A novel method for the determination of the hydrogen solubility in aluminum and aluminum alloy melts

A novel method for the determination of the hydrogen solubility in aluminum and aluminum alloy melts

Szokefalvi-Nagy, A

The CHAPEL method, which was developed for the direct and continuous determination of the hydrogen activity in aluminum melts by measuring the equilibrium hydrogen pressure, has been modified for the determination of the hydrogen solubility in aluminum and aluminum alloy melts. The change of the hydrogen equilibrium pressure due to addition or removal of a given amount of hydrogen from a melt of known mass yields directly the Sieverts constant. Such experiments provide reliable data only if no additional gas exchange takes place between the melt and the gas phase. In the present work, a quasi-impermeable interface between the melt and the surrounding hydrogen atmosphere has been realized by maintaining the hydrogen pressure above the melt continuously at the level of the hydrogen equilibrium pressure. By this technique, which is performed by a relatively simple experimental setup, fast determination of the hydrogen solubility is possible. The main advantage of this novel method is the fact that it can be applied also for aluminum alloys without protective oxide layer on the surface. Preliminary results on pure aluminum and Al-Cu alloy melts show good agreement with the data obtained by the classical method of Sieverts that is not well suited for routine determinations on a wide range of alloy composition and temperature since it is very time-consuming. The method can also be applied for the investigation of other metal and alloy melts.


Hydrogen solubility data determined by various authors with absolute methods deviate remarkably from each other even for pure aluminum melts.[6,7,8] This inconsistency results from systematic errors of the methods used. After accurate redetermination of the solubility in the temperature range between 943 and 1123 K and the critical assessment of other data available, Talbot and Anyalebechi[9 concluded that the most reliable determination of the solubility was provided by the original method of Sieverts,[10]where the volume of hydrogen absorbed by a metal sample of known mass is measured directly. This method, however, is extremely time-consuming and requires meticulous corrections for experimental errors. Therefore, it is not well suited for fast determination of the Sieverts constant of alloys in a wide range of temperature and composition as required for the direct-reading control methods in the casting practice. Systematic measurements are rare and restricted to narrow ranges of alloy composition. For example, for aluminum alloys, little experimental data are available.[bl Therefore, extrapolations from these few data are often combined with empirical estimations to evaluate the hydrogen content of aluminum alloys when direct-reading control methods are used.[12]

Hydrogen solubility values obtained by the method of isothermal desorption of hydrogen from saturated samples into a low pressure system[8] are significantly lower than those obtained by Sieverts’ method. This may be caused by hydrogen losses during the evacuation process between the absorption and desorption steps. These hydrogen losses can hardly be estimated and corrected, particularly in the case of molten samples, where sudden exhaustion of the atmosphere above the melt can result in bubble formation. In the present work, a method is described that uses the relatively simple experimental setup of the CHAPEL method (cf Section II-B) for the determination of the hydrogen solubility in metal and alloy melts. Independently from the existence of an impermeable surface layer, the permeation of hydrogen through the crucible wall and the melt surface is avoided and cannot influence the accuracy of the solubility measurement as in Sieverts’ method. The method enables a rather fast determination of the hydrogen solubility. Therefore, it is an attractive alternative to provide reliable Sieverts’ constant values for the casting practice.


A. Determination of the Hydrogen Solubility by Sieverts’ Method

The classical method of Sieverts[10]has been widely used for the determination of the solubility of gases in solids.[69 ‘3] The amount of hydrogen absorbed by a degassed metal sample of known mass is determined by the difference of the volumes of hydrogen and an inert gas necessary to attain the same equilibrium pressure at a given temperature. The very low solubility of hydrogen in aluminum causes errors, since a small difference between two relatively large volumes has to be determined. Therefore, large samples are used, which slow down the kinetics of the process. Carefully degassed samples are needed, and thus, extended degassing periods must precede each measuring run. The hydrogen uptake is retarded by the oxide layer covering the sample surface. The difference between the thermal properties of the inert gas used and those of hydrogen and the hydrogen loss due to permeation of hydrogen through the wall of the absorption bulb are also to be considered. Since the lifetime of the absorption bulbs used is also limited, the determination of the solubility as a function of temperature and composition becomes extremely time-consuming.[9]

B. Continuous Determination of the Hydrogen Activity in Melts by the CHAPEL Method

The CHAPEL method (Continuous Hydrogen Activity measurement by Pressure Evaluation in Liquid metals) developed by is a direct-reading method for the continuous determination of the hydrogen activity in melts by measuring the equilibrium hydrogen pressure. An oxidefree surface within the melt is realized by immersing a porous graphite body into the melt. The porous body is permeable for gases but not for the melt. In this artificial bubble, the hydrogen partial pressure in equilibrium with the hydrogen dissolved in the melt is measured. In the case of aluminum, only hydrogen is present in the gas phase, and thus, it suffices to measure the absolute pressure. The sensor cell consists of the porous graphite body and a vacuum gage connected by a gas-tight Al^ sub2^O^ sub 3^ tube. The imental setup of the CHAPEL method[4] is schematically shown in Figure 1.

