Chapter 5 Cooling of Power Switching Semiconductor Devices 134

Conduction heat transfer

Conductive heat transfer occurs when energy exchange takes place, by direct impact of molecules, from
a high temperature region to a low temperature region.
Conductive heat loading on a system may occur through lead wires, mounting screws, etc., which form a
thermal path from the device being cooled to the heat sink or ambient environment.
The one-dimensional model for general molecular (non-radiation) heat transfer is given by
δT δT
Pd = −λ A + ρm A A (W) (5.2)
δA δt

5
where δT =T2 -T1 or ∆T, is the temperature difference between two regions of heat transfer
λ is thermal conductivity, W/m K, see Appendix 5.24
ρm is density of the heatsink material
cp is specific heat capacity, J/kg K, such that ∆T = W/mcp (W is energy, m is mass)
A is distance (thickness).
Equation (5.2) shows that the thermal power generated Pd is balanced by the stored thermal power (first
term on the right hand side) and the thermally dissipated power (second term on the right hand side).

Cooling Assuming steady-state heat dissipation conditions, then δT / δ t = 0 in equation (5.2).

of Conduction through a homogeneous solid, from Fourier’s law of heat conduction, is therefore given by
λ
Power Switching Semiconductor Devices Pd =
A
A ∆T (W) (5.3)

Convection heat transfer

When the temperature of a fluid (a gas or liquid) flowing over a solid object differs from that of the object
surface, heat transfer occurs. The amount of heat transfer varies depending on the fluid flow rate.
Convective heat loads are generally a result of natural (or free) convection. This is the case when gas
flow is not artificially created as by a fan or pump (forced convection), but rather occurs naturally from
Semiconductor power losses are dissipated in the form of heat, which must be transferred away from the varying density in the gas caused by the temperature difference between the object being cooled
the switching junction, if efficient switching is to be maintained. The reliability and life expectancy of any and the gas. Heat transfer processes that involve change of phase of a fluid (for example evaporation or
power semiconductor are directly related to the maximum device junction temperature experienced. It is condensation) are also considered to be convection.
therefore essential that the thermal design determine accurately the maximum junction temperature from The convective loading is a function of the exposed area and the difference in temperature between the
the device power dissipation. Every 10°C junction temperature decrease, doubles device lifetime. load and the surrounding gas. Convective loading is usually most significant in systems operating in a
gaseous environment with small active loads or large temperature differences.
i. Heat load
The heat load may be active or passive, then there is a combination of the two. An active load is the Convection heat transfer through a fluid or air, under steady-state conditions in equation (5.2), is given
heat dissipated by the device being cooled. It is generally equal to the input power to the device, for by Newton’s law of cooling, that is
example, P = V × I = I 2 × R = V 2 / R . Passive heat loads are indirect, are parasitic in nature, and may Pd = h A ∆T (W) (5.4)
consist of radiation, convection or conduction. The convection heat transfer coefficient h (= λ / A ), W/m2K, depends on the heat transfer mechanism
Heat energy, due to a temperature difference, can be transferred by any of, or a combination of, three used and various factors involved in that particular mechanism. It is not a property of the fluid.
mechanisms, viz., Natural or free convection is essentially still to slightly stirred air with h values ranging from 1 to 25.
• Convection - heat transferred to a moving fluid which takes the heat away Forced convection is air moved by a fan or other active method, giving h values ranging from 10 to 100.
• Conduction - heat flows through a thermal conducting material, away from the heat source Values for forced liquid convection are 50 to 20,000, while the h range for boiling and condensation is
• Radiation - heat flow by long-wave electromagnetic radiation, e.g. infrared. 2,500 to 100,000.
Electromagnetic thermal radiation heat transfer For natural vertical convection in free air, the losses for a plane surface may be approximated by the
When two objects at different temperatures come within proximity of each other, heat is exchanged following empirical formula
between them. Electromagnetic wave propagation, radiation is emitted from one object and absorbed by
the other. As a result of the temperature difference, the hot object experiences a net heat loss and the ∆T 5 ∆T
Pd = 1.35 A 4
= 1.35 A 4 ∆T = hA ∆T (W) (5.5)
cold object undergoes a net heat gain. This is termed thermal radiation. A A
where ℓ is the vertical height in the direction of the airflow and h is of the form
Radiation heat loads are usually considered insignificant when the system is operated in a gaseous ¼
environment since other passive heat loads are usually greater. Radiation loading is usually significant  ∆T 
h =K   (5.6)
in systems with small active loads and large temperature differences, especially when operating in a  A 
vacuum environment, where convection processes are absent. Two cases occur for forced airflow, and the empirical losses are
Electromagnetic thermal radiation heat loading (for a grey body, ε < 1) is given by • for laminar flow
Pd = σ ε A (T 14 − T 24 ) (5.1) v
Pd = h A ∆T = 3.9 A ∆T (W) (5.7)
where Pd is the rate of radiated heat transfer (that is, the power dissipated), W A
-8 2 4
σ is the Stefan-Boltzmann constant (5.667×10 W/m K ) • for turbulent flow
ε is a surface property, termed emissivity, 0 ≤ ε ≤ 1, see Table 5.6 and Appendix 5.25
5v
4
A is the surface area involved in the heat transfer, m2 Pd = h A ∆T = 6.0 A ∆T (W) (5.8)
T is absolute temperature, K A
where v is the velocity of the vertical airflow.
The ideal emitter, or black body, is one which gives off radiant energy with ε = 1 in equation (5.1).

BWW
135 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 136

Combined convection and conduction heat transfer For convection, from equation (5.4), the effective thermal resistance is
Heat Loss (or gain) - through the walls of an insulated container (combined convection and conduction, ∆T 1
estimation) is Rθ = = (5.13)
Pd hA
∆T
Pd = A × For radiation, from equation (5.1), the effective thermal resistance of radiation is
A /λ + 1 /h ∆T 1 1
where Pd is the heat lost or gained, W Rθ = = = (5.14)
Pd σε A × (T 1 + T 2 ) (T 12 + T 22 ) hr A
ℓ is the thickness of insulation, m 2
λ is the thermal conductivity of the insulation material, W/m K where the radiation heat transfer coefficient, hr, in W/m K, is
A is the outside surface area of the container, m2. hr = σε × (T 1 + T 2 ) (T 12 + T 22 ) ≈ 4 × σε ×T mean
3
(5.15)
h is the convection heat transfer coefficient of the surface material, W/m2 K
where Tmean is the arithmetic mean of T1 and T2, specifically ½( T1 + T2).
∆T= To/s - Ti/s
To/s is the outside temperature, °C
Ti/s is the inside temperature, °C
5.1 Thermal resistances

A general thermal dissipation model or thermal equivalent circuit for a mounted semiconductor is shown
ii. Transient heating in figure 5.1. The total thermal resistance from the virtual junction to the open air (ambient), Rθ j-a, is
Some designs require a set amount of time to reach the desired temperature.
R × (Rθ c-s + Rθ c-a )
The estimated time required to heat (or cool) an object (also known as Newton’s Law of Cooling) is Rθ j-a = Rθ j-c + θ c-a (K/W) (5.16)
m × c p × ∆T Rθ c-a + Rθ c-s + Rθ s-a
t =
P
P is the mean heat added (or being removed) from the object, W, watts In applications where the average power dissipation is of the order of a watt or so, power
m is the mass (weight) of the object, kg (density x volume) semiconductors can be mounted with little or no heat sinking, whence
cp is the specific heat of the object material, J/kg K Rθ j-a = Rθ j-c + Rθ c-a (K/W) (5.17)
t is the time required to cool down (or heat up) the object in seconds Generally, when employing heat sinking, Rθ c-a is large compared with the other model components and
∆T = To – Tf equation (5.16) can be simplified to three series components:
To is the starting temperature, °C Rθ j-a = Rθ j-c + Rθ c-s + Rθ s-a (K/W) (5.18)
Tf is the final temperature, °C
(
P = ½ Pt o + Ptt )
Pto is the initial heat pumping capacity when the temperature difference across the cooler is zero.
Ptt is the heat pumping capacity when the desired temperature difference is reached and heat-pumping virtual Tjunction
capacity is decreased. junction Rθ c-a
Heat loading may occur through one or more of four modes: active, radiation, convection or conduction.
package Tcase Rθ j-c
By utilizing these equations, the heat load can be estimated. case at
mount
iii. Thermal resistance Rθ c-hs
heatsink Theatsink
It is generally more convenient to work in terms of thermal resistance, which is defined as the ratio of
temperature change to power. Thermal capacity is the reciprocal of thermal resistance. For conduction,
from equation (5.4), thermal resistance Rθ is
∆T 1 A
Rθ = = = (K/W) (5.9)
Pd hA λA
where the conduction thermal heat transfer coefficient, h, is Rθ hs-a
λ
h= (5.10) ambient Tambient
A
The average power dissipation Pd and maximum junction temperature Tlj , in conjunction with the
ambient temperature Ta, determine the necessary heat sink, according to equation (5.9)
∆T Tl − T a Figure 5.1. Semiconductor thermal dissipation equivalent circuit.
Pd = = j (W) (5.11)
Rθ j-a Rθ j-a
where Rθ j-a is the total thermal resistance from the junction to the ambient air. The device user is
restricted by the thermal properties from the junction to the case for a particular package, material, and 5.2 Contact thermal resistance, Rθ c-s
header mount according to
The case-to-heat-sink thermal resistance Rθ c-s (case means the device thermal mounting interface
∆T Tl − T c surface) depends on the package type, interface flatness and finish, mounting pressure, and whether
Pd = = j (W) (5.12)
Rθ j-c Rθ j-c thermal-conducting grease and/or an insulating material (thermal interface material, TIM) is used. In
where Tc is the case temperature, K and general, increased mounting pressure decreases the interface thermal resistance, and no insulation
Rθ j-c is the package junction-to-case mounting thermal resistance, K/W. other than thermal grease results in minimum Rθ c-s. Common electrical insulators are mica, aluminium
oxide, and beryllium oxide in descending order of thermal resistance, for a given thickness and area.
An analogy between the thermal (and magnetic) equations and Ohm’s law and Kirchhoff’s laws is often Table 5.1 shows typical contact thermal resistance values for smaller power device packages, with
made to form models of heat flow. The temperature difference ∆T could be thought of as a voltage drop various insulating and silicone grease conditions. Silicon based greases are best, for example Assmann
∆V, thermal resistance Rθ corresponds to electrical resistance R, and power dissipation Pd is analogous V6515, spread at a thickness of 100µm to 150µm, on both surfaces. Grease in excess of this will be
to electrical current I. [viz., ∆T = Pd Rθ ≡ ∆V = IR]. See Table 5.10. squeezed out under clamping pressure. Initial grease thermal resistance decreases slightly after a few
normal deep thermal cycles.
137 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 138

