Thermal conductivity of saturated liquid toluene by use of anodized tantalum hot wires at high temperatures

Headnote

Absolute measurements of the thermal conductivity of a distilled and dried sample of toluene near saturation. are reported. The transient hot-wire technique with an anodized tantalum hot wire was used. The thermal conductivities were measured at temperatures from 300 K to 550 K at different applied power levels to assess the uncertainty with which it is possible to measure liquid thermal conductivity over wide temperature ranges with an anodized tantalum wire. The wire resistance versus temperature was monitored throughout the measurements to study the stability of the wire calibration. The relative expanded uncertainty of the resulting data at the level of 2 standard deviations (coverage factor k = 2) is 0.5 % up to 480 K and 1.5 % between 480 K and 550 K, and is limited by drift in the wire calibration at temperatures above 450 K. Significant thermal-radiation effects are observed at the highest temperatures. The radiation-corrected results agree well with data from transient hot-wire measurements with bare platinum hot wires as well as with data derived from thermal diffusivities obtained using light-scattering techniques.

Key words: anodized tantalum wire; liquid; thermal conductivity; thermal radiation; toluene; transient hot-wire.

Accepted: February 4, 2000

Available online: http://www.nist.gov/jres

1. Introduction

Saturated liquid toluene has been widely studied and is recommended by the International Union of Applied Chemistry (IUPAC) as a reference standard for thermal conductivity from 189 K to 360 K [1]. Efforts to extend this temperature range to 553 K have been recently reported by Ramires et al. [2]. The barriers to obtaining reliable high-temperature reference standards for thermal conductivity are a lack of data from multiple experimental techniques, and increased uncertainty due to the effects of thermal radiation. Both transient and steady-state measurement techniques for the determination of thermal conductivity are susceptible to errors due to thermal-radiative heat transfer at high temperatures since temperature gradients are imposed during the measurement, and fluids such as toluene absorb and emit the associated thermal radiation [3-5]. Even though thermal-radiation errors may be present in data from transient and steady-state techniques, agreement between radiation-corrected data from these two different techniques would provide evidence of the accuracy of the data. Unfortunately, the relative uncertainty (at the level of 2 standard deviations) of available steady-state thermal conductivity data that have been corrected for radiation exceeds 1 %, which is desired for the development of reference standards [2]. As a result, only data from a single transient hot-wire instrument [4] were designated as primary data during the development of the previous reference standard by Ramires et al. [2].

Recently, thermal diffusivity has been measured for saturated liquid toluene from 293 K to 523 K using light scattering [6]. These light-scattering data have a relative uncertainty of 2.5 %; and because there are no significant thermal gradients in the sample, thermal radiation errors are not present. The thermal conductivity .l can be calculated from the thermal diffusivity a by using

2. Experimental

The transient hot-wire technique is widely recognized as an accurate method to measure the thermal conductivity and thermal diffusivity of fluids. The present measurements are absolute and require only knowledge of the geometry of the hot wires, the applied power, the resistance-versus-temperature characteristics of the wires, and time. The ideal working equation is based on the heat transfer from an infinitely long line source into an infinite medium. The temperature rise of the fluid at the surface of the wire, where r = ro, is given [9] by

2.1 Hot-Wire Cell

The hot-wire cell used in these measurements was designed for measurements on corrosive solutions at temperatures from 300 K to 550 K at pressures up to 70 MPa. The design of the pressure vessel and temperature control system is the same as for our previous hightemperature cells [4], so it will only be briefly described here. The pressure vessel and the internal components of the hot-wire cell are constructed of nickel alloy UNS-N10276, which is particularly resistant to halides and halide salts. The cell contains long and short hot wires located within long and short cavities with diameters of 9.5 mm. The total volume of the cell and supporting pressure system is relatively small, about 50 cm3, to facilitate measurements on scarce or hazardous materials. There are separate voltage and current leads to each end of each hot wire to eliminate the effects of lead resistance during the calibration process. All electrical connections in the cell are spot welded. In the assembly of the tantalum hot wires (25.4 (mu)m diameter), polytetrafluoroethane-insulated nickel/chromium alloy wire (254 (mu)m diameter) and polyimide-insulated platinum (76 (mu)m diameter) wire were used to make electrical connections inside the pressure system. The connections were welded, then all the bare leads were coated three times with a polyimide/polytetrafluoroethane resin, and the assembly was baked at 550 K for several hours. The baking both cured the polymer resin and annealed the tantalum hot wires from their initial hard-drawn condition. The tantalum hot wires were then anodized in aqueous citric acid with up to 50 V to produce a film of tantalum pentoxide with an estimated thickness of 70 nm. Although the temperature control and pressure systems are rated to 750 K, the upper operating temperature of the present tantalum hot-wire cell assembly is limited to 550 K because of the melting point of polytetrafluoroethane used to electrically insulate the lead wires.

