2. used to cool a heated surface. In testing, it was shown by Wait, et al. that the device’s
effect on the convection coefficient is largely dependent on orientation of the fan
(cantilever).[11]
Method 1: Impingement
Here, impingement is the idea that with microsized jets, air, or other such fluids, can be
impinged on a heated surface at high velocities to cool a surface. Instead of requiring a large area of
flow over something like a single chip, a jet’s nozzle (outlet) could be engineered to impinge on
specific areas. This is an appealing solution to the problem of heat, since heat transfer rates are
dominated by the metrics of the impinging device, which are diameter of jet, spacing of jets, and
location of jets (i.e. upon which areas are being impinged). Wang et al. scrutinized the physics of
impingement cooling on this scale, specifically evaluating the effect of the different flow regimes on
heat transfer rates. A profile view of a jet impinging on a surface is shown in fig. 2. At a height h away
from the surface, the flow will go through four regimes while on a surface: a stagnation or
impingement zone, a laminar region, a transition region, and a turbulent zone. All of these zones and
their relative locations from each other are shown in fig. 2. While all of the zones had effects on heat
transfer rates, Wang et al. concluded it was the stagnation zone that must be minimized to optimally
use impinging jets, which means packing as many arrays of jets would not be an effective means to
transfer heat.[1]
Figure 2: the different regimes of flow from a microscale impingement system in a
steady state environment. The fluid is impinged onto the hot surface by means of a
circular orifice jet. The flow cools the surface as the flow transitions from laminar to
turbulent. Different types of flow determine the effectiveness of heat transfer. Wang,
et al. developed models for 3D microjet flows and found that stagnation regions need
to be minimized to have desired convection coefficient.[1]
4.
Figure 4: Brunschwiler, et al.’s impingement design. The most disadvantageous aspect
of using jet impingement is the high pressure drop in the jet. This design employs a
bifurcation system, where the inlets and outlets branch across the heated area. The
top left figure shows a 2D side view of the bifurcated inlet and outlet jets. In the top
right figure, the light blue surface is being cooled by the impinged air from the
bifurcated inlet jet (dark blue) and at the same time, air is being removed from the
surface by the bifurcated outlet jet (red). The jets without the heated surface is shown
in an isometric view in the bottom left figure. The bottom right figure shows a
topdown diagram of the system meshed together. Using bifurcation, a much lower
pressure drop occurs across the array. The shape of the array is a compromise between
using more space to make use of bifurcation and minimizing space by slotting the inlets
and outlets together.[2]
Method 2: Heat Exchangers & Sinks
Just as heat can be transferred between fluids to remove thermal energy from a device or
passively sinked into material to increase overall convection on the macroscale, MEMS devices can
utilize similar physics to cool on the microscale. Cooligy developed a closed loop liquid cooling system,
which is made up of a counterflow heat rejector, electrokinetic (EK) pump, and a microheat
exchanger. The microheat exchanger was designed to handle heat fluxes greater than an impressive
500 W/cm2
and engineered to be mounted on the back of a processor.[7]
However, the cost of a large
heat flux is the size requirement of the rejector and the pump, making the system not much smaller
than a standard heat dissipation device.
5.
Figure 5: a diagram the microheat exchanger in Cooligy’s closed loop Liquid Cooling
System (Datta, et al.). The cooling liquid flows in from the inlet on the housing on the
right side of the diagram before filtering into the high surface to volume ratio
microstructure (HSVRM) illustrated here in gray. The heat sinks into the HSVRM
before leaving out the outlet on the the left side of the diagram, effectively cooling
the circulating liquid. The heat exchanger is a critical component in Cooligy’s design
because it is a closedloop system. The effectiveness of the system is entirely
dependent on the HSVRM.[5]
Datta et al.’s analysis of Cooligy’s microheat exchanger (fig. 5) attributed the large heat flux to
a microheat exchanger with high surface to volume ratio microstructures (HSVRM).[5]
The HSVRM
allows the the EK pump to work optimally at a high flow rate with low current. This is a critical
functional requirement as high pressure drops in these microchannel heat exchangers are a defining
factor in terms of effectiveness.[7]
On the subject of these pumping requirements for microchannel
heat sinks, Garimella et al. published work analyzing the pumping requirements for a regime of
microchannel hydraulic diameters (318 μm to 564 μm). They found that the requirements for pressure
and flow rates were high, and developed an analytical model for minimum operating region for heat
sink pumping. This region is shown in fig. 6.[9]
Figure 6: Garimella, et al. developed a graph of the analytic model to determine the
“minimum operating limit” of a microchannel heat sink. The model could then predict
if the heat sink met the pumping requirement of the system. If the pumping pressure
or flow rate of the fluid through the heat sink were too demanding, the designers
could use this model to optimize the dimensions of the microchannels.[9]
7.