Table 5.1: Typical case-to-heat-sink thermal resistance value for various small packages • Cure-in-Place Thermally Conductive Compounds
A thermally conductive compound again incorporates thermally conductive ceramic fillers, shown
Rθc-s (K / W) in Table 5.3, but unlike thermal greases, the binder is a rubber material. When first applied, the
Package Insulating washer paste-like compound flows into the interstices between the mating surfaces. Then, when
Silicone grease
with without subjected to heat, it cures into a dry rubber film. Besides its thermal properties, this film also
No insulating washer
serves as an adhesive, allowing a tight, void-free joint without the need for additional fasteners.
0.10 0.3
TO-3 Teflon 0.7-0.8 1.25-1.45 Thermally conductive compounds can fill larger gaps in situations where thermal greases might
Mica (50 - 100 µm) 0.5-0.7 1.2-1.5 ooze from the joint. Although application and performance is similar to that of thermal grease,
cleanup is easier, simply involving removal of the excess cured rubber film.
No insulating washer 0.3-0.5 1.5-2.0
TO-220
Mica (50 - 100 µm) 2.0-2.5 4.0-6.0
• Thermally Conductive Elastomeric Pads
No insulating washer 0.1-0.2 0.4-1.0
TO-247
0.5-0.7 1.2-1.5 A thermally conductive elastomeric pad consists of a silicone elastomer filled with thermally
Mica (50 - 100 µm)
conductive ceramic particles and may incorporate woven glass fibre or dielectric film
SOT-227 No insulating washer 0.1-0.2 0.3-0.4
0.5-0.7 1.0-1.2
reinforcement. Typically ranging in thickness from 0.1 to 5 mm and in hardness from 5 to 85
ISOTOP Mica (50 - 100 µm)
Shore A, these pads provide both electrical insulation and thermal conductivity, making them
useful in applications requiring electrical isolation. Thicker pads prove useful when large gaps
must be filled. During application, the pads are compressed between the mating surfaces to make
The thermal resistance of a heat-conducting layer is inversely proportion to heat conductivity of the
them conform to surface irregularities. Mounting pressure must be adjusted according to the
material and in direct ratio to its thickness. If the clamping pressure is increased, the layer thermal
hardness of the elastomer to ensure that voids are filled. A mechanical fastener is essential to
resistance falls. In figure 5.2, the exemplary dependence of the gasket thermal resistance per surface
maintain the joint once assembled.
unit on pressure is shown. However, with a growth of pressure it is necessary to find an optimum, as the
clamping effort should not exceed a package recommended value or introduce differential thermal
• Thermally Conductive Adhesive Tapes
expansion problems into the clamping arrangement.
A thermally conductive adhesive tape is a double-sided pressure sensitive adhesive film filled with
thermally conductive ceramic powder. To facilitate handling, aluminium foil or a polyamide film
K/m W pu

may support the tape; the latter material also provides electrical insulation. When applied between
1 mating surfaces, the tape must be subjected to pressure to conform to the surfaces. Once the
joint is made, the adhesive holds it together permanently, eliminating the need for supplemental
2

0.9
fasteners. No bond curing is needed. One limitation of thermally conductive tapes is that they
0.8
cannot fill large gaps between mating surfaces as well as liquids; hence, the convenience of tape
Thermal resistance

0.7 mounting is traded against a nominal sacrifice in thermal performance.

0.6
• Phase Change Materials
0.5 Solid at room temperature, phase change materials, shown in Table 5.3, melt (that is, undergo a
phase change) as the temperature rises to the 40° to 70°C range. This makes the material (0.13
mm thick in its dry film form) as easy to handle as a pad, while assuring, when subjected to heat
0 50 100 150 200
during the assembly process, the melt flows into voids between mating surfaces as effectively as
Pressure psi a thermal grease. Applying power to the power electronic component introduces the needed heat
Figure 5.2. Exemplary dependence of the gasket thermal resistance on clamping pressure.
for the phase change to occur, establishing a stable thermal joint. These materials consist of
organic binders (that is, a polymer and a low-melt-point crystalline component, such as a wax),
thermally conductive ceramic fillers, and, if necessary, a supporting substrate, such as aluminium
5.2.1 Thermal Interface Materials foil or woven glass mesh. See section 5.2.2 for further details.

To be effective, heatsinks require intimate surface-to-surface contact with the component to be cooled.
Unfortunately, irregular surface areas, both on the electronic components and on the heatsink mating
surface prevent good contact. Up to 99% of the surfaces are separated by a layer of interstitial air, which Table 5.2: Thermal Interface Material (TIM) thermal resistances
is a poor conductor of heat thus presents a thermal barrier. Therefore, a thermally conductive interface
material is necessary to fill the interstices and microvoids between the mating surfaces. To ensure that Thickness Thermal Conductivity Thermal resistance Thermal resistance
electrical problems are not inadvertently introduced while solving the thermal problems, it is often Interface λ Rθc-s pu area
essential that the thermal interface materials also perform an electrical isolation function. mm W / m.K K/W K cm2/W
Thermal interface materials TIMs vary widely in terms of performance (that is, thermal, electrical, and Dry Joint n/a n/a 2.9 1.8 - 2
physical properties), general appearance, and mode of application. Among the most commonly used
classes of thermal interface materials are: thermal greases, cure-in-place thermally conductive Thermal Grease 0.076 0.7 0.9 0.5 - 1.1
compounds, gap filling thermally conductive elastomeric pads, thermally conductive adhesive tapes, and Thermal Compound 0.127 1.2 0.8 0.2 - 0.7
phase change materials, all of which are summarised in Table 5.2 and are briefly described.
Elastomer 0.254 5.0 1.8 1-2
• Thermal Greases Adhesive Tape 0.229 0.7 2.7 0.5 - 1.5
Comprised of thermally conductive ceramic fillers in silicone or hydrocarbon (organic) oils, as
shown in Tables 5.3a and b, a thermal grease is a paste, which is applied to at least one of the Eutectic (soldering)
61 0.1
Sn(91) Zn(9)
two mating surfaces. When the surfaces are pressed together, the grease spreads to fill the void.
During compression, excess grease squeezes out from between the mated surfaces. Some form
of clip or other mounting hardware is needed to secure the joint. Although it is comparatively
inexpensive and thermally effective, thermal grease is not an electrical insulator.
Disadvantageously, it can be inconvenient to dispense and apply, and requires cleanup to prevent
contamination problems.
 139 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 140

Table 5.3: Thermal Interface material (TIM) parameters liquid spreads, the molten PCM displaces the interstitial air and the distance between the surfaces
decreases. Both of these processes act to reduce the thermal resistance of the interface and the
temperature difference is seen to decrease rapidly, reaching the performance of thermal grease, 4°C. In
Thermal Temperature effect, the solid PCM film has turned into thermal grease and a grease-like joint has been formed. The
Material parameters Resistivity Thickness Lifetime
Conductivity range
next time the thermal load is activated, the interface does not experience the large temperature
(a) Thermal grease λ, W / mK ρ, Ω.cm °C mm difference because the void free thermal joint has already been established.
Zinc Oxide/Silicone 0.74 > 5×1014 < 150 5 years @ 25C
13
Aℓ203 / Non-Silicone 2 > 10 < 150 1 year @ 25°C
AℓN / Non-Silicone 4 > 10
13
< 150 1 year @ 25°C 15

13
Non-silicone paste 2 > 1×10 n/a 1 year @ 25°C

°C
Dry interface joint
14
Silicone paste > 4.0 > 1×10 n/a 1 year @ 25°C
10
Non-curing paste 4 > 1×1013 n/a n/a

- Theatsink
Non-silicon Phase Change
> 7.5 1Å -50…+150 n/a
polysynthetic oils Material
thermal phase change 0.90 1×1014 -50…+200 n/a 5
13

Tcase
Silicone 0.8-1.2 1×10 -60…+180 5 years
Grease
Tensile
(b) Heat conducting
Strength 0
gaskets
psi
14
0 300 600
Silicon/Alumina filled 0.38 1×10 -60…+150 0,08 3000
Time seconds
14
Silicon/Boron Nitride 2.07 1×10 -60…+200 0,25 1000
Aℓ2O3 filled 1.7 n/a < +150 0,10 15…300 Figure 5.3. Performance of a PCM, compared to a dry interface joint and thermal grease.

AℓN filled 3.6 n/a < +150 0,10 15…300

The thermal resistance across an interface depends on the thermal conductivity of the PCM in the
97% Graphite filled 5 n/a < +450 0,13 15…300
interface and its conductive path length. Thermal conductivity is a function of the type and level of the
(c) Phase changing
Clamping
Phase Change ceramic filler in the formulation, typically between 0.7 and 1.5 W/m.K. The amount of filler that can be
heat conducting Temperature added is limited by the need to keep the viscosity as low as possible to achieve proper flow of the PCM
psi
gaskets °C in the interface. The thickness of the interface formed by a PCM is determined by the flatness of the
Black/self-adhesive layer 0.7 n/a < +120 0.077 10…200 < +120 mating surfaces, the clamping pressure, and the viscosity and rheology of the molten PCM. Most
15 commercial surfaces deviate from true flatness by as much as 2µm/mm. This means that the thermal
Grey/self-adhesive layer 1.0 5×10 -60…+125 0.13 20…60 -60…+125
path between a module surface and the heatsink may be as much as 100µm, and more with large
Grey/self-adhesive layer 0.6 5×1014 -60…+125 0.18 50…300 -60…+125 heatsinks. Critical applications may require a better surface flatness through additional machining
operations to reduce the thermal path.
Grey/self-adhesive layer 3 n/a n/a 0.11 n/a n/a
The viscosity and the rheology of the PCM above its melt point represent another factor determining the
thickness of the interface. As the PCM melts in the interface, the pressure applied by the mounting
clamps forces the liquid to spread, eliminating the interstitial air and allowing the space between the two
surfaces to decease. If the viscosity is high, the low force of the clamps will be insufficient to cause
5.2.2 Phase Change Gasket Materials (solid to liquid) sufficient spreading and the conduction path will be long. Low viscosity on the other hand will allow the
The inavertable heat produced by power electronics necessitates a carefully designed thermal path liquid to fill most of the joint, resulting in the thinnest joint. Using a stronger mounting force will aid the
along which all of the thermal resistances are minimized. For the case-to-heatsink interface, this spreading process, but there is a package limit as to the amount of pressure that can be applied.
requires that thermal grease be used to minimize the interface resistance. Phase change materials, Phase change materials offer the same thermal performance as thermal grease without the mess and
PCMs, are an alternative to the messy application and migration problems associated with thermal contamination associated with grease. They can be supplied attached to a heatsink as a dry film. As
grease. soon as they are heated above their phase change temperature, they melt and perform as well as, or
The term phase change describes a class of materials that are solid at room temperature and change to often better than, thermal grease. Once this interface has been formed, it remains stable until the sink is
a liquid as temperature increases. This phase change, or melting, occurs in the range of 40 to 70°C. physically separated from the power module case-mounting surface.
PCMs are composed of a mixture of organic binders, fine particle ceramic fillers for thermal
enhancement, and, optionally, a supporting substrate, such as aluminium foil or a woven glass mesh.
The organic binder is a blend of a polymer and a low-melt-point crystalline component, such as a wax.
The ceramic fillers may be Aℓ203, BN, AℓN or Zn0. 5.3 Heat-sinking thermal resistance, Rθ s-a
The way a PCM performs compared to a dry interface joint and thermal grease is illustrated in figure 5.3,
where the case to heatsink temperature difference is plotted against elapsed time after the The thermal resistance for a flat square plate heat sink may be approximated by
commencement of power dissipated. The curve representing the dry interface shows rapid thermal 3.3 ¼ 650
equilibrium at about 13°C. The curve involving the use of thermal grease shows the same rapid rise to Rθ s-a = Cf + Cf (K/W) (5.19)
λb A
thermal steady-state but at a lower temperature difference of 4°C. The thermal grease significantly
reduces the interface resistance by eliminating the interstitial air.
The PCM - a 0.1 mm thick dry film - behaves as a combination of the two interfaces. Initially at power Typical values of heatsink thermal conductance λ in W/K cm at 350 K, are shown in Appendix 5.24 and
up, the cool components give the dry interface behaviour, with the temperature difference rapidly b is the thickness of the heat sink, mm
increasing to about 12°C. As the system temperature increases, the PCM melts and the clamping A is the area of the heat sink, cm2
pressure exerted by the clamping mechanism forces the liquid to spread in the thermal joint. As the Cf is a correction factor for the position and surface emissivity of the
heat-sink orientation according to Table 5.4.
141 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 142

