Measurement of the Thermal Conductivity of Powders by DSC

The first measurements of the thermal conductivity of powders [1] showed that powders can be an interesting alternative to vacuum systems for achieving good thermal insulation. Currently powders of different materials (ceramics or polymers) are used in packaging or for building insulation. On the other hand, the low thermal conductivity of powders entails serious risks in the production and manipulation of energetic powders intended for pyrotechnics or explosives. Knowledge of the thermal conductivity of powders is therefore crucial to avoid spontaneous ignition.

Introduction

Many industrial activities (ceramics, powder metallurgy, food, etc.) involve the heating or cooling of powders.

We have developed a new method for measuring the thermal conductivity, κ, of powders using heat flow DSC. The measurement of κ is made at discrete temperatures that correspond to the melting points of selected pure metal references. Our results indicate that κ can be determined within an error bar of ±10%.

Several methods are currently available for measuring the thermal conductivity, κ, of thin solid slabs by DSC [2, 3, 4]. Camirand’s method [3] involves measuring the melting peak of a reference metal placed on top of a slab of the sample to be analyzed. The thermal conductivity, κ, is obtained from the slope of the low temperature side of the peak.

The difficulty with powders is that they have to be packed inside a pan. This causes a dramatic change in the geometry of the path of thermal conduction. We have successfully adapted Camirand’s method to these particular samples.

A spherical reference metal bead is placed on top of the powder filling the pan, as shown in Figure 1. When the pan is heated at a constant rate, the slope of the low-temperature side of the melting peak will depend on the resistance of the powder, R, as given by eq 1:

⎟slope⎟ = 1 / R+RDSC (1)

where RDSC is the thermal resistance of the DSC sensor measured without the powder. It was shown that after an initial transient period the constant slope given by eq 1 is valid for any pan geometry [5].

For simple geometries such as those of commercially available cylindrical pans [5] or home-made hemispherical pans, κ can be easily determined from R [6] using eq 2:

κ = ε – 1 /π R (1/D_m - 1/D_p),     (2)  

where D_m is the diameter of the reference metal bead and D_p is the diameter of the pan. The value of the constant, ε, for a hemispherical pan is 1.

The value of ε for cylindrical pans was obtained by solving the heat transport equation in the steady state for a range of different pans with different height-todiameter ratios (H/D_p) and different bead sizes that cover all practical situations. Some particular values for the correction factor, ε, are given in Table 1.

Finally, we would like to point out that the slope given by eq 1 is only reached after a transient period whose duration depends on several experimental conditions [5]. Figure 2 shows that the melting time for a large metal bead (Dm = 2.26 mm in Figure 2) is longer than the transient period and a constant slope is obtained. Consequently, in this case, R can be correctly measured.

On the other hand, a smaller bead (D_m = 1.29 mm in Figure 2) melts too quickly. In general, it is best to use a low heating rate in order to achieve a constant slope according to eq 1.

Conclusions

Our method allows the κ value of powders to be measured with an accuracy of about ±10%. It is important to choose the measurement parameters so that the slope of the melting peak reaches a constant value. Otherwise the value of κ will be too small.

_{Measurement of the Thermal Conductivity of Powders by DSC | Thermal Analysis Application No. UC 463 | Application published in METTLER TOLEDO Thermal Analysis UserCom 46}