The volume of the sensor cell can be made small and enables a relatively fast determination of the hydrogen activity in a melt sample without remarkable changes of the concentration. After short evacuation of the sensor cell, the new equilibrium pressure is attained within about 10 to 20 minutes, depending mainly on the surface area of the graphite body. The value of the equilibrium pressure determines the hydrogen concentration in the melt according to Eq. [1].

The CHAPEL method provides direct and continuous control of the hydrogen activity of aluminum melts even under plant floor conditions. It has been used for studies on the kinetics of hydrogen uptake through the oxide skin,[14] for the investigation of the inter-relation between the hydrogen content in the melt and the porosity in aluminum cast alloys,[15] and in a modified arrangement that allows the selective measurement of the hydrogen partial pressure in a gas mixture, for the determination of the hydrogen activity in copper melts[16,17]

The advantage of the CHAPEL technique when compared to other direct-reading methods is that hydrogen can also be added to or withdrawn from a test melt through the sensor cell. Thus, not only continuous control of the hydrogen content but also controlled doping or degassing of a melt is possible.


B. Apparatus

Figure 4 shows the experimental setup of the modified CHAPEL arrangement for the measurement of hydrogen solubility in melts. The A1^ sub 2^O^ sub 3^ crucible containing the test melt of known mass is heated by a resistance furnace in a receiver where the hydrogen pressure can be adjusted. The melt temperature is controlled by a standard PID controller with a sealed thermocouple immersed into the melt. The receiver is relatively large and serves as a pressure reservoir for short periods of time. The sensor cell consists of a small cylindrical porous body of degassed electrographite (1.2cm diameter, 1.5-cm height, and 16 to 18 vol pct porosity) and a vacuum-tight Al^ sub 2^O^ sub 3^ tube, which connects it to a pressure gage. The materials are chemically and thermally stable in the melt. The receiver and the sensor cell can be evacuated and filled with hydrogen gas separately.

The volume of the sensor cell, V^ sub cell^ has been determined in a separate run from the pressure change caused by opening the valve that connects the sensor cell to the evacuated calibrated volume, V^ sub cal^,. In the experiments, test melts with a mass of 80 g were used. With a sensor-cell volume of a few cm,3 readily measurable equilibrium pressures were obtained after rapid evacuation of the cell. About one-fourth of the hydrogen dissolved in the melt was withdrawn by each evacuation step. The hydrogen pressure in the cell and in the receiver is monitored continuously by piezoelectric pressure gages. In the present set of experiments, the hydrogen pressure in the receiver was adjusted manually to the actual value of the hydrogen pressure in the sensor cell by handling the needle valves between the receiver and the hydrogen container or the vacuum pump, respectively. Automation of the pressure adjustment is, however, easily possible.

The advantages of this technique compared to Sieverts’ method[lo] or to the isothermal desorption method[8] are as follows.

(1) The experimental error of the sensor cell volume is smaller than that of the volume difference used in Sieverts’ method.[10]

(2) In the original CHAPEL arrangement, the sensor cell volume must be kept as small as possible to reduce changes in the hydrogen content of small melt volumes. However, if the CHAPEL arrangement is used for solubility determination, the amount of hydrogen transferred to the evacuated sensor cell must produce significant change of the equilibrium pressure.

(3) Uncontrolled hydrogen losses of the melt due to bubble formation cannot occur during the evacuation step, since evacuation takes place below the melt surface through the sensor cell and not by sudden exhaustion of the atmosphere above the melt as in the isothermal desorption method[8]

(4) Sampling of the melt is not necessary; thus, the main difficulties of the extraction methods[‘] are avoided namely, errors caused by quenching or surface contaminations. If the hydrogen pressure in the receiver is properly adjusted to the hydrogen activity in the melt, the permeability of the melt surface does not play any role, because in the absence of an activity gradient as the driving force, no gas exchange can take place.

(5) Once immersed into the melt, the sensor cell can remain there for a long time without damage. Therefore, in contrast to Sieverts’ method, series of measurements can be performed at various temperatures and pressures with the same test melt and volume calibration.

(6) In Sieverts’ method,[10] the total amount of dissolved hydrogen is determined for each data point. This requires a carefully degassed sample and a long absorption period. In the method proposed here, only the change of the hydrogen concentration and the equilibrium pressures at the initial and final states must be known. The knowledge of the actual dissolved hydrogen concentration in the melt is not necessary and the equilibration process is much faster than in Sieverts’ method. The equilibration can be further accelerated by the doping capability of the CHAPEL sensor cell.t4 Accurately determined the Sieverts constants for an alloy in an extended temperature range can be obtained within a few days.


