the ideas
also a rotary degasser near the end of the launder (Fig. 10). offered here, but even so it is possible to consider the
A probe from a DPM unit was placed in the launder after practical implications. During the past 30 years our
the degasser. During a period of nearly two months, industry has focused mainly on RID equipment and head
continuous measurements showed that the hydrogen gas in design; and the use of flux treatments; but there has been
the metal was between 0.06 and 0.08 cc/100g. no work on shielding the melt from moisture in the
atmosphere. We need to start focusing on this aspect of the
Those are the lowest gas levels I have ever seen in my process. This appears to be the best way to make further
lifetime, and I wonder now how important the surface advances: lower gas and higher quality castings. Better
reaction was in that long launder. covers for degassers and the practical use of shield gases
are simple engineering solutions, and may provide
By the way, although the DPM equipment worked, it significant benefit in the foundry.
proved to be temperamental. After several years of
working on it, I abandoned the approach as impractical for In short, I am proposing that we change our mindset, when
use in the foundry. the degassing process is considered.

7
References 17. G.K. Sigworth: "A New Method for Gas Measurement
in Aluminum," AFS Transactions, Vol. 101, pp. 541-
1. P.D. Hess and G.K. Turnbull: "Effects of Hydrogen on 8, (1993).
Properties of Aluminum Alloys," Internat. Conference
on Hydrogen in Metals, Amer. Soc. Metals (1974),
277-298.
2. D.E.J. Talbot: "Effects of Hydrogen in Aluminum,
Magnesium, Copper and Their Alloys," Internat.
Metals Reviews, 20(1975), 166-184.
3. G. K. Sigworth and T. A. Engh: "Chemical and
Kinetic Factors Related to Hydrogen Removal from
Aluminum," Metallurgical Transactions B, 13B
(1982), 447-460.
4. G.K. Sigworth: "A Scientific Basis for Degassing
Aluminum," AFS Transactions, 95(1987), 73-78.
5. G.K. Sigworth: "Gas Fluxing of Molten Aluminum,
Part 1: Hydrogen Removal," Light Metals 1999, 641-
649.
6. A.G. Szekely: "Degassing Molten Metals," U.S. Patent
3,227,546, 1966.
7. M. Nilmani, P.K. Thay and C.J. Simensen: "A
Comparative Study of Impellor Performance," Light
Metals 1992, pp. 939-946.
8. W.C. Eister and W.R. Krumme: "An Evaluation of a
SNIF Unit as an Inclusion Removal and Degassing
Device," Light Metals 1991, 1171-1177.
9. S. T. Johansen et al.: "The Bubble Size and Mass
Transfer Mechanisms in Rotor Stirred Reactors," Light
Metals 1997, 663-666.
10. J.J.J. Chen and J.C. Zhao: "Bubble Distribution in a
Melt Treatment Water Model," Light Metals 1995,
1227-1231.
11. Shinji Nagata: Mixing: Principles and Applications,
(New York, Kodanh Ltd. and John Wiley and Sons,
1975).
12. G.K. Sigworth and B. Began: “Control and
Measurement of Hydrogen in Aluminum,” AFS
Transactions, Vol. 116, paper 12-002 (2012).
13. T. Engh and T. Pedersen: “Removal of Hydrogen from
Molten Aluminum by Gas Purging,” Light Metals, pp.
1329-1344 [1984].
14. T. A. Engh, private communication.
15. K. E. Höner and Z. Youling: “Influence of Ca and Sr
on the Hydrogen Pickup in Aluminum Alloy Melts, by
Example of G-AlSi12,” Giessereiforschung, Vol. 39,
pp. 34-48 (1987).
16. J.G. Stephens: “A Mathematical/Computer Model of
the Hydrogen Removal in the Alcoa 622 Process,”
Internal report of Alcoa Technical Center, February,
1991.