Table 5.6: Emissivity coefficient of As the base flow height H is increased, the air at the top of a vertical heatsink is hotter than that entering
various surface treatments at 100°C at the bottom. If the fin depth L is increased, there is more mutual radiation between fins, and as the
Table 5.4: Heatsink correction factor spacing is reduced, mutual radiation increases further. Airflow is also restricted because of the smaller
Emissivity physical area for air to pass, since more of the available space is occupied by the heatsink itself.
Surface position Surface The performance of a heatsink is linearly proportional to the base width Whs of the sink in the direction
shiny blackened ε
Cf Polished aluminium 0.05 perpendicular to the flow and proportional to the square root of the fin base length H in the direction of
vertical 0.85 0.43 the airflow. (The heat transfer coefficient h is inversely related to H). Therefore it is better to increase the
Polished copper 0.07
horizontal 1.0 0.5 width rather than the length, provided the width is not already excessive compared to the length.
Rolled sheet steel 0.66
Oxidised copper 0.70 Heat transfer coefficient h can be defined in a number of ways. If it is defined referenced to the inlet fluid
temperature of the heatsink, the heatsink thermal resistance is calculated by
Table 5.5: Fin spacing versus flow and fin length Black anodised aluminium 0.70 - 0.90
1
Black air-drying enamel 0.85 - 0.91 Rθ hs −a = (5.21)
ηf hA
Fin length (mm) H 75 150 225 300 Dark varnish 0.89 - 0.93
where A is the total surface area of fins and base between fins and
Airflow (m/s) v Fin Spacing (mm) s Black oil (organic) paint 0.92 - 0.96 ηf is the fin efficiency, defined as
natural convection 6.5 7.5 10 13 Al203 0.15 tanh ( mf × H ) h × P,
1.0 4.0 5.0 6.0 7.0 ηf = in which mf = (5.22)
mf × H λ × Ax
2.5 2.5 3.3 4.0 5.0
where H is the base height of the fin, m
5.0 2.0 2.5 3.0 3.5
P□ is the fin perimeter, m
Ax is the fin cross sectional area, m2
mf is the mass flow rate, equal to ρℓ ×Vf ×s×L, kg/s
The correction factor Cf illustrates the fact that black surfaces are better heat radiators and that warm air
ρℓ fluid density (= 1/ν specific volume), kg/m3
rises, creating a ′chimney′ effect. Equation (5.19) is valid for one power-dissipating device, in the centre
Vf is the velocity between the fins
of the sink, at a static ambient temperature up to about 45°C, without other radiators in the near vicinity. If the heat transfer coefficient is defined based on the temperature between the fins, the thermal
In order to decrease thermal resistance, inferred by equation (5.9), finned-type heat sinks are employed resistance expression involves a heat capacitance component:
which increase sink surface area. Figure 5.6 illustrates graphs of thermal performance against length for
1 1
a typical aluminium finned heat sink. This figure shows that Rθ s-a decreases with increased sink length. Rθ hs −a = + (5.23)
ηhA 2mf c p
If the fin thickness, t, is small relative to the fin spacing, s, the following equation can be used for where cp is the fluid specific heat capacitance at constant pressure, kJ/kg.K.
estimating the thermal resistance of a vertical heat sink in natural convection.
1 1 Estimating radiation heat transfer from an extruded heat sink
Rθ hs −a = = (5.20) The effect of radiation heat transfer (hence emissivity, ε) is important in natural convection, as it can be
h × total fin area h × ( 2nf LH ) responsible for up to 40% of the total heat dissipation. Unless the heatsink is facing a hotter surface
where a fin efficiency of unity has been assumed (see equation (5.22)) and the number of fins, nf, is nearby, it is imperative to have the heat sink surfaces thinly painted or correctly anodised to enhance
Whs radiation. In natural convection situations where the convective heat transfer coefficient is relatively low,
nf = based on the dimensional parameters in figure 5.4a, the radiation heat transfer from all surfaces of the
s +t
extruded heat sink can be calculated using
Minimal thermal reduction results from excessively increasing base length, H, as shown in figure 5.6b. 1
Rθ hs −a = (5.24)
The maximum distance between fins, s, depends on the fin depth, L, and width of the fins, t, with deep
finned heat sinks needing more space between adjacent fins than a shallow design, unless fan cooling { }
(nf − 1) ε a s + ε a nf t + 2 (L + B ) H σ (T s + T A ) (T s2 + T A2 )
is used. The minimum spacing s is determined by fin depth, L, and airflow. If the fins are packed too where nf is the number of fins
closely, the flow through them is significantly reduced and therefore the heat transfer coefficient, h, εa is the apparent emissivity of a channel
decreases. The deeper the fins, L, the more space needed between them since a portion of the heat is Ts is the heatsink surface temperature and
radiated to adjacent fins, which helps to stabilise the temperature, but does little to dispose of the heat TA is the ambient temperature
(in figure 5.4a, about 30% of the heat is radiation transferred fin-to-fin, hence not all dissipated).
3
L/H
Heat transfer co-efficient h Thermal resistance Rθ 1

εa
b 2.5
Whs decreases due to close decreases due to reduction
proximity of adjacent fins in sink surface area
Emissivity ε = 0.8 0.8
L 2

Apparent Emissivity
0.4
Heat sink thermal resistance

1.5 0.2
H
1 0
Rθhs-a

Ts
0.5
Optimal Emissivity ε = 0.08
fin spacing 0
0 2 4 6 8 10
L / Whs
s t g (a) (b) Inside fin spacing s
Figure 5.5. Apparent emissivity εa of a channel heatsink of two different surface emissivities
Figure 5.4. Heat sink dimension parameters and thermal resistance dependence on fin spacing. for different number of fins and dimensions.
143 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 144

The apparent emissivity is a function of heat sink dimensions and surface emittances of the sink The chimney effect results in an airflow velocity v, which increases further up the heatsink stack. This
material, as shown in figure 5.5, for two values of the surface emissivity, namely ε equals 0.08 and 0.8. and the air density increase results in the upper heatsink being the coolest, even though the passing air
The apparent emissivity εa is based on enclosure theory and assumes a diffused grey surface and is the warmest.
constant surface temperature.
The emissivity coefficient, ε, indicates the radiation of heat from a body according the Stefan-Boltzmann

mK/W
law, compared with the heat radiation from an ideal black body where the emissivity coefficient is ε = 1.
60
Regardless of the composition of the emitting surface, the microscopic (and macroscopic) roughness of
the surface causes differences in emissivity because a rougher surface has a larger emitting area.

typical substrate thermal resistance

Generally, the emissivity of most opaque emitting surfaces increases as wavelength becomes shorter. 50
The emissivity coefficient, ε, for some common surface qualities of aluminium and copper can be found
in the Table 5.6 and in Appendix 5.25.
The low emissivity coefficients of untreated, polished aluminium and copper means they have surface 40
finishes that limit the radiated heat from a body. Two thin coats of flat white Krylon #1502 (or equivalent)
which has an emissivity of 0.96, should be used on all untreated (emissivity-wise) areas. Al202
substrate
30
Unless otherwise stated, the heat sink is assumed anodised black (emissivity of up to 0.97) and Al3N4
vertically mounted with negligible thermal resistance from case to sink. In accordance with the data in substrate
Table 5.4, a general derating of 10 to 15 per cent for a bright surface and 15 to 20 per cent in the case
of a horizontal mounting position, are usually adopted. Figure 5.6b also shows the improvement effects 0 1 2 3 4 5 6
on dissipation due to the high thermal conductivity (heat spreader effect) of oxidised copper. water flow l/min

Figure 5.7. Improved cooling with:

(a) forced air cooled heat-sink - relative thermal resistance improvement with surface airflow and
anodised (b) thermal resistance dependence on liquid cooling flow rate and substrate material.

5.4 Modes of power dissipation

fan Al
cooled
Cu For long, >1ms, high duty cycle pulses the peak junction temperature is nearly equal to the average
junction temperature. Fortunately, in many applications, a calculation of the average junction
temperature is sufficient and the concept of thermal resistance is valid.
Other applications, notably switches driving highly reactive loads, may create severe current-crowding
conditions which render the traditional concepts of thermal design invalid. In these cases, transistor safe
operating area or thyristor di/dt limits must be observed, as applicable.
Cu Cu
Al Al In yet other applications, heat cycling can cause power module faults, hence device failure, due to
• thermal cycling – is associated with large base plate (case) temperature changes, ∆Tc
• power cycling – is associated with large junction temperature changes, ∆Tj

The die is connected to a low thermal impedance substrate, usually utilising copper in the form of so-
called direct copper bonding, DCB, as shown in figure 5.8a and the forced water cooling effectiveness is
shown in figure 5.7b.
Direct copper bonding
Direct copper bonding DCB is a process in which copper (on each side) and a ceramic material, usually
either aluminium oxide (Aℓ2O3) or aluminium nitride (AℓN), are fused (bonded) together at high
temperature.
Figure 5.6. Heat-sink typical data (for aluminium and copper): The properties of DCB substrates are
(a) cross-section view; (b) heat-sink length versus thermal resistance for a matt black surface finish; • High mechanical strength and stability
(c) temperature rise versus dissipation for an anodised finish and different lengths; and
• Good cohesion and corrosion resistance
(d) as for (c) but with a matt black surface finish.
• High electrical insulation
• Excellent thermal conductivity
Thermal resistance increases with altitude, z, above sea level, as air density decreases, according • Reliable thermal cycling stability
to Rθ ( z ) = RO metres / (1 − 5 × 10 −5 z ) . For example: a 1°C/W heatsink degrades to 1.11°C/W at an altitude of • Matched thermal expansion coefficient to silicon and gallium arsenide
2,000 metres, or 1.18° C/W at 3,000 metres. • Good heat spreading
The effective sink thermal resistance can be significantly reduced by forced air cooling, as indicated in • Processable, e.g. copper is etchable and millable like a pcb
Table 5.5, figure 5.7 and by equations (5.7) and (5.8). If the airflow is • Environmentally friendly
• laminar, heat loss is proportional to the square root of air velocity, equation (5.7); • High copper current density
• turbulent, heat loss is proportional to velocity to the power of 0.8, equation (5.8).
The advantages of DBC substrates are high current carrying capability due to thick copper metallization
When heatsinks (dissipating a total power of PDtotal) are vertically stack to share the same vertical natural and a thermal expansion close to the silicon at the copper surface due to high bond strength of copper
convention airflow, the air temperature of the flow at the upper heatsink, after passing n-1 heatsinks, is to ceramic. The DCB process yields a super-thin base and eliminates the need for the thick, heavy
n −1 c v
copper bases that were used prior to this process. Because power modules with DCB bases have fewer
T air = T amb + (5.25) layers, they have a much lower thermal resistance. Because the expansion coefficient matches silicon,
n PD total they have much better thermal cycling capabilities (up to 50,000 cycles). (See Appendix 5.27)
 145 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 146

