Using new thin-film techniques to manage thermal and energy problems in embedded designs

Seri Lee, Nextreme Thermal Solutions Inc.

May 9, 2007

Seri Lee, Nextreme Thermal Solutions Inc.

Solid-state cooling and power generation have long been sought after as a solution for challenging thermal-management and energy problems. Thus, to address some of these issues, thermoelectric modules have been available for decades.

The core component of a thermoelectric module is a thermocouple, which consists of two dissimilar semiconductors (referred to as p-type and n-type to describe dissimilar electrical conduction mechanisms in the two materials) connected together by a metal plate.

Electrical connections at the top complete an electric circuit. Thermoelectric cooling (TEC) occurs when current is supplied, in which case the thermocouple cools on one side and heats on the other by the Peltier effect.

Thermoelectric generation (TEG) occurs when the couple is put in a thermal gradient (i.e. the top is hotter than the bottom), in which case, the device generates current, converting heat energy into electrical power by the Seebeck effect (Figure 1, below).

Figure 1: TEC occurs when current is supplied (left). TEG occurs when the couple is put in a thermal gradient, in which case, the device generates current, converting heat energy into electrical power by the Seebeck effect (right).

In practical applications, a large number of p- and n-type pellets are cut from boules of thermoelectric material, and thermocouples are assembled together (electrically in series, thermally in parallel) to form a TEC or a TEG. Conventional modules are referred to as "bulk" modules because of their size and fabrication approach (Figure 2, below).

Figure 2: Conventional modules are referred to as "bulk" modules because of their size and fabrication approach.

Such devices found early (and continued) use for power generation in aerospace applications, and for cooling and temperature control in instrumentation, telecommunications and other specialty applications.

While thermoelectric modules deliver the benefit of solid-state operation, they suffer from many disadvantages. They are generally inefficient, fragile and big.

The very size and discrete nature of conventional thermoelectric modules severely limit their implementation.

More recently, a significant amount of development has been focused on thin-film thermoelectric devices. Thin-film thermoelectric materials can be grown by conventional semiconductor deposition methods, and devices can be fabricated using conventional semiconductor micro-fabrication techniques.

The resulting devices are much smaller than conventional offerings and are suitable for direct integration into modern manufacturing methods (Figure 3 below).

Figure 3: A thin-film thermoelectric module  measures 3.5mm x 3.0mm x 0.1mm in size.

The thin-film TEC is about 6x smaller in its x-y dimensions and about18x smaller in the z dimension. Thus, on a volume basis, the thin-film TEC is about 110x smaller in size.

Thin-film vs. bulk
While it is interesting to compare thin-film and bulk thermoelectric modules on the basis of size, it is much more useful to compare them on the basis of performance. The most common measure of the performance of a thermoelectric module is its load line.

Figure 4: (a) The thin-film TEC pumps almost 4x the power of the bulk device, although it holds off only about 60 percent of the temperature difference; (b) In the cooling mode, thin-film thermoelectric devices deliver good power-density pumping capability (Q/area) combined with a good temperature gradient.

A load line is generated at fixed operating current and specified reference temperature by plotting the temperature difference ( Delta T) that the device can achieve between its top and bottom plate. As a function of the power (Q), it can pump against the temperature gradient.

Figure 4a above shows the load lines of the thin film and bulk devices, both measured at a reference temperature of 25 degrees Centigrade. In this case, the characteristic load line for a module is shown at its maximum operating current (I_max).

Under I_max conditions, the Delta T at zero Q is referred to as DeltaT_max, and the power pumped at zero Delta T is referred to as Q_max.

Neither Delta T_max nor Q_max is a practical operating condition for the device. However, both define the performance envelope of a device and are often used as a basis of comparison.

The performance comparison shown in Figure 4a above is interesting - the thin-film TEC pumps almost 4x the power of the bulk device, although it holds off only about 60 percent of the temperature difference.

Yet, if we consider dimensional differences, the intrinsic performance of the thin-film TEC is quite amazing. The thin-film TEC holds off a maximum of 40°C (Delta T_max) across the thickness of a piece of paper.

And it pumps a maximum of almost 16W (Q_max) through the area of a piece of confetti. The respective response time of the bulk and thin-film devices is not shown. The thermal response time of bulk devices is in seconds (e.g. to achieve a cooling target), while the response time of thin- film TECs is in milliseconds.

Figure 4a shows the measured load lines of the bulk and thin- film TECs, recast to account for dimensional performance differences.

In this case, Delta T is divided by the thickness over which it is achieved, and Q is divided by the area over which it is achieved. The vertical axis represents the temperature gradient that the device can hold off across the thickness, and the horizontal axis shows the power density that the device can pump over its area. The complete bulk-thin film comparison is summarized in Table 1 below.

Table 1: Thermal response time of bulk devices is in seconds, while that of thin-film TECs is in milliseconds

The comparison in Figure 4b earlier illustrates a fundamentally new operating regime offered by thin-film TECs.

In the cooling mode, thin-film thermoelectric devices deliver unprecedented power-density pumping capability (Q/area) combined with an extraordinary temperature gradient (Delta T/H). Similarly, in power generating mode, these devices achieve unmatched performance characteristics.

Resulting implications
Nevertheless, to take advantage of the characteristics shown in Figure 4b, the thin-film devices need to be put into the right thermal and dimensional form-factor environment.

Due to their small size, the devices can be integrated directly into a semiconductor or optoelectronic package to achieve localized cooling and temperature control hat. The devices are also small enough and pump enough power to enable new applications in power generation, medical devices and instrumentation.

Finally, the concept of integrated, localized cooling and power generation is now available to make temperature control or energy delivery a complementary function in an electric circuit.

Seri Lee is Chief Technology Officer at Nextreme Thermal Solutions. To read a PDF version of this story, go to "Thin-film tech sets new performance paradigm."