Figure 8: is a profile view of V. K. Pamula, et al.’s device. Similar in fundamental
concept shown in fig. 7, this design utilized control electrodes to control wettability.
Control electrodes on the bottomplate in conjunction with a ground electrode on the
topplate governs droplets movement hydrophobic surfaces coating both plates.
Unlike other electric fieldbased liquid movement systems, this one grants the
designer ability to implement control loops which increases the effectiveness of the
device.[8]
Other Promising MEMS Cooling Technologies
Presented in this section are two other relevant cooling technologies worth mentioning. The
first is an air propulsion system that takes points from both impingement and electric field designs.
Yang et al. investigated a corona driven air propulsion system to explore “possibility of building an
electrostatic air pump...from microelectronic devices and MEMS.”[6]
Corona driven air propulsion
works by ionizing gas molecules using a high intensity electric field creating an ion stream. As the
ionized particles flow to a collector electrode, a flow of air is formed. This effect is shown in fig. 9A.
The downside of ionizing air is the requirement of a substantial electrical potential, but if this idea
were to be used in a MEMS device, there are three advantages over other MEMS designs. The first
two advantages are a low noise dynamic airflow profiles and versatile shapes and sizes could be
made, but the third, which is particularly relevant for cooling, is that corona driven flow ensures
greater flow closer to the wall than pressuredriven flow due to the constant coulombic force applied
to the fluid (fig. 9B).[6]
(A) (B)
8. Figure 9: Fig. 9A illustrates the operation of the coronadriven air propulsion system.
By setting a large electrical potential to the positive corona electrode (left), positively
ionized air particles are repulsed towards a collector electrode (right) via an electric
field. The relative velocity of the particles is shown by the size of the arrows. This
stream of ions translates into airflow and the forced flow increases the convection
coefficient. Fig. 9B shows the velocity profile of the flow induced by a pressure
difference (red) compared to one induced by the coronadriven air pump (green). The
position on the yaxis is relates to position along width of channel and the vaxis
relates to the relative velocity of the flow in each case. Coronadriven flow allows
greater flow closer to the channel walls due to the constant coulombic force applied
to the air. The larger flow field closer to the wall decreases the boundary layer
thickness, increasing the convection coefficient across the wall.[6]
The second is Huang et al.’s microthermoelectric cooler (μTEC). This group researched,
constructed, and tested two μTECs: a bridgetype polysiliconbased cooler and a columntype
telluridebased cooler. Throughout their analysis and testing, the bridgetype (fig. 10A) had a far
better cooling performance over the columntype. They attributed this to the bridgetype having a
larger thermal resistance (due to its smaller area per length), a cooling performance of 136.5 K/A at
an operating current of 80 mA, and a working temperature difference of 5.6 K (fig. 10B).[10]
The benefit
of such a device comes from effective coverage area and the power needed to run.
9.
(A)
(B)
Figure 10: Fig. 10A shows the manufacturing steps of building a low cost μTEC. The
left column shows the μTEC construction from the front, while the right column
10. shows a side view. A combination of eight materials was used, each shown at the top
of the figure along with its graphical representation. This bridgetype had a much
smaller area to length ratio, which increased the thermal resistance, making it the
better manufacturing choice for μTECs. Fig. 10B shows a heatmap (topdown view) of
the temperature distribution across the bridgetype polysiliconbased
microthermoelectric cooler (μTEC) Huang, et al. built. Dark blue is the coldest of
regions and the warmer regions are increasingly more red. Whereas the columntype
telluridebased μTEC achieved a 1.2K temperature difference across the heated
surface, this bridgetype achieved a 5.6K temperature difference. The μTEC works by
employing the Peltier effect, where heat is transferred between two different
materials via a DC voltage. [10]
Conclusion
Heat dissipation in laptops and other such electronics is becoming an increasingly important
issue with the advent of more and more powerful processors combined with the ubiquity of
computing. In this paper, we have identified and briefly evaluated some of the MEMS technologies
designed to cool electrical components for laptops. Impingement, heat exchangers and sinks, and
electric field based liquid cooling are presented, and the paper ends with some other promising
technologies. Impingement and heat sinks offer the highest heat transfer rates while the
controllability of liquids using electric fields has its own benefits. Although impingement and heat
sinks have the advantage of higher heat transfer rates, 420 W/cm2
with impingement bifurcation and
up to 500 W/cm2
with a closed loop microheat exchanger system, they both suffer from problems
with high pressure drops. Electrohydrodynamic cooling avoids this problem, but tend to have much
lower heat transfer rates. Some other promising technologies use high voltages to ionize air to induce
flow, while others avoid convection as the primary mode of heat dissipation using the Peltier effect
(μTECs). Further research involving the specific constraints of laptops (such as space and fluid flux)
when incorporating these systems should be done before adoption.