The drawback of standard DCB substrates for high voltage applications is a start of partial discharge at Power cycling
relatively low voltages. Therefore substrates using expensive metal brazing technologies (AMB) are Rapid cycling of the chip junction temperature causes mechanical stress around the silicon chip to
mainly used in high voltage semiconductor modules for traction applications. The initiation voltage for a aluminium wire bond interface, due to their different linear expansion temperature co-efficients.
ceramic thickness of 0.63mm is less than 4kV. Main causes for this behaviour are small voids between Eventually a crack occurs on the silicon side of the interface, as indicated in figure 5.8a. Short rapid
the copper and ceramic and blurred straight border lines of the copper conductors at the copper/ceramic junction temperature changes, over tens of seconds, of ∆Tj =100K, can lead to failure within 2500
interface. Precision etching technology can alleviate these disadvantages. The other disadvantage of cycles. The number of cycles to failure increases by just over an order for every 10°C decrease in ∆Tj.
DCB is its deficiency for thermal shock because of the large residual stress on the substrate surface due In a related thermal application, where the power dissipated in the semiconductor consists of pulses at a
to the coefficient of thermal expansion CTE mismatch of alumina and copper. low duty cycle, the instantaneous or peak junction temperature, not average temperature, may be the
limiting condition. Figure 5.9 shows by comparison such a condition, where the operating frequency, not
Thermal cycling the maximum power dissipated, is dominant in determining junction temperature. In this case thermal
Intermittent equipment operation, start-up, and shutdown in extreme temperature conditions may cause impedance Zθ j-c is used instead of thermal resistance Rθ j-c such that Zθ j-c = r(tp) Rθ j-c, where r(tp) is the
power module thermal stresses due to the different linear expansion temperature co-efficients of the normalising factor yielded from the normalised transient thermal impedance curves for the particular
materials associated with the soldered substrate mounting to the copper base plate in multi-chip large device. Appropriate values for the rectangular power pulse width tp and duty cycle factor δ are used.
area packages (see Tables 5.32 and 5.33 in Appendix 5.24). Large base plate (case) temperature
changes in excess of 80K over a few minutes, stress the hard solder bonding between the copper base The power devices employed in power electronics are usually used in some form of on/off power pulse
plate and the insulating substrate (usually AℓN or Aℓ203), as shown in figure 5.8a. This fatigue leads to waveform mode. The following power waveforms are analysed:
eventual crack failure after a finite number of cycles N, as shown in figure 5.8c, approximated by • Periodic rectangular power pulses (steady-state thermal response);
k • Single rectangular power pulse;
N = (5.26) • Composite rectangular superimposed power pulses;
A × ∆T 2

where A is the die area and ∆T is the thermal shock temperature change. The constant k depends on • A burst of rectangular power pulses; and
the package, type of hard soldering, etc. Large, multiple die IGBT modules suffer from thermal shock • Non-rectangular power pulses.
limitations and relatively low reliability, because of the sheer large number of die soldered to the
substrate over a large base plate copper area in the module. 5.4.1 Steady-state response
Figure 5.8b shows how the number of thermal cycles to fracture for DCB substrates varies with copper
thickness, when cycled between -40°C to +110°C. For a case temperature change of ∆T = 80K, lifetime Large cycle-by-cycle junction temperature fluctuations occur at low frequencies, figure 5.9a. As
can be as low as 3,500 cycles and may only involve powering up and shutting down the associated frequency increases, thermal inertia of the junction smoothes out instantaneous temperature
equipment. Thermal cycling is normally performed by cycling the inactive package between the fluctuations, as shown in figure 5.9b, and the junction responds more to average, rather than peak
maximum and minimum storage temperatures. Although Aℓ/SiC is far superior to copper from a power dissipation. At frequencies above a kilohertz and duty cycles above 20 per cent, cycle-by-cycle
differential thermal expansion perspective, its thermal conductivity is only a little better than that of temperature fluctuations usually become small, and the peak junction temperature rise approaches the
aluminium. Floating silicon wafers in disc type packages suffer to a much lesser extent (an order) from average power dissipation multiplied by the steady-state junction-to-case thermal resistance, within a
the effects of differential thermal expansion when thermally cycled. few per cent.
power cycle Because of thermal inertia (long thermal time constant), the heat sink and package case respond only to
crack average power dissipation, except at ultra low frequencies, < 1Hz. The steady-state thermal conditions
Si chip Aluminium
wire
for the case-mount and junction (equation (5.12)) are given by
solder layer
Tl − T c Tc −Ta
Cu foil Pd = j = (W) (5.27)
thermal cycle
Rθ j −c Rθ c-s + Rθ s-a
insulation substrate with
adhered copper foil both crack where Pd is the average power dissipation, which is the maximum power multiplied by the on-time duty
sides cycle δ for rectangular power pulses. The difficulty in applying equation (5.27) often lies in determining
Cu foil
the average power dissipation.
solder layer
Cu base plate

(a)
tp t2
4
10
T2 T2
number of thermal cycles until failure

(b) (c) 80ºC

T1 T1
3
10

T
2
10

10

1
0 0.2 0.4 0.6 0.8 1 1.2
copper thickness mm

Figure 5.8. Direct copper bonding:

Figure 5.9. Waveforms illustrating that peak junction temperature is a function of switching
(a) sectional view of power module substrate showing boundary regions where power cycle cracking
frequency: (a) lower switching frequency with 10 ms pulse and a 20 per cent duty cycle and
and thermal cycle cracking, occur; (b) copper thickness affect on power failure; and
(b) high frequency and 1 ms pulse with a duty cycle the same as in (a).
(c) power life thermal cycling.
147 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 148

5.4.2 Pulse response which yields ∆Tl = 551.6K. That is, after a long period (>>10ms) the junction temperature would
increase by 551.6K. From equation (5.30), this temperature rise corresponds to continuous power of
When a junction dissipates power associated with a single pulse, the junction temperature increases
during the pulse and decays to the original temperature after the energy pulse ceases. The junction ∆Tl 551.6 K
Pd = = = 1103.3 W
temperature variation may vary from ambient temperature to a level above the normal maximum Rθ 0.5 K/W
operating limit, a change of over 150°C. The upper temperature due to the power pulse can cause In 10ms the temperature must only rise 100K, hence, from equation (5.32) the transient thermal
silicon damage, if the maximum allowable limit is exceeded too often or by a large amount on just a impedance Zθ is
single occasion. ∆T 100 K
Z θ = r t p Rθ = ( )
Pd
=
1103.3 W
= 0.091K/W
Equation (5.2) is valid for one dimensional steady state and transient thermal conditions, and the
transient temperature equation is given by the first order solution to
Thus the thermal resistance Rθ is modified, or normalized, by
λA δT
Pd = − T + ρm AA (W) (5.28) Z 0.0001K/W
A δt r (10ms ) = θ = = 0.181
The time domain solution for the temperature rise is
Rθ 0.5 K/W
Table 5.7 shows the normalised thermal impedance factor, r(tp), for other pulse durations, which are
∆T (t ) = ∆Tl × 1 − e −t / τ ( ) (5.29)
plotted in the accompanying figure. Notice the similarity of the single pulse results given for a practical
where the maximum temperature eventually attained if the power pulse were maintained, above power device in figure 5.10.
ambient, is
P A P Table 5.7: Single pulse data
∆Tl = d = d = Pd Rθ
λA hA
(K ) (5.30)

and the thermal time constant ∆Tl Pd Zθ r (tp)

tp
t →∞ = ∆Tl / Rθ = ∆T / Pd = Zθ / Rθ
ρ A2 thermal capacity, J/K
τ = m =
λ power per K, W/K
(s ) (5.31)
pulse time temperature rise power dissipated thermal impedance normalised
s K W K/W pu
The transient thermal impedance Zθ is defined as 1 100 200 0.5 1
∆T ∆Tl × 1 − e p
−t / τ
( )
( )
Z θ = r t p Rθ =
Pd
=
∆Tl
−t / τ
= 1 − e p Rθ ( ) (5.32)
0.1
0.01
116
552
231
1103
0.432
0.091
0.86
0.181
Rθ
0.001 5050 10100 0.0099 0.0198
That is, thermal resistance Rθ is modified by the factor r(tp) to yield transient thermal impedance Zθ:
r (t p ) = 1 − e ( −t p / τ
) (5.33)
0.0001
0.00001
50050
500050
100100
1000100
0.0010
0.0001
0.0020
0.0002
This one-dimensional solution assumes a homogeneous thermal conducting material with a single point
heat source, producing a uniform heat flow path. Since the practical case is far from ideal,
manufacturers provide data for dynamic temperature effects based on the concept termed thermal
Th e r m a l Im p e d a n c e
impedance. The thermal solution given by equation (5.29) gives acceptable results when applied to solid
carbon resistors (being a homogeneous material), as considered in Chapter 25 (specifically, see 1
Example 25.7).

Example 5.1: Semiconductor single power pulse capability 0.1

pu
A semiconductor has a thermal capacity (mc) of 0.1J/K and a steady state thermal resistance to its case
of Rθ = 0.5 K/W. If the junction temperature is not to exceed 125°C in a 25°C ambient, determine the single pulse
0.01
allowable power dissipation, hence transient thermal impedance, as a function of single power pulse 1−e
−t p / 0.05s

r = Z/R
duration. Plot the results for five time decades, decreasing from 1s.

Solution 0.001

The power dissipation per K is

1 1
Pd / K = = = 2 W/K 0.0001
Rθ 0.5 K/W 0.00001 0.0001 0.001 0.01 0.1 1
From equation (5.31) the thermal time constant τ is given by sin g le p u lse w id th , t p (s)
thermal capacity, J/K 0.1 J/K
τ = = = 0.05s ♣
power per K, W/K 2 W/K
After time tp, the junction temperature rise from 25°C must not exceed 125°C, that is ∆T(tp) = 100K, thus
equation (5.29) gives Figure 5.10 shows the thermal impedance curves for a power-switching device, normalised with respect
( ) ( )
∆T t = ∆Tl × 1 − e −t / τ = ∆Tl × 1 − e p
p
−t / 0.05s
(= 100K ) to the steady-state thermal resistance Rθ j-c. The curve labelled ′single pulse′ shows the rise of junction
temperature per watt of power dissipated as a function of pulse duration. The thermal impedance for
As a specific example of the procedure, consider a tp = 10ms energy pulse. repetitive pulses Z, of duty cycle δ, can be determined from the single pulse value z according to
(
∆T (10ms ) = ∆Tl × 1 − e −10ms / 0.05s = 100K ) Z θ (t p , δ ) = δ + (1 − δ ) z (t p ) (K/W) (5.34)
149 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 150

Equation (5.12) becomes 5.4.3 Repetitive transient response

Tl − T c Tl j − T c Minimal temperature variation occurs if the power switching period T is shorter than the junction to case
Pp = j = (W) (5.35) mount thermal time constant, T < τ, whence the concept of steady state thermal resistance is applicable,
Z θ (t p , δ ) r (t p )Rθ j −c
as presented in 5.4.1. When the relative magnitudes are reversed such that T > 5 τ, then the temperature
Note that the peak power Pp is employed, and then only for thermal analysis from the junction to the effects of the power pulse die away, and the single pulse transient thermal impedance approach
case thermal mounting. That is, Zθ j-c is the only thermal impedance term that exists. See problem 5.8. presented in 5.4.2 is applicable.
The transition or boundary between junction operation that can be assumed steady-state junction
Figure 5.10 shows that at the pulse width minimum extreme, tp < 1µs, as z(tp→0) →1 in equation (5.34): temperature operation (T < τ) and that of a series of discrete non-interacting single pulses (T > 5 τ) can
lim Z θ ( 0, δ ) = δ Rθ j −c = r (t p = 0 ) Rθ j −c (5.36) be analysed by extending the one-dimensional thermal transient equation (5.29) in conjunction with
t p →0

that is, r(tp→0)→δ. figure 5.9a. Figure 5.9a shows how the temperature increases from T1 to T2 during the time tp when
power is dissipated, and decreases from T2 to T1 during time t2 when no power is being dissipated by
Figure 5.10 also shows that at the pulse maximum extreme, that is, tp > 1s or continuous power the virtual junction. This increasing and decreasing of the junction temperature occurs cyclically over
dissipation, as z(tp→1) →0 in equation (5.34): each period T.
lim Z θ ( ∞, δ ) = Rθ j −c (5.37) Based on equation (5.29) the junction temperature increases exponentially according to
t p →∞

that is, r(tp→∞)→1, independent of duty cycle, that is, for all duty cycles.
(
T (t ) = ∆Tl − ∆Tl − T e −t /τ 1 ) (5.38)
and decreases exponentially according to
T (t ) = T 2 e −t /τ (5.39)
where the thermal time constant τ and maximum possible junction temperature rise are defined by
equations (5.31) and (5.30), respectively. Since these temperature variations are in steady state, the
temperature constants T1 to T2 can be solve using the boundary conditions. This gives
−t / τ
1−e p
T 2 = ∆Tl and T1 = T 2 e −t 2 /τ (5.40)
1 + e −T /τ

The junction temperature swing, ∆T is

∆T j = T 2 − T1 = ∆Tl
(1 − e −t p / τ
) (1 − e −t 2 / τ
)
(5.41)
1+e −T / τ

The maximum variation in junction temperature occurs for square-wave power, that is tp = t2 = ½T, δ=½:
T 
∆T jmax = ∆Tl tanh   (5.42)
 4τ 
Figure 5.10. Transient thermal impedance curves;
This equation highlights that the magnitude of the temperature change is highly dependant on the power
normalised with respect to the steady state thermal resistance, Rθ j-c.
switching frequency 1/T relative to the thermal time constant τ of the semiconductor package.

