Sie sind auf Seite 1von 3

APPLIED PHYSICS LETTERS 86, 041901 2005

Physical properties of liquid and undercooled tungsten by levitation techniques


Paul-Franois Paradisa and Takehiko Ishikawa
Japan Aerospace Exploration Agency, 2-1-1 Sengen, Tsukuba, Ibaraki, Japan

Ryuichi Fujii
Chiba Institute of Technology, Tsudanuma, Narashino, Chiba, Japan

Shinichi Yoda
Japan Aerospace Exploration Agency, 2-1-1 Sengen, Tsukuba, Ibaraki, Japan

Received 15 September 2004; accepted 23 November 2004; published online 18 January 2005 Maintaining deep undercooling melts represents a formidable challenge when dealing with tungsten due to its high vapor pressure, its melting temperature Tm = 3695 K , and the risk of contamination. Using electrostatic levitation, properties of liquid tungsten were measured above the melting temperature as well as in the undercooled phase. Over the 3125 3710 K interval, the density was measured as T = 1.67 104 1.08 T Tm kg m3. Similarly, the surface tension was measured as T = 2.478 103 0.31 T Tm over the 3360 3700 K range. At Tm, the data agree well with the literature values. The excellent processing conditions also offer opportunities to achieve reproducible and controlled formation of metastable phases. 2005 American Institute of Physics. DOI: 10.1063/1.1853513 Tungsten plays a vital role in geology as it helps to time the formation of telluric planet core.1 On a more applied aspect, its ductility and high tensile strength at high temperature explain why it has been used for a long time in metallurgy, in particular as laments in light bulbs. Recent advances in tungsten photonic lattice technology also show promise in low-energy consumption light bulbs.2 Measurement of the physical properties of liquid tungsten is challenging due its high vapor pressure, its melting temperature 3695 K ,3 the highest among metals, and the associated risks of contamination. Despite the fact that it has been known for more than two centuries,3 density and surface tension data are scarce and limited at its melting temperature. In discussing the nature and behavior of liquid metals, the data of these properties are essential to understand metallurgical processes. In particular, density is needed to calculate mass balance in rening operations and thermal natural convection in furnaces,4 to model planetary core composition, magnetism, and convections,5 and to describe the radial distribution function.6 In addition, knowledge of volume changes at the melting point and temperature dependence are of critical importance to understand solidication processes. Knowledge of the surface tension is equally vital in smelting and rening operations where it dominates in gas adsorption, nucleation of bubbles, nucleation, and growth of nonmetallic inclusions, and metal/slag reactions.4 To obtain a drop of liquid tungsten above or below the melting point adequate for property measurements is a redoubtable technical task. Wouch et al.7 succeeded using an electromagnetic levitator but did not perform any measurements. The highly oblate and unsymmetrical shape of their sample would make density and surface tension analysis rather difcult. The electrostatic levitation technique pioneered by Rhim et al.8 has been successfully used to handle and maintain deep undercooled metallic millimeter-size
a

droplets of materials up to zirconium Tm = 2128 K while performing property measurements. However, their electrode design and heating conguration could not offer sample position stability for more refractory and higher vapor pressure metal samples e.g., Re, Ta, and W . Here, two horizontal electrodes, a concave bottom electrode 30 mm diameter and a at top electrode 10 mm diameter with a through hole Fig. 1 , improved sample vertical position control as well as horizontal position control restoring force towards the center provided by the conical electrical eld distribution . In addition, a attened tetrahedral laser heating conguration was implemented to counter photon induced and evaporation forces Fig. 1 .9 This, together with four secondary electrodes at the height of a levitated sample, solved sample instability problems. A feedback control system and an 12 kV eld between the main electrodes ensured levitation of 2 mm diameter sample charged by thermionic emission in vacuum 105 Pa .10 Spinning the sample with rotating magnetic elds prior to melting provided uniform heating. The radiance temperature was measured with two singlecolor pyrometers 0.96 and 0.98 m working wavelength, 10 and 120 Hz acquisition rates, 50 K covering a

Electronic mail: paradis.paulfrancois@jaxa.jp

FIG. 1. Electrode arrangement and heating conguration three 50 W focused CO2 laser beams 10.6 m in a same plane separated by 120 and one 500 W focused Nd:YAG laser beam 1.064 m directly from the top .

0003-6951/2005/86 4 /041901/3/$22.50 86, 041901-1 2005 American Institute of Physics Downloaded 04 Sep 2009 to 129.186.151.225. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp

041901-2

Paradis et al.

Appl. Phys. Lett. 86, 041901 2005

FIG. 2. Temperature prole for a W sample diameter: 2.10 mm; mass: 84.39 mg showing heating to at least 3820 K, radiative cooling, undercooling 500 K , and solidication.