2.2 Sample Purification

The toluene sample used in these measurements was prepared from spectroscopic-grade toluene. The toluene was further purified by distillation over calcium hydride. Calcium hydride reacts irreversibly with any water in the sample to form calcium hydroxide precipitate, which remains in the distillation flask. The principal impurity is benzene, which has a lower boiling temperature, so the initial condensate is discarded. The sample used for measurement is then collected when the inlet to the condenser is stable at the boiling temperature of toluene. The purified sample was analyzed by gas chromatography and found to have less than 50 ng/g of benzene and less than 100 ng/g of water. The sample preparation procedure was the same as in our previous measurements [4] using bare platinum hot wires.

2.3 Wire Calibration

Platinum is preferred for use in resistance thermometry when it is properly annealed and is free from stress because its resistance is very stable for prolonged periods of time. Tantalum has not been used widely for resistance thermometry, so the stability of its resistance must be carefully examined and characterized. The situation is further complicated since the present tantalum wires have been anodized to form a protective layer of tantalum pentoxide. The electrical resistivity of tantalum from 273 K to 1273 K has been shown to increase in proportion to the concentration of oxygen in the sample [14]. Since the concentration of oxygen is not uniform through the wire's cross section, there is a possibility that oxygen from the anodized layer might diffuse into the bulk tantalum wire and alter its resistance. Any oxygen-diffusion process would be enhanced at higher temperatures. The present system is ideally suited for characterizing the anodized tantalum wires since the wire resistance is measured for each wire, along with the temperature from the reference PRT and the pressure from the quartz pressure transducer, during the balance cycle for each measurement. The instrument also maintains a record of the time and date of each measurement to allow examination of the stability of the wire calibration. Since the present measurements were made along the saturation line of toluene, at pressures less than 3.3 MPa, there is not enough pressure range to allow characterization of the pressure dependence. The electrical resistivity of tantalum is known to decrease in proportion to the pressure on the sample [15].

To eliminate uncertainty due to the resistance of the lead wires, there are separate current-supply and voltage-sensing leads to each end of each hot wire. The resistance is measured during the balance cycle by measuring the voltage drop across standard resistors in each leg of the Wheatstone bridge containing the hot wires, together with the voltage drop across each hot wire. The measured value is the average of five readings with a forward current of about 0.3 mA and five readings with the current reversed. This process minimizes uncertainty due to thermoelectric voltages at weld junctions and electrical connectors. The uncertainty of the resistance measurement is about 0.003 omega. The measurements were made first with temperature increasing from 300 K to 550 K. Then, measurements were made at 550 K in the morning and evening for a period of 4 days. Finally, measurements were made with temperature decreasing from 550 K to 300 K so that hysteresis effects could be examined. The electrical resistance of the long (188.08 mm) and short (49.07 mm) hot wires are shown in Fig. I during this temperature cycle. It is apparent in the figure that the resistances of both the long and short hot wires increased with the elapsed time at 550 K. Although the resistance of the long wire increased more than that of the short wire, the increase was not in proportion to the wire lengths, as would be expected if the process were uniform over the entire length of each wire.

However, the increase in resistance during the 4 days at 550 K is quite dramatic and unexpected, based on our previous experience with pure platinum hot wires [4]. With pure platinum the normal behavior is a slight decrease in resistance due to annealing and stress release in the hot wires if the wires have not been at this temperature recently. Figure 2 clearly shows that the increase in resistance was not consistent with wire length. Thus, resistance increases at the welds must be considered a possibility. The only weld locations that can contribute to the measured resistance during a four terminal measurement occur where the ends of the 25 (mu)m diameter tantalum wires are joined to the 254 (mu)m diameter nickel-chromium alloy lead wires. Each hot wire has two welds which could potentially contribute to the measured resistance, but it is not possible to separate contributions due to the welds from those due to changes in the wires. The use of all-tantalum lead wires should be examined in the future to see whether this resistance increases within the welds or within the wires themselves.