Example 5.2: A single rectangular power pulse

Example 5.3: Semiconductor transient repetitive power capability
A semiconductor with a junction to case thermal resistance of 1 K/W absorbs a single 100W power
pulse for 20µs. Based on the thermal impedance characteristics in figure 5.10, what is the expected A semiconductor with a thermal capacity of 0.02J/K and a thermal resistance from the junction to the
junction temperature rise, assuming the case-mount temperature does not respond to this short pulse? case of ½K/W, dissipates 100W at a repetition rate of
i. 50Hz
Solution ii. 300Hz.
By calculating the worst-case junction temperature variation, indicate whether steady-state constant
The period for a single power pulse is infinite T → ∞, therefore the duty cycle t p / T is zero, δ = 0. junction temperature-based analysis (a thermal resistance approach) is a valid assumption.
∆T j −c = P × Z θ j −c = P × r (t p ) × Rθ j −c Solution
From figure 5.10, for a single 20µs pulse r(tp = 20µs) = 0.035. The junction temperature change is The long-term junction temperature rise with 100W continuous is given by equation (5.30), which yields
therefore
∆Tl = P R = 100W × ½K/W = 50K
∆T j −c = P × r (t p = 20µs ) × Rθ j −c
d θ

The thermal time constant τ is given by equation (5.31), giving

= 100W × 0.035 × 1K/W = 3.5K thermal capacity, J/K 0.02
The peak junction temperature will rise to 3.5K above the case mount temperature at the end of the
τ =
power per K, W/K
=
1
= 0.01 (s )
100W rectangular power pulse. ½
♣ Worst case temperature variation occurs with a 50% power duty cycle, as given by equation (5.42)
T   T 
The basic single rectangular power pulse approach can be extended to analyse composite rectangular ∆T jmax = ∆Tl tanh   = 50K × tanh  
 4τ   4 × 0.01s 
power pulses by algebraic superposition of a series of accumulating rectangular pulses. Because each
composite power pulse extends to the end of the temperature-calculating period, any positive From this equation:
rectangular pulse is subsequently cancelled by a negative power pulse. The technique is illustrated in at 50Hz, T = 20ms, ∆T jmax = 23.1K
example 5.4. at 300Hz, T = 3.33ms, ∆T jmax = 4.1K
The temperature variation of 4.1K at 300Hz is small compared to the maximum allowable junction
151 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 152

temperature, typical between 125°C and 175ºC, thus thermal analysis of this device in a 300Hz Table 5.8: Rectangular, composite pulse example data
application, can be thermal resistance calculation based as presented in 5.4.1. On the other hand, the
same device used in a 50Hz application will experience 5.6 times the junction temperature swing. This tx ty
23.1K variation represents a significant portion of the allowable junction operating temperature, and
tp1 tp2 tp3 tp4 tp5 tp6 tp7 tp8
could mean a thermal resistance approach is unsafe. A thermal impedance design approach is Pulse duration
recommended, as in example 5.2 and 5.4.2. 180µs 170µs 150µs 30µs 480µs 360µs 330µs 10µs
♣ single pulse δ = tp / T 0 0 0 0 0 0 0 0

T→∞ r(tp) 0.06 0.055 0.05 0.025 0.10 0.085 0.08 0.015
Example 5.4: Composite rectangular power pulses
period δ = tp / T 0.36 0.34 0.30 0.06 0.96 0.72 0.66 0.02
A semiconductor with a junction to case thermal resistance of 1 K/W absorbs the composite power pulse T = 500µs r(tp) 0.38 0.36 0.35 0.075 0.96 0.73 0.73 0.035
shown in figure 5.11. Based on the thermal impedance characteristics in figure 5.12, what is the
expected junction temperature rise at indicate times tx and ty, assuming the case temperature does not
In figure 5.11b, power pulse P1 = 40W lasts for 180µs, which represents a duty cycle of δ =180µs/500µs
respond to this short pulse? That is, the heatsink-case interface temperature is held constant.
= 0.36. The thermal impedance normalised factor of r(tp1=80µs) = 0.38 corresponds to δ = 0.45 in figure
What is the average junction to case temperature rise, in the repetitive case, f=2kHz? t p2 t p6
5.12.
r(tp) t p4 δ = tp / T t p4 t p3 t p1 t p5
Solution t p7
↓
1
tx
ty
Rectangular power pulses

0.5 0.5

normalising thermal factor

100

r(tp) = Zθj-c (tp, δ) / Rθj-c

80
0.2
(W)

60
40
0.2
20 0.1
0 0.1
0.05
P

0 50 100 150 200 450 500 550 0.05

0.02
time µs t
0.01
(a) 0.02
Single pulse
t1 = 180µs
0.01
t2 = 170µs
Composite rectangular power pulses

t3 = 150µs 10µs 20µs 50µs 100µs 200µs 400µs

t4 = 30µs square-wave pulse duration tp
160
140
(W)

120
P3 = 100W
100 Figure 5.12. Normalise junction to case-mount thermal impedance characteristics.
80
60 P2 = 20W
P

40
20 P1 = 40W
0 The junction temperature (rise) at tx, is given by
-20 P1 = 40W 450
-40
50 100 150 500 450 T jt−xc = P1Z θtpj 1−c − P1Z θtpj 2−c + P2 Z θtpj 3−c − P2 Z θtpj 4−c + P3 Z θtpj 4−c
-60 P2 = 20W time µs t

The junction temperature (rise) at ty, is given by

(b) t
T j −yc = P2 Z θtpj 5−c − P2Z θtpj 6−c + P3 Z θtpj 6−c − P3Z θtpj 7−c + P1Z θtpj 8−c
Composite rectangular power pulses

160
140 P1 = 40W
120
100 tx repetitive
80
T jt−xc = 40 × Z θtpj 1−c − 40 × Z θtpj 2−c + 20 × Z θtpj 3−c − 20 × Z θtpj 4−c + 100 × Z θtpj 4−c
(W)

60 P3 = 100W
40
20 P2 = 20W = 40 × 0.38 × 1K/W − 40 × 0.36 × 1K/W + 20 × 0.35 × 1K/W − 20 × 0.075 × 1K/W + 100 × 0.075 × 1K/W
0
-20 P2 = 20W = 13.8K
P

-40
50 100 150 200 350 400 450
-60 time t
-80 P3 = 100W tx single pulse
-100 T jt−xc = 40 × Z θtpj 1−c − 40 × Z θtpj 2−c + 20 × Z θtpj 3−c − 20 × Z θtpj 4−c + 100 × Z θtpj 4−c
-120

T = 500µs
= 40 × 0.06 × 1K/W − 40 × 0.055 × 1K/W + 20 × 0.05 × 1K/W − 20 × 0.025 × 1K/W + 100 × 0.025 × 1K/W
t5 = 480µs = 2.7K
t6 = 360µs (c)
t7 = 330µs
t8 = 10µs ty repetitive
t
T j −yc = 20 × Z θtpj 5−c − 20 × Z θtpj 6−c + 100 × Z θtpj 6−c − 100 × Z θtpj 7−c + 100 × Z θtpj 8−c
Figure 5.11. Composite power pulses: (a) original rectangular pulse;
(b) composite rectangular pulse, reference tx; and (c) composite rectangular pulse, reference ty. = 20 × 0.96 × 1K/W − 20 × 0.73 × 1K/W + 100 × 0.73 × 1K/W − 100 × 0.67 × 1K/W + 40 × 0.035 × 1K/W
= 12.2K
153 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 154

Figure 5.14 also shows the two rectangular pulse representations, where successive 50µs portions of
ty single pulse the triangle are represent by pulses, 100W, 37.5µs and 50W, 25µs, such that the total area is
t
T j −yc = 20 × Z θtpj 5−c − 20 × Z θtpj 6−c + 100 × Z θtpj 6−c − 100 × Z θtpj 7−c + 100 × Z θtpj 8−c maintained and the peak junction temperature rise occurs at the end of the power pulse sequence. The
two pulses are subsequently decomposed into three equivalent composite rectangular power pulses,
= 20 × 0.10 × 1K/W − 20 × 0.085 × 1K/W + 100 × 0.085 × 1K/W − 100 × 0.08 × 1K/W + 40 × 0.015 × 1K/W which sum at any time to give the original two rectangular pulses.
= 1.4K

In the repetitive composite pulse case, the average power dissipated over 500µs is

Power dissipated, Pd W

Power dissipated, Pd W
100µs, 100W 100µs, 100W
10µs × 40W + 120µs × 20W + 30µs × 100W 100
50µs, 62.5µs,
= 11.6W 100W 100 100W
500µs 80 80
The average junction to case mounting temperature rise is 60 60 25µs,
50µs, 50W

T j −c = Pave × Rθ j −c = 11.6W × 1K/W = 11.6K 50W

40 40
♣
20 20
0 0
Non-rectangular power pulses
The concept and characterisation of thermal impedance is based on rectangular power pulses. Non- 0 20 40 60 80 100 0 20 40 60 80 100
rectangular pulses are converted to equivalent energy, rectangular pulses having the same peak power, time t µs time t µs
Pp, of duration tp, as shown in figure 5.13. The resultant rectangular power pulse will raise the junction tp3
temperature higher than any other wave shape with the same peak and average values, since it tp1
concentrates its heating effects into a shorter period of time, thus minimising cooling during the pulse.
tp2
Worst case semiconductor thermal conditions result. Improved thermal accuracy is obtained if each non- 160 160
rectangular pulse is further sub-divided into numerous equivalent total energy rectangular pulses, as 140 140
considered in example 5.5. 120 tp 120

W
100 100 P2

Power composites, P

Power composites, P
80 80
60 60
40 40
20 20 P1
0 0
-20 0 20 40 60 80 100 -20 40 60 80 100
equal
time t µs P1 t µs
areas -40 -40
-60 -60

Figure 5.14. Transient thermal impedance curves;

normalised with respect to the steady state thermal resistance, Rθ j-c.

The thermal impedance normalising factor r(tp) for the applicable device can be read from figure 5.15,
using the pulse periods and duty cycles shown in figure 5.14, and are shown in Table 5.9.
Figure 5.13. Conversion of non-rectangular power pulse (a) into equivalent rectangular pulse (b).