FIG. 3. Density of W as a function of temperature.

temperature and can be tted with a condence interval of 95% by the relationship: T = 1.67 104 0.023 104 kg m3 , 1

1200 K to 3820 K interval. Calibration to true temperature was obtained with the help of Plancks law after adjusting the melting plateau to correspond to the known melting point of tungsten.3 Figure 2 shows a typical thermal prole for a W sample being laser heated to at least 3820 K upper limit of pyrometers and then radiatively cooled by blocking simultaneously all four laser beams. The sample exhibited an average radiative cooling rate of 1820 K / s, an undercooling of nearly 500 K, and a sudden release of the heat of fusion at crystallization recalescence . Although all processed samples mass from 45 to 90 mg displayed similar undercooling levels, which infers that the homogeneous nucleation occurred, the levels reached were far from those suggested by the classical nucleation theory for homogeneous nucleation to be the triggering event of solidication.11 Oxide or nitride patches were not observed on the undercooled melts but traces in the bulk were revealed by electron probe microanalysis. In addition to impurities present in the starting material, they constitute the most possible cause of heterogeneous nucleation. This is supported by Paine et al. who reported that even a minute amount of oxygen could be favorable to the nucleation of a metastable phase in pure refractory metals.12 The presence of oxygen could also explain why either bcc-fcc or A15-fcc phase transitions were not observed. The undercooled levels were, however, identical with those reported in containerless microgravity experiments carried out with a drop tube when experimental uncertainties were considered.13 Optical microscopy did not show any evidence of cavities or voids on the surface or in the bulk of the samples. Upon melting, a sample took a spherical shape due to surface tension, and its images and its temperature were simultaneously recorded. Magnied images of a sample backlit with an UV source were obtained by a high-resolution charged-coupled-device camera equipped with a telephoto objective.10 After the experiment, the sample radius was extracted from each digitized video images and matched to a temperature prole Fig. 2 . Since the sample was axisymmetric and because its mass was known, the density was found as a function of temperature. Calibration was done by levitating a sphere of precisely known diameter under identical conditions. The density measurements of liquid tungsten 99.95 wt. % purity , taken over the 3125 3710 K temperature range and covering the undercooled region by nearly T , like that of 570 K, are shown in Fig. 3. The density other pure metals, exhibited a linear behavior as a function of

1.08 0.08 T Tm

where Tm is the melting temperature 3695 K . In these experiments, the uncertainty was estimated to be less than 2% from the resolution of the video grabbing capability 640 480 pixels and from the uncertainty in mass measurement 0.0001 g . These measurements cover the undercooled region. From these data, the volume expansion coefcient was calculated as 6.5 105 K1. The values that appeared in the literature were also plotted in Fig. 3 for comparison. At the melting point, our density datum was, within respective experimental uncertainties, identical to that measured by Seydel et al.14 with the isobaric expansion method, 2.7% higher than that measured by Shaner et al.15 with the isobaric expansion technique, and 3.1% higher than that obtained by Vinet et al.16 with the pendant drop method in vacuum. It was 4.6% lower than that calculated by Allen,17 5.1% lower than that estimated by Calverley,18 and 5.7% lower than that measured with the drop volume by Pekarev.19 Our temperature coefcient was nearly 1.38 times larger than that calculated by Steinberg20 and 1.2 times larger compared with that of Pekarev.19 Because levitation and heating were uncoupled, it was possible to maintain deep undercooled melts. Taking advantage of the sample stability, the surface tension was determined by the oscillation drop technique.2123 In this method, a mode-2 drop oscillation was induced to the molten sample by superimposing a small sinusoidal electric eld on the levitation eld. Deliberate sample rotation ensure that excitation of modes other than mode-2 was suppressed. The transient signal that followed the termination of the excitation eld was detected and analyzed. This was done several times at a given temperature and repeated over a large range. Clear drop oscillations could be obtained only on fully molten samples. Using the characteristic oscillation frequency c of the signal after correcting for nonuniform surface charge distribution,24 the surface tension can be found from:22
2 c

= 8 /r3 Y , o

Downloaded 04 Sep 2009 to 129.186.151.225. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp

where ro is the radius of the sample, is the density, and Y is a correction factor depending on the drop charge, the permittivity of vacuum, and the applied electric eld. Real-time values of the radius and density data were used in Eq. 2 to prevent distortion due to sample evaporation. The measure-

041901-3

Paradis et al.

Appl. Phys. Lett. 86, 041901 2005

FIG. 4. Surface tension of W as a function of temperature.

ments, taken over the 3360 3700 K temperature range and covering the undercooled region by more than 330 K, are shown in Fig. 4 along with the literature data. The surface tension exhibited a linear behavior as a function of temperature and can be tted with a condence interval of 95% by the relationship: T = 2.48 103 0.28 103 mN m1 . 3