It has been demonstrated that hydrogen exchange between a test melt and the gas atmosphere can be suppressed if the melt is placed in a receiver where the hydrogen pressure is continuously adjusted to the actual hydrogen activity in the melt. The continuous activity control is accomplished by the CHAPEL sensor system. Quantitative changes of the hydrogen content of the test melt can be achieved by addition or removal of hydrogen through the sensor cell.

Since this method is relatively simple and fast, reliable determination of the Sieverts constants for a large number of different alloy compositions and temperatures is feasible. The knowledge of these data are essential for the use of the direct-reading hydrogen analysis techniques Telegas,[3] CHAPEL,[4 and Alcan[5] at the plant floor.

The determination of Ks in several successive runs for pure aluminum and for Al-2 wt pet Cu and Al-4 wt pct Cu alloy melts shows good agreement with the data obtained by means of the precise but time-consuming method of Sieverts.[b]

The modified CHAPEL principle can also be applied to other metal and alloy melts and for other gases; however, depending on the specific thermal and chemical conditions, the materials of the porous body and that of the tube connecting it to the pressure gage may be different. If, unlike in the case of aluminum melts, other gases than hydrogen also can be evolved from the melt, pressure gages must be used that are sensitive to the partial pressure of the gas studied.

Although in the present experiments the hydrogen pressure in the receiver was regulated manually, the construction of an automatic pressure controller with an on-line data evaluation program is rather straightforward. ACKNOWLEDGMENT

One of the authors (LS) thanks the financial support of the DAAD (Bonn, Germany).


1. C.E. Ransley and D.E.J. Talbot: J. Inst. Met., 1955-56, vol. 84, pp. 445-52.

2. P.N. Anyalebechi: Cast Met., 1991, vol. 3, pp. 182-201.

3. C.E. Ransley, D.E.J. Talbot, and H.C. Barlow: J Inst. Met., 195758, vol. 86, pp. 212-19.

4. E. Fromm: Aluminium, 1989, vol. 65, pp. 1240-43.

5. J.-P. Martin, F. Tremblay, and G. Dube: Light Met.1989, AIME, New

York, NY, 1989, pp. 903-12.

6. W.R. Opie and N.J. Grant: Trans. AIME, 1955, vol. 188, pp. 123741.

7. C.E. Ransley and H. Neufeld: J. Inst. Met., 1948, vol. 74, pp. 599620.

8. W. Eichenauer, K. Hattenbach, and A. Pebler: Z. Metallkd., 1961, vol. 52, pp. 682-84.

9. D.E.J. Talbot and P.N. Anyalebechi: Mater. Sci. Technol., 1988, vol. 4, pp. 1-4.

10. A. Sieverts: Z. Phys. Chem., 1914, vol. 88, p. 451.

11. P.N. Anyalebechi, D.E.J. Talbot, and D.A. Granger: Metall. Trans. B,

1988, vol. 19B, pp. 227-32.

12. M. Heckler: Aluminium, 1967, vol. 43, pp. 239-43.

13. A. Sieverts: Z. Metallkd., 1929, vol. 21, pp. 37-46. 14. J. Weigel and E. Fromm: Metall. Trans. B, 1990, vol. 21B, pp. 85560.

15. X.-G. Chen and S. Engler: Giesserei, 1989, vol. 76, pp. 767-74.

16. A. Szokefalvi-Nagy, J. Weigel, L. Stojanova, and E. Fromm: J

Mater. Sci. Technol., 1994, vol. 2, pp. 10-25.

17. A. Szokefalvi-Nagy and E. Fromm: Z. Metallkd., 1994, vol. 85, pp. 185-91.

18. W. Hehn and E. Fromm: Aluminium, 1988, vol. 64, pp. 180-86.

19. W. Eichenauer and J. Markopoulos: Z. Metallkd., 1974, vol. 65, pp. 649-52.

A. SZOKEFALVI-NAGY, Research Scientist, and E. FROMM, Senior Scientist, are with the Max-Planck-Institut fair Metallforschung, D-70174 Stuttgart, Germany. L. STOJANOVA, formerly Visiting Scientist, MaxPlanck-Institut fir Metallforschung, is Senior Scientist, Institute for Metal Science, Bulgarian Academy of Sciences, 1574 Sofia, Bulgaria. Manuscript submitted February 4, 1997.

Copyright Minerals, Metals & Materials Society and ASM International Apr 1998

Provided by ProQuest Information and Learning Company. All rights Reserved