Example 5.5: Non-rectangular power pulses r(tp) δ = tp / T t p2 t p3 t p1

↓
Switch losses are a series of triangular power pulses rising linearly to 100W in 100µs after switch turn- 1
on. If the thermal resistance of the junction to case mounting is 1 K/W and the thermal impedance 0.5
0.5
characteristics are represented by figure 5.15, calculate the case to junction peak temperature rise for

normalising thermal factor

r(tp) = Zθj-c (tp, δ) / Rθj-c
0.2
i. A single pulse, T → ∞ 0.2
ii. 50% power duty cycle, T = 200µs 0.1
iii. 10% duty cycle, T = 1000µs 0.1
0.05
0.05
0.02
Represent the triangular power pulses by firstly one equivalent rectangular power pulse, secondly two
equivalent rectangular power pulses, and compare the predicted peak junction temperature rise results. 0.01
Assume the case temperature is maintained at a constant temperature. 0.02
Single pulse
Where applicable, calculate the average junction to case thermal mounting temperature drop, Tj-c. 0.01
10µs 20µs 50µs 100µs 200µs
Solution
square-wave pulse duration tp
Each saw-tooth power pulse is represented by a single rectangular power pulse, 100W and 50µs
duration in figure 5.14, therein fulfilling the requirements of the same maximum power occurring Figure 5.15. Transient thermal impedance curves;
simultaneously in both waveforms and both containing the same energy, area. normalised with respect to the steady state thermal resistance, Rθ j-c.
155 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 156

Table 5.9: Non-rectangular, composite pulse example data sawtooth area

Pd =
T
½ × 100W × 100µs
One composite power pulse Two composite power pulses = = δ × 50 W
T
Thus the average case to junction temperature drop is
Pulse duration tp tp1 t p2 t p3
T j −c = P d × Rθ j −c
50µs 75µs 37.5µs 50µs
= δ × 50W × 1K/W = 50 × δ K
Single pulse δ = tp / T 0 0 0 0 For 50% and 10% duty cycles, this gives average temperature drops of 25K and 5K respectively.
T→∞ r(tp) 0.045 0.04 0.040 0.045
Both rectangular composite power pulse decomposition assumptions produce similar thermal results. At
50% duty cycle δ = tp / T ¼ ⅜ 0.188 ¼ cycle frequencies of 5kHz and 1kHz, together with high duty cycles, the peak junction temperature is
T = 200µs r(tp) 0.32 0.40 0.22 0.32 marginally higher than the average junction temperature. Using the concept of thermal resistance is
adequate under the switching frequency and duty cycle conditions of this problem.
10% duty cycle δ = tp / T 0.05 0.075 0.0375 0.05 ♣
T = 1000µs r(tp) 0.08 0.12 0.066 0.08

5.5 Average power dissipation

For a single pulse rectangular power waveform Two commonly used empirical methods for determining power dissipation Pd are
∆T j −c = P × Z θ j −c = P × r (t p ) × Rθ j −c • graphical integration and
• power superposition.
= 100W × r (t p = 50µs ) × 1K/W
For a single pulse, δ =0 5.5.1 Graphical integration
∆T j −c = 100W × r (t p = 50µs ) × 1K/W
Graphical integration may be formulated by digitally storing a complete cycle of test device voltage and
= 100W × 0.045 × 1K/W = 4.5K current under limiting steady-state temperature conditions. Each voltage and current time-corresponding
For a 50% duty cycle, δ = 0.5 pair are multiplied together to give instantaneous values of power loss. Numerical integration techniques
∆T j −c = 100W × r (t p = 50µs ) × 1K/W are then employed to give the average power dissipation.
= 100W × 0.32 × 1K/W = 32K 5.5.2 Practical superposition
For a 10% duty cycle, δ = 0.1
∆T j −c = 100W × r (t p = 50µs ) × 1K/W This technique is based on substituting a smooth dc voltage source for a complex waveform. A two-pole,
two-position switching arrangement is used, which firstly allows operation of the load with the device
= 100W × 0.08 × 1K/W = 8K
under test, until the monitored case temperature stabilises. Then, by throwing the switch to the test
mode position, the device under test (DUT) is connected to a dc power supply, while the other pole of
For a two pulse rectangular power waveform representation the switch supplies the normal power to the load to keep it operating at full power level conditions. The
∆T j −c = P1 × Z θ j −c t 1 − P1 × Z θ j −c t 3 + P2 × Z θ j −c t 2 dc supply is adjusted so that the semiconductor case temperature remains approximately constant when
= P1 × r (t p 1 ) × Rθ j −c t1 − P1 × r (t p 3 ) × Rθ j −c t3 + P2 × r (t p 2 ) × Z θ j −c t2
the switch is thrown to each position for about 10 seconds. The dc source voltage and current values
are multiplied together to obtain the average power dissipated.
= 50W × r (t p 1 = 75µs ) × 1K/W − 50W × r (t p 3 = 50µs ) × 1K/W + 100W × r (t p 2 = 37.5µs ) × 1K/W

For a single pulse, δ = 0 5.6 Power losses from manufacturers’ data sheets
∆T j −c = 50W × r (t p 1 = 75µs ) × 1K/W − 50W × r (t p 3 = 50µs ) × 1K/W + 100W × r (t p 2 = 37.5µs ) × 1K/W The total power dissipation Pd is the sum of the switching transition loss Ps, the on-conduction loss Pd,
= 50W × 0.04 × 1K/W − 50W × 0.045 × 1K/W + 100W × 0.04 × 1K/W drive input device loss PG, and the off-state leakage loss Pℓ.
= 3.75K The average total power loss is given by
1/fs
For a 50% duty cycle, δ = 0.5 Pd = f s ∫ v (t ) i (t ) dt (W) (5.43)
∆T j −c = 50W × r (t p 1 = 75µs ) × 1K/W − 50W × r (t p 3 = 50µs ) × 1K/W + 100W × r (t p 2 = 37.5µs ) × 1K/W
0

where fs is the switching frequency and v(t) and i(t) are the device instantaneous voltage and current
= 50W × 0.40 × 1K/W − 50W × 0.32 × 1K/W + 100W × 0.22 × 1K/W over one complete cycle of period 1/fs. The usual technique for determining total power loss is to
= 26K evaluate and sum together each of the individual average power loss components.
For a 10% duty cycle, δ = 0.1
5.6.1 Switching transition power loss, Ps
∆T j −c = 50W × r (t p 1 = 75µs ) × 1K/W − 50W × r (t p 3 = 50µs ) × 1K/W + 100W × r (t p 2 = 37.5µs ) × 1K/W
= 50W × 0.12 × 1K/W − 50W × 0.08 × 1K/W + 100W × 0.066 × 1K/W Figure 5.16 shows typical power device voltage-current switching waveforms. Normally an exact solution
is not required and an approximation based on straight-line switching intervals is usually adequate.
= 8.6K
For a resistive load, as derived in Chapter 6
The average junction to case temperature for a single pulse is zero, and the average junction Ps = 61V s I mτ f s (W) (5.44)
temperature is the heatsink/ambient temperature. and for an inductive load, as derived in Chapter 6
The average junction to case temperature during repetitive operation is independent of whether one or Ps = ½V s I mτ f s (W) (5.45)
two composite rectangular pulses are used to analyse the saw-tooth pulse pulses, since both model the
where τ is the period of the switching interval (both on and off), and Vs and Im are the maximum voltage
same original power waveform, each having the same waveform area, energy. The junction to case
and current levels as shown in figure 5.16. Switching losses occur at both turn-on and turn-off.
temperature is dependent on the duty cycle, which specifies the average power dissipation.
157 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 158

5.6.2 Off-state leakage power loss, Pℓ 5.7 Heat-sinking design cases

During the switched-off period, a small, exponentially temperature dependent current Iℓ, will flow through Heat-sink design is essentially the same for all power devices, but the method of determining power loss
the switch. The loss due to this leakage current is varies significantly from device type to device type. The information given in data sheets, in conjunction
PA = IAV s (1 − δ ) (W) (5.46) with the appropriate equation in Table 5.11, allows the designer to calculate power semiconductor
where δ is the on-time duty cycle of the switch. Normally Pℓ is only a small part of the total loss so that thermal rating for a variety of conditions.
the error in neglecting Pℓ is not significant. Generally, heatsink design is more readily visualised if a thermal equivalent electrical circuit model
approach is adopted, as shown in figure 5.1. The equivalence of parameters is shown in Table 5.10. The
examples to follow illustrate the approach.

Table 5.10: Thermal equivalent electrical circuit parameters

thermal parameter thermo-electric model magnetic model

temperature degrees potential magneto motive Amp-
∆T Volts ∆V ℑ
drop Kelvin difference force turns
power
Watts P current flow Amps I flux Wb Φ
dissipated
thermal Ohm’s Amp-
K/W Rθ Ohms R reluctance ℜ
resistance resistance turns/Wb

Figure 5.16. Typical voltage and current at turn-off switching transition for:
(a) an inductive load and (b) a resistive load. Current and voltage are interchanged at turn-on. 5.7.1 Heat-sinking for diodes and thyristors

At low switching frequencies (<100 Hz), switching loss can be ignored, so that in the case of rectifying
5.6.3 Conduction power loss, Pc diodes or converter-grade thyristors, 50 to 60 Hz, switching loss can usually be ignored. Fast-recovery
power diodes switching at less than 500Hz can also have switching losses neglected at low VA levels.
The average conduction power loss under a steady-state current condition is given by
Pc = δ I onVon (W) (5.47) 5.7.1i - Low-frequency switching
although equation (5.43) is valid in the general case when the integration is performed over the interval At a given current level IF and on-time duty cycle δ, on-state power loss can be read directly from the
corresponding to δ. manufacturers’ data. Figure 5.17a illustrates loss for square-wave power pulses, while figure 5.17b
The conduction loss for the MOSFET is usually expressed in terms of its on-state resistance (equations illustrates loss in the case of half-wave sinusoidal current. Figure 5.17b gives energy loss per cycle,
(3.16) and (4.12)) which may be converted to power when multiplied by the sinusoidal pulse frequency.
Pc = δ I d2(rms )Rds (on ) Thyristor loss due to the current waveform initial rate of rise of current, di/dt, can be incorporated and its
T −25°C contribution is added into the manufacturers’ conduction loss data for a given device type.
 α j (5.48)
 δ I d2(rms )Rds (on ) (25°C) 1 +  (W) 5.7.1ii - High-frequency switching
 100 
At device operating frequencies greater than about 100 Hz, fast-recovery diodes are normally employed
where α is the temperature coefficient of the on-state resistance, which is positive. A linear resistance
and at about 500Hz, switching losses must be added to the on-state conduction loss. Diode turn-off loss
approximation of equation (5.48) is accurate above 25°C if α is small, such that Pc can be approximated
is usually more significant than turn-on loss. Manufacturers provide maximum reverse recovery charge,
by
QR, characteristics as shown in figure 5.18. The reverse recovery charge is a linear function of tempera-
Pc ≈ δ I d2(rms )Rds (on ) (25°C) {1 + α (T j − 25°C)} (W) (5.49) ture and between the given junction temperatures of 25°C in figure 5.18a and 150°C in figure 5.18b,
interpolation of QR is used.
5.6.4 Drive input device power loss, PG
The reverse recovery W.s/pulse, JR, can be approximated by
A portion of the drive power is dissipated in the controlling junction or, in the case of the MOSFET, in the J R = V RQ R (J) (5.52)
internal gate resistance. Usually more power is dissipated in the actual external drive circuit resistance. where VR is the reverse voltage applied to the diode just after turn-off. Losses are lower since the diode
Drive input loss is normally small and insignificant compared with other losses, and can usually be only supports voltage once peak reverse recovery has occurred (at the peak of the reverse current). The
ignored. reverse recovery average power loss is given by
Two possible exceptions are:
• One notable exception is in the case of the power GTO thyristor, where continuous gate drive is
Ps ≈ VRQR f s (W) (5.53)
used to avoid loss of latching or when the holding current is high. The holding current can be 3%
of the anode current thus, the gate to cathode junction loss can be included in the total loss
calculation for better accuracy. Thus, for a gate junction voltage VGC the gate losses are given by The total average power loss is the algebraic sum of the steady-state conduction loss and the recovery
Pg = δ I G VGC (5.50) loss.
The recovery loss of the gate commutated thyristor (GCT) cathode junction can be included since
it is significant because the full anode current is extracted from the gate, thus is involved in
recovery of the cathode junction.
• A second exception is the MOSFET and IGBT at high switching frequencies, >50kHz, when the
loss in the device, associated with providing the gate charge QT is given by equation (4.36):
RG int
PG (Rint ) = Vgg QT f s (W) (5.51)
RG int + RGext
159 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 160

Table 5.11: Power rating equations based on thermal considerations

(c) Constant high

frequency and
constant duty
cycle

P T
T
δ = tp / T
tp
(d) Constant low Tj Ttp
Pd Ttp – Ta = Pd zθ(tp)
frequency and
constant duty
cycle

t
Figure 5.17. Diode on-state energy loss at low frequency as a function of forward current for: Ta t
0 tp T
(a) squarewave power pulses and (b) sinusoidal power pulses.