the samples, and level of vacuum. Ways to implement VUV and x-ray imaging systems to improve contrast of samples at temperatures above 3700 K are currently being sought. Temperature measurements too are subject to errors. Besides the fact that the pyrometer range was limited to 3820 K, recent measurements on W with a fast polarimeter revealed that over the 3207 4400 K range, the emissivity decreased from about 0.396 to 0.379.28 This implies that the highest attained temperature in our thermal curves would have been underestimated. No emissivity data for undercooled W have been reported yet but such a trend is plausible. In conclusions, the density and surface tension of liquid W were measured with a vacuum electrostatic levitation furnace above as well as below the melting temperature. Analysis of the microstructure of the droplets solidied from deep undercooled states and comparison with droplets solidied in drop tubes will also appear elsewhere and will hopefully clarify the role of microgravity in the solidication process.
D.-C. Lee and A. N. Halliday, Nature London 388, 854 1997 . J. G. Fleming, S. Y. Lin, I. El-Kady, R. Biswas, and K. M. Ho, Nature London 417, 52 2002 . 3 D. R. Lide and H. P. R. Frederikse, CRC Handbook of Chemistry and Physics, 78th ed. CRC, Boca Raton, FL, 1997 . 4 T. Iida and R. I. L. Guthrie, The Physical Properties of Liquid Metals Clarendon, Oxford 1988 . 5 J.-P. Poirier, Geophys. J. 92, 99 1988 . 6 M. Shimoji, Liquid Metals Academic, London, UK, 1977 . 7 G. Wouch, E. C. Okress, R. T. Frost, and D. J. Rutecki, Rev. Sci. Instrum. 46, 1122 1975 . 8 W.-K. Rhim, S.-K. Chung, D. Barber, K.-F. Man, G. Gutt, A. A. Rulison, and R. E. Spjut, Rev. Sci. Instrum. 64, 2961 1993 . 9 P.-F. Paradis, T. Ishikawa, and S. Yoda, Appl. Phys. Lett. 83, 4047 2003 . 10 T. Ishikawa, P.-F. Paradis, and S. Yoda, Rev. Sci. Instrum. 72, 2490 2001 . 11 D. Turnbull, J. Appl. Phys. 21, 1022 1950 . 12 D. C. Paine, J. C. Bravman, and C. Y. Yang, Appl. Phys. Lett. 50, 498 1987 . 13 B. Vinet, L. Cortella, and J. J. Favier, Appl. Phys. Lett. 58, 97 1991 . 14 U. Seydel and W. J. Kitzel, J. Phys. F: Met. Phys. 9, L153 1979 . 15 J. W. Shaner, G. P. Gathers, and C. Minichino, High Temp. - High Press. 8, 425 1976 . 16 B. Vinet, J.-P. Garandet, and L. Cortella, J. Appl. Phys. 73, 3830 2002 . 17 B. C. Allen, Trans. Soc. Min. Eng. AIME 227, 1175 1963 . 18 A. Calverley, Proc. Phys. Soc. 70, 1040 1957 . 19 V. Pekarev, Izv. Vys. Uch. Sav. Tsvetn. Met. USSR 6, 111 1963 . 20 D. J. Steinberg, Metall. Trans. 5, 1341 1974 . 21 Lord Rayleigh, Proc. R. Soc. London 14, 184 1882 . 22 W.-K. Rhim, K. Ohsaka, P.-F. Paradis, and R. E. Spjut, Rev. Sci. Instrum. 70, 2796 1999 . 23 S. Sauerland, G. Lohofer, and I. Egry, J. Non-Cryst. Solids 156158, 833 1993 . 24 J. Q. Feng and K. V. Beard, Proc. R. Soc. London, Ser. A 430, 133 1990 . 25 P. S. Martsenyuk, Yu. N. Ivaschenko, and V. N. Eremenko, Tep. Vys. Temp. USSR 12, 1310 1974 . 26 A. D. Agaev, Dissertation, Moscow Steel and Alloys Institute, 1973. 27 O. Flint, J. Nucl. Mater. 16, 260 1965 . 28 C. Cagran, C. Brunner, A. Seifter, and G. Pottlacher, High Temp. - High Press. 34, 669 2002 .
2 1

0.31 0.08 T Tm

The uncertainty of the surface tension measurements was estimated to be better than 5% from the response of the oscillation detector and from the density measurements. These measurements cover a large temperature range. Compared with the measurements obtained by several investigators with the pendant drop method, our surface tension datum at the Tm agrees generally well. When the experimental uncertainties were considered, our datum at the melting temperature was identical to that measured by Allen17 and systematically higher than those reported by Martsenyuk et al. 6.95% ,25 Vinet et al. 7.2% ,16 Calverley 7.2% ,18 Agaev 7.7% ,26 and Flint 24% .27 Our measured temperature coefcient is less than 7% higher than that estimated by Allen.17 The discrepancies between our density and surface tension results and those of other investigators can be attributed to the difference in processing techniques. In this work, levitation method in high vacuum isolated the samples from container walls and gases whereas other authors employed the pendant drop or drop weight methods for which chemical reactions between the highly reactive molten metal and a support could have occurred and contaminated the samples. This would explain that their surface tension results are systematically lower although our data are possibly lower than the true values due to the presence of oxygen. Additional sources of error include sample imaging, image digitizing, sample mass measurements, material purity, gas solubility in

Downloaded 04 Sep 2009 to 129.186.151.225. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp

Das könnte Ihnen auch gefallen