Given that the increase in resistance with time is significant at 515 K, the best possible calibration must be determined, and the influence of uncertainty in the Given that the increase in resistance with time is significant at 515 K, the best possible calibration must be determined, and the influence of uncertainty in the calibration on the thermal conductivity results must be assessed. The rise in temperature at any elapsed time during a measurement is obtained from the change in resistance of the hot wires using the derivative of the calibration curve for resistance versus temperature:

Thus, the uncertainty of the wire's temperature rise, and of the measured thermal conductivity, is directly related to the uncertainty of the derivative of wire resistance with respect to temperature. This resistance derivative divided by length is plotted as a function of temperature in Fig. 3, which shows the effect of the change in sign of A^sub 3^ when resistance data at temperatures above 480 K are included in the fit. Based on data below 480 K, where there is confidence that the calibration is stable, the resistance derivative decreases by 1 % for a temperature increase of 180 K. Since this temperature dependence is quite small, and the region of extrapolation (480 K to 550 K) is less than half this temperature range, it is anticipated that extrapolation errors should be less than 1 % if only data below 480 K are used for the calibration. If resistance data from temperatures above 480 K are used in the calibration, the thermal conductivity results at 550 K will be about 4 % higher at 550 K and results at 300 K will be 2 % too low. Based on these considerations, the wire calibration from 300 K to 480 K is used in the subsequent data analysis, with the estimate that the relative uncertainty of the measured thermal conductivity increases above 480 K by the additive amount of 1.0 % due to extrapolation in the wire calibration.

3. Results

The thermal conductivity results for the purified toluene sample are given in Table 2 and are shown in Fig. 4. The results in Table 2 have been corrected for thermal radiation, as were our previous results using bare platinum hot wires [4], and with the same empirical optical parameters that were found for the fluid using the previous platinum hot wires [5]. Since the power divided by length was not equal for the long and short wires after the isotherm at 550 K, as shown in Fig. 2, the data from the decreasing temperature portion of the temperature cycle are not reported. There are 184 thermal conductivity data points at temperatures from 300 K to 550 K.

Fig. 1.

Fig. 2.

Table 1.

Fig. 3.

3.1 Repeatability at High Temperatures

The isotherm at 550 K is quite interesting since it includes several replications, with a wide range of power levels, over the course of four days. During this time, the resistance of the long hot wire increased by 5.6 omega and that of the short hot wire increased by 1.9 omega. In addition, since this increase in resistance was not proportional to the wire lengths, the power generation of the two wires differed by up to 1.6 %. Dispite these complications, there seems to be little additional scatter in the thermal conductivity results measured during this 4 day period. In Fig. 5, deviations between the results for the 550 K isotherm are plotted relative to the reference standard of Ramires et al. [21 as a function of applied power level at 550 K. The mean deviation of the data (solid line) is 1.45 % higher than the earlier reference standard, and the scatter (dashed lines) is +/-0.6 % at the level of 95 % confidence. No trend is noted with respect to time throughout this four-day period. Convection in the sample occurs at shorter times for experiments with higher powers and correspondingly larger temperature rises. Consistency between thermal conductivity results at different power levels is considered a good indication that there was no significant convection during the measurements and that compensation for the wire's end effects was achieved. Data at power levels above 0.5 (mu)m^sup -1^ appear to have some influence due to the onset of convection since this isotherm is relatively close to the critical point of toluene at 593.95 K. The contribution of convection on the apparent thermal conductivity appears to be less than 0.5 % at even the highest power levels.