(e)

2 2

Example 5.6: Heat-sink design for a diode

1 1 A fast-recovery diode switches 60 A rectangular current pulses at 10kHz. The off-state bias is 400V and
the external circuit inductance limits the reverse dIF /dt to 100A/µs. If the device junction-to-case thermal
resistance is 0.7K/W, calculate the minimum heat-sink size requirement with a 50 per cent conduction
duty cycle, if the maximum ambient temperature is 40°C.

20 50 100 500 1000 20 50 100 500 1000 Solution

The steady-state loss given from figure 5.17a is about 40 W when using IF(AV) = 30A for δ = 0.5.
Minimum possible heat-sinking thermal resistance requirements occur when Tj is a maximum, that is
150°C from figure 5.18b. From figure 5.18b, for dIF /dt = 100 A/µs and IF = 60A, the maximum reverse
recovery charge is 1.3µC. The switching power loss (over estimate) is given by
Figure 5.18. Reverse recovery charge as a function of forward current and dIF/dt at: Ps = QRVR f s
(a) 25°C and (b) 150°C junction temperature.
= 1.3µC×400V×10kHz = 5.2W
161 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 162

The total power loss is therefore 5.7.3 Heat-sinking for power MOSFETs
Pd = 40 + 5.2 = 45.2W
Since the frequency and duty cycle are both high, the concept of thermal resistance is appropriate; that Switching losses in MOSFETs tend to be low at frequencies below 20 kHz and therefore may be
is neglected, along with gate and off-state losses. Conduction loss is generally expressed in terms of the
T j = T a + Pd (Rθ j-c + Rθ c-a ) on-state resistance as I2R loss. The first step in the thermal design is to determine the total power
dissipation in the device, which is generally dominated by the conduction loss. Determination of this loss
is not trivial since, while the power dissipation determines junction temperature, the power dissipation
Therefore 150V = 40V + 45.2A × (0.7Ω + Rθ c-a ) (in units of the electrical dual) itself is a function of junction temperature, because the on-state resistance increases with temperature,
as shown in figure 3.13.
whence Rθ c-a = 1.73 K/W

Figure 5.6b shows that a minimum of 50mm length of matt black heat sink is required. This assumes Example 5.8: Heat-sink for a MOSFET - repetitive operation at high peak current, low duty cycle
that the case-to-sink thermal resistance is negligible. In order to improve device reliability and lifetime,
Find the thermal resistance of the heat sink needed for a MOSFET conducting a repetitive 20A
operation at Tlj is avoided. A derating of 40 to 50°C significantly reduces junction thermal fatigue and
rectangular current waveform. On-time is 10µs, duty cycle is 0.1 per cent, and the maximum ambient
can result in a tenfold improvement in reliability. To restrict Tl j to 100°C, Rθ c-a= 0.7K/W, necessitating
temperature is 40°C. Assume Rds (on) at 150°C and 20 A is 5 Ohms, and Rθ j-c = 1.5 K/W.
120 mm of the heat sink as characterised in figure 5.6b. The flatness of the Rθ s-a curve means that the
effectiveness of the heat sink is diminished and either a wider sink of the same length or a shorter length Solution
of a profile offering lower thermal resistance would be more effective in reducing device thermal fatigue.
♣ Since the on-state duty cycle and switching frequency are both low, the peak junction
temperature at the end of the on-period will be significantly different from the average junction
temperature. The concept of thermal resistance from the junction to the case is therefore invalid;
5.7.2 Heat-sinking for IGBTs rather the concept of thermal impedance is used.
The peak power per pulse = Pp = I2R = 202 × 5Ω = 2 × 103 W
Externally, the IGBT conduction loss is related to the gate voltage and the collector current magnitude, Using a thermal impedance basis, the case temperature is given by
which specify the on-state voltage. No simple power loss characteristic is possible, as in figure 5.17 for T j = T c + Pp × Z θ j-c
the diode and thyristor. Fortunately, the power switching IGBT is used in such a way that its on-state
= T c + Pp r (t p ) Rθ j-c
collector-emitter voltage is near constant, whence conduction loss is given by
where r(tp) is the transient thermal impedance factor for the junction-to-case. For a 10 µs pulse
Pc = δ v ce I c (W) (5.54)
from figure 5.10, r(tp) = 0.03, assuming δ = 0.001≈ a single pulse condition, thus
150°C = T c + 2 × 103 × 0.03 × 1.5 = T c + 90°C
that is T c = 60°C
Example 5.7: Heat-sink design for an IGBT- repetitive operation at a high duty cycle
The average junction temperature is
A power IGBT is used to switch a 20A, 100V inductive load at 10 kHz. The transistor maximum on-state T j = T c + P d Rθ j −c = T c + δ Pd Rθ j −c
duty cycle is 90 per cent and the device has a junction-to-case thermal resistance of 0.7K/W. The = 60°C + 0.1% × 2 × 103 W × 1.5°C/W
transistor on-state voltage is maintained at 2V and the switch-on and switch-off times are 1 and 2 µs
respectively. If the junction temperature is not to exceed 125°C with a maximum ambient temperature of = 60°C + 3°C = 63°C
35°C, what is the minimum heat-sink requirement? Assume that the transistor is in a T0247 package, Although the average junction temperature is only 3°C above the case temperature of 60°C, the peak
which is mounted directly on the heat sink but with interface silicone grease used. junction temperature reaches 150°C.
Because of the heat-sink thermal inertia, the concept of thermal resistance and average power are
Solution used for calculations involving the heatsink. That is
T c = T a + P d Rθ c-a = T a + δ Pd Rθ c-a
Since both the duty cycle and switching frequency are high, the peak junction temperature is
approximated by the average junction temperature. That is, the concept of thermal resistance is 60°C = 40°C + 0.001 × 2 × 103 × Rθ c-a
valid. thus Rθ c-a = 10 K/W
The on-state power loss is given by The heat sink of cross-section shown in figure 5.6a is not suitable in this application, and one of a
Pc = δ v ce I c = 0.9 × 2V × 20A = 36 W much smaller surface area is applicable. A heatsink may not be necessary since the package
From equation (5.45), the switching losses for an inductive load are thermal resistance Rθ c-a, shown in figure 5.1, may be less than 10K/W, there in satisfying equation
(5.17). See problem 5.6.
Ps = Ps (on) + Ps (off)
♣
= ½×100V × 20A × (1µs + 2µs) × 10 kHz = 30 W
Total power losses Pd are 36W+30W = 66 W.
From If the junction operating temperature is unknown but can be assumed greater than 25°C, from equation
Tl j = T a + Pd (Rθ j-c + Rθ c-s + Rθ s-a ) (5.49), the total power loss can be expressed as
Pd = Po + I d2(rms)Rds (on) (25°C) {1 + α (T j − 25°C)} (W) (5.55)
125°C = 35°C + 66W × ( 0.7 + 0.1 + Rθ s-a )
where Po represents all losses other than the conduction loss, and is assumed temperature
Therefore Rθ c-a = 0.56 K/W independent. The temperature coefficient α for Rds (on)(25°C) is positive, typically 1 per cent/K as
indicated in figure 3.13. The usual thermal equality holds, that is
The case-to-heat-sink thermal resistance value of 0.1K/W for a T0247 non-insulated case using T j = T a + Rθ j-a Pd (K) (5.56)
silicone thermal grease was obtained from Table 5.1. To obtain the minimum heat-sink thermal Combining equations (5.55) and (5.56) by eliminating Tj yields
resistance of 0.56K/W, 150 mm of the heat sink with cross-section shown in figure 5.6a is required.
P + I d2(rms)Rds (on) (25°C) {1 + α (T a − 25°C)}
Clearly, a sink profile that has a lower thermal resistance per unit length would be more suitable. Pd = o (W) (5.57)
♣ 1 − I d2(rms)Rds (on) (25°C)α Rθ j-a
163 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 164

The denominator yields an asymptotic maximum drain current of T hs − T a = 42°C − 30°C = Rθ hs −a × ( 40W + 20W )
TDiode TTransistor
1
I d (rms) = (A) (5.58) Rθ hs −a = 42°C − 30°C = 12°C = 0.2 K/W
Rds (on) (25°C) α Rθ j-c 40W + 20W 60W 20W PD PT 40W
at which current thermal runaway would result. In practice, insufficient gate voltage is available and the RθDj-c RθTj-c
ii. Assume device losses are not affected by temperature and the
device would leave the constant-resistance region and enter the constant-current region, where the 0.8K/W 0.7K/W
heatsink thermal resistance is decreased to ½×0.2 = 0.1 K/W, then
above analysis is invalid.
T hs − T a = T hs − 30°C = 0.1K/W × ( 40W + 20W ) RθDc-hsc RθTc-hs
0.6K/W 0.5K/W
T hs = 0.1K/W × ( 40W + 20W ) + 30°C = 36°C Ths
Example 5.9: Heat-sink design for a MOSFET - repetitive operation at high duty cycle The device junction temperatures are given by
A power MOSFET switches 5 A rms at 10 kHz with a maximum on-state duty cycle of 90 per cent. The T Dj − 36°C = 20W × ( 0.8K/W + 0.6K/W ) that is T Dj = 64°C Rθhs-ac
junction-to-case thermal resistance is 0.7 K/W, the maximum ambient temperature 35°C, and on-state TTj − 36°C = 40W × ( 0.7K/W + 0.5K/W ) that is TTj = 84°C Ptotal
resistance at 25°C is 1 Ohm. If the heat-sink arrangement yields an effective case-to-ambient thermal 64W Ta
=30°C
resistance of 1.3 K/W and α = 0.01 /K, what is the junction operating temperature?
Solution The junction temperature of each device has decreased by about 6°C, so although the lifetime will have
increased, lifetime improvement is not doubled. Device package thermal properties are more dominant
Since the switching frequency and duty cycle are both relatively high, the thermal resistance concept than the heatsink in determining junction temperatures.
based on average junction power dissipation is valid.
iii. If the on-state duty cycle is δ and the instantaneous device losses are both P (since the on-state
Assuming zero losses other than conduction losses, then Po = 0. Equations (5.55) and (5.56) voltage is the same for both devices and the current is constant hence the same for both when each
rearranged to eliminate Pd yield device conducts) then
mosfet δ P = 40W
T a + Rθ j-aI d2(rms)Rds (on) (25°C) {1 − 25α }
Tj = (W) (5.59) diode (1 − δ ) P = 20W
1 − α Rθ j-aI d2(rms)Rds (on) (25°C)
Assuming typical α = 0.01/K and Rθ j-a= Rθ j-c+ Rθ c-a Summing these two equations gives an instantaneous loss of P = 60W, whence a switch on-state duty
cycle of δ = , that is the switch conducts for 66% of the cycle period. The diode on-state voltage is
35°C + 2 × 5 × 1Ω × (1 - 25 × 0.01)
2
therefore 60W/30A = 2.0V and the MOSFET on-state resistance is 60W/30A2 = 67mΩ.
Tj= = 145°C
1 - 0.0l × 2 × 52 × lΩ ♣