3.2 Uncertainty Assessment

4. Conclusions

The present measurements demonstrate that anodized-tantalum hot wires can be used to make absolute measurements of the thermal conductivity a liquid from 300 K to 550 K. Previous studies with anodized-tantalum hot wires have been limited to temperatures below 370 K [8]. The present transient hot-wire measurements using anodized-tantalum hot wires have a larger uncertainty in the temperature extremes than our previous measurements using bare tantalum hot wires [4] over the same temperature range. This is primarily due to drift in the resistance calibration of the anodized tantalum hot wires at high temperatures. Use of tantalum lead wires may reduce or eliminate this problem in the future and allow accurate measurements of the fluid thermal diffusivity. It was also noted during the experiments that convection occurs earlier, and at lower power levels, as the wire diameter increases. Thus, experiments must be done with lower levels of applied power (smaller temperature rises) when the larger tantalum hot wires are used.

Fig. 4.

Table 2.

Fig. 5.

Fig. 6.

The big advantage of anodized-tantalum hot wires is for measurement of electrically conducting fluids, and this is not a problem in the case of toluene. The anodized tantalum hot wires have a geometry and emissivity different from those of platinum, so the present measurements support the validity of the thermal-radiation correction for absorbing fluids [5]. There is no significant temperature trend in deviations between the present radiation-corrected thermal-conductivity data and thermal-diffusivity data derived from light-scattering measurements of thermal diffusivity. This again supports the validity of the radiation correction [5], since the contribution of thermal radiation is expected to increase with T3.

Acknowledgments

We gratefully acknowledge the financial support of this work by the United States Department of Energy, Division of Engineering and Geosciences, Office of Basic Energy Sciences, through DE-AI02-96ER14597. Both M. L. V Ramires and C. A. Nieto de Castro thank the Faculty of Sciences of the University of Lisbon for a leave of absence and the Physical and Chemical Properties Division of the National Institute of Standards and Technology for support as guest researchers. M. L. V Ramires also thanks JNICT, Portugal, for an additional grant.

References

6. References

References

[1] C. A. Nieto de Castro, S. F. Y. Li, A. Nagashima, R. D. Trengove, and W. A. Wakeham, J. Phys. Chem. Ref. Data 15, 1078 (1986).

[21 M. L. V. Ramires, C. A. Nieto de Castro, and R. A. Perkins, High Temp. High Press. 25, 269 (1993).

[3] W. A. Wakeham, A. Nagashima, and J. V. Sengers, eds., Measurement of the Transport Properties of Fluids, Experimental Thermodynamics Volume III, Blackwell Scientific Publications, Oxford (1991).

[4] R. A. Perkins, H. M. Ruder, and C. A. Nieto de Castro, J. Res. Natl. Inst. Stand. Technol. 96, 247 (1991).

[5] C. A. Nieto de Castro, R. A. Perkins, and H. M. Ruder, Int. J. Thermophys. 12, 985 (1991).

[6] K. Kraft, M. Matos Lopes, and A. Leipertz, Int. J. Thermophys. 16, 423 (1995).

[7] R. D. Goodwin, J. Phys. Chem. Ref. Data 18, 1565 (1989).

[8] M. L. V. Ramires, J. M. N. A. Fareleira, C. A. Nieto de Castro, M. Dix, and W. A. Wakeham, Int. J. Thermophys. 14, 1119 (1993).

[9] H. S. Carslaw and J. C. Jaeger, Conduction of heat in solids, 2nd ed., Oxford University Press, London (1959).

[10] C. A. Nieto de Castro, B. Tams, H. M. Ruder, and W. A. Wakeham, Int. J. Thermophys. 9, 293 (1992).

[11] J. J. Healy, J. J. de Grout, and J. Kestin, Physica 82C, 392 (1976).

[12] Y. Nagasaka and A. Nagashima, J. Phys. E: Sci. Instrum. 14, 1435 (1981).

[13] M. J. Assael, L. Karagiannidis, S. M. Richardson, and W. A. Wakeham, Int. J. Thermophys. 13, 223 (1992).

[14] E. Gebhardt and H. D. Seghezzi, Z. Metallk. 48, 430 (1957). [15] P. W. Bridgman, Proc. Amer. Acad. Arts Sci. 81, 165 (1952). [16] J. Kestin, and W. A. Wakeham, Physica 92A, 102 (1978). [17] Y. S. Touloukian, ed., Thermophysical properties of high

temperature solid materials-Volume 1: Elements, Macmillan, New York, (1967).

[18] T. Yamada, T. Yaguchi, Y. Nagasaka, and A. Nagashima, High Temp. High Press. 25, 513 (1993).