♣
Example 5.11: Six thermal elements (on a common substrate) in a common package

Example 5.10: Two thermal elements on a common heatsink A three-phase full-wave diode rectifier package consists of six-diode die within a single module. The
junction-to-case thermal resistance of each die is 0.24K/W. The module is mounted on a heatsink with a
A dc chopper has a MOSFET switch that dissipates 40W and a load freewheel diode that dissipates module-to-heatsink contact thermal resistance of 0.2K/W and a heatsink-to-ambient thermal resistance
20W. Each power device is mounted on a common heatsink. The MOSFET has a junction-to-case of 0.1K/W. The maximum ambient temperature is 30°C and the highly inductive load current is constant
thermal resistance of 0.7K/W and a case-to-heatsink thermal resistance of 0.5K/W. The diode has a at 100A. If the diode on-state voltage is 1V, determine
junction-to-case thermal resistance of 0.8K/W and a case-to-heatsink thermal resistance of 0.6K/W.
i. the diode junction temperature
i. Determine the maximum heatsink thermal resistance that maintains both junction
ii. the current to double the rectifier lifetime (decrease junction temperature by 10°C)
temperatures below 90°C in a 30°C ambient.
iii. the heatsink to double the rectifier bridge lifetime (at 100A).
ii. Semiconductor lifetime approximately doubles for every 10°C decrease in junction
temperature. If the heatsink in the previous case is fan cooled, estimate the
lifetime improvement if the heatsink thermal impedance is halved with fan cooling.
Solution
iii. If the load current is constant (25A) and the switch and diode on-state voltages are
the same, determine the chopper on-time duty cycle and device instantaneous
i. During rectification, two diodes always conduct therefore total module conduction losses are
losses assuming no switching losses (only on-state losses).
PM = 2 × I o ×V Don = 2 × 100A × 1V = 200W
Tj1 Tj2 Tj3 Tj4 Tj5 Tj6 Tj
Solution
The figure shows how the six thermal paths can be PM/6 PM/6 PM/6 PM/6 PM/6 PM/6 PM 200W
i. Applying Kirchhoff’s voltage law to each loop of the equivalent thermal circuit shown gives: reduced to the simplified equivalent thermal model
T Dj − T hs = 20W × ( 0.8K/W + 0.6K/W ) = 28°C on the right.
Rθj-c Rθj-c Rθj-c Rθj-c Rθj-c Rθj-c = 0.24K/W Rθj-c/6 0.04K/W
TTj − T hs = 40W × ( 0.7K/W + 0.5K/W ) = 48°C
Applying Kirchhoff’s voltage law
Since both semiconductor devices are mounted on the same heatsink, Ths is the same in each case, the Tc Tc
MOSFET virtual junction will operate 20°C hotter than the diode junction. Therefore the MOSFET junction T j − T a = PM × ( 1 6 Rθ j −c + Rθ c −hs + Rθ hs −a )
0.2K/W Rθc-hs 0.2K/W
temperature should not exceed 90°C, that is
T j − 30°C = 200W × ( 1 6 × 0.24K/W + 0.2K/W + 0.1K/W ) Ths Ths
90°C − T hs = 40W × ( 0.7K/W + 0.5K/W ) = 48°C
⇒ T j = 98°C 0.1K/W Rθhs-a 0.1K/W
giving a heat sink surface temperature of 90°C - 48°C = 42°C and a diode junction temperature of 42°C 30°C Ta Ta 30°C
+ 28°C = 70°C. The heatsink thermal resistance requirement is
165 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 166

ii. If the current is reduced so as to decrease the diode junction temperature by 10°C then
T j − T a = PM × ( 1 6 Rθ j −c + Rθ c −hs + Rθ hs −a ) Heat Sink

88°C − 30°C=PM × ( 1
6 × 0.24K/W + 0.2K/W + 0.1K/W ) ⇒ PM = 170.6W
External
Assuming the diode on-state voltage drop is independent of current, that is remains 1V then
PM = 2 × I o ×V Don
170.6W = 2 × I o × 1V ⇒ I o = 85.3A External
interface
iii. When the junction temperature is reduced by 10°C to 88°C by decreasing the heatsink thermal TIM2 Internal
resistance, and the constant load current is maintained at 100A
T j − T a = PM × ( 1 6 Rθ j −c + Rθ c −hs + Rθ hs −a ) Lid
Chip
88°C − 30°C = 200W × ( 0.04K/W + 0.2K/W + Rθ hs −a ) ⇒ Rθ hs −a = 0.5K/W Internal interface TIM1

♣ Figure 5.20. Power semiconductor package with thermal conduction path to heat sink via TIMs.

5.8 High-performance cooling for power electronics For high-power applications, the interface thermal resistance becomes an important constraint. Direct
soldering (for example, reflow soldering) is often difficult, particularly when copper is used because of
In many instances, standard finned aluminium heat sinks, even with fan assistance, cannot achieve the the large co-efficient of thermal expansion CTE mismatch between Cu and Aℓ203 and in turn, Si.
required cooling performance due to physical limitations in heat transfer capabilities, namely the limited Diamond-filled greases have an effective thermal conductivity of over 20W/mK. Also possible is a
thermal conductivity of air for convection and copper and aluminium for conduction. nanostructured foil, which utilizes a fast exothermic reaction to create a soldered connection virtually at
Figure 5.19 shows a comparison of various cooling techniques as a function of the attainable heat room temperature.
transfer in terms of the heat transfer coefficient, h. For example, using equation (5.4), to accommodate a Heat spreading is an effective method of mitigating the need for complicated high-heat flux cooling
heat flux of 100W/cm2 at a temperature difference of 50K requires an effective heat transfer coefficient options. To be effective the benefits of decreasing the heat flux density by increasing the area should
(including a possible area enlarging factor) of 20kW/m2K (h = 100W/cm2/50K). From Figure 5.19 it can outweigh the penalty of adding another thermal layer through which the heat must conduct. Other than a
be concluded that liquid cooling can play an important role in thermal management. traditional copper heat spreading base plate, the alternative is to use advanced heat spreading materials
such as carbonaceous materials, metal-matrix composites, ceramic matrix composites (for example,
diamond-particle-reinforced silicon carbide), or ScD (Skeleton cemented Diamond), all with higher
thermal conductivities than copper, are much lighter, and have tuneable CTEs.
2
air He 20kW/m K By employing heat spreaders, cooling methods such as loop heat pipes and low-flow liquid cooling may
FC be augmented to accommodate higher heat flux applications. Figure 5.21a shows heat spreading results
for a 300W heat source of 2cm2 area as a function of thermal conductivity λ, thickness t, and cooling
air jet
boundary condition (that is, heat transfer coefficient h). Heat spreading is a complex phenomenon
water because the conduction and convection effects cannot be separated and the two effects compete:
increasing the thickness increases the through-plane thermal resistance but decreases the in-plane
thermal resistance. For example, comparing the two upper curves with the two lower curves, their order
water jet
is changed. The figure also shows that heat spreaders can be used to decrease the required fluid-side
legend water boiling heat transfer coefficient to manageable values, below 5kW/m2K, which can be realized with
natural convection He water condensation hydrofluoroether (HFE) cooling fluids. For example, using an 8×8=64cm2 heat spreader of an advanced
forced convection
composite with a thermal conductivity λ of 800W/mK and a thickness t of 4mm results in a temperature
jet water boiling
rise of 40°C with a heat transfer coefficient h of 2500W/m2K.
water condensation
For a single heat energy source, minimal thermal gain results from a Cu base plate thickness in excess
0 1 2 3
of 6mm. In the case of power IGBT modules, this boundary is complicated by the fact that many heat
10 10 10 10 104 105 106 sources, die, are adjacently bonded to a given copper coated ceramic substrate, which is then bonded
Heat transfer coefficient h W/m K 2 to a Cu spreader base plate.
The relative heat spreading resistance for varying spreader thicknesses in aluminium, copper, silver,
and Cusil (72%Ag+28%Cu) is shown in figure 5.21b. Aluminium is extensively used (also for air-cooled
Figure 5.19. Heat transfer coefficient h attainable with natural convection, single-phase liquid forced heat sinks) even though its relative thermal conductivity is the poorest of the materials shown; because
convection, and boiling for different coolants. (see figure 5.50) of its:
• lowest cost
• ease of fabrication
• performance adequacy
5.9 Conduction and heat spreading

In all power electronic cooling applications, heat from the device sources must be transmit via thermal 5.10 Heat-sinks
conduction to the surfaces exposed to the cooling fluid before it can be rejected to the coolant. As
shown in Figure 5.20, heat must be conducted from the die to the Aℓ203 substrate and Cu base plate to Heat sinks are devices that enhance heat dissipation from a hot surface, usually the case of a heat-
the heat sink before it can be rejected to the flowing air. A thermal interface material (TIM) may be used generating component, to a cooler ambient, usually air. Air is assumed to be the cooling fluid. In most
to facilitate thermal conduction from the die to the base plate and from the base plate to the heat sink. In situations, heat transfer across the interface between the solid surface and the coolant air is the least
power electronics, heat spreaders (heat sinks without any cooling fins and less bulky structures) in the efficient within the system, and the solid-air interface represents the greatest barrier to heat dissipation.
form of a flat plate with good thermal conductivity may be interposed between the die mounted substrate A heat sink lowers this barrier by increasing the surface area that is in direct contact with the coolant.
and heatsink, to facilitate spreading of the heat from its small source. Vapour chambers are also used to This allows more heat to be dissipated and/or lowers the device operating temperature. The primary
spread heat from a concentrated die or module heat source to a larger heat sink. purpose of a heat sink is to maintain the device junction temperature below the maximum allowable
temperature specified by the device manufacturer.
167 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 168

5.10.2 Heat-sink selection

°C
110
die heat source 300W, 1×2cm Area λ h In selecting an appropriate heat sink that meets the required thermal criteria, it is necessary to examine
100 various parameters that affect not only the heat sink performance itself, but also the overall performance
cm2 W/mK W/m2K
∆T 90
of the system. The choice of a particular heat sink type depends largely on the thermal budget allowed
4x4 800 2500 for the heat sink and external conditions surrounding the heat sink. There is not a single thermal
80
resistance value assigned to a given heat sink, since thermal resistance varies with external cooling
Temperature rise

conditions.
70 When selecting a heat sink, it is necessary to classify the airflow as natural, low flow mixed, or high flow
forced convection. Natural convection occurs when there is no externally induced flow and heat transfer
4x4 400 5000
60 relies solely on the free buoyant flow of air surrounding the heat sink. Forced convection occurs when
the flow of air is induced by mechanical means, usually a fan or blower. There is no clear distinction on
50 6x6 800 2500 the flow velocity that separates the mixed and forced flow regimes. Generally the effect of buoyant force
on the overall heat transfer diminishes to a negligible level (under 5%) when the induced airflow velocity
40
6x6 400 5000
excess 1 to 2m/s.
8x8 800 2500
Next, the required volume of a heat sink is determined. Table 15.12 shows approximate ranges of
30 8x8 400 5000
volumetric thermal resistance of a typical heat sink under different flow conditions.
0 4 8 12 16
Thickness t mm Table 15.12: Range of volumetric thermal resistance
(a) flow condition volumetric resistance
3
m/s cm °C/W
natural convection 500-800
Relative spreading resistance

0.25
die heat source, 1×1cm
1.0 150-250
(b)
0.20 2.5 80-150
5.0 50-80
0.15 Aluminium Al
The heat sink volume for a given low flow condition is obtained by dividing the volumetric thermal
0.10 resistance by the required thermal resistance. Table 5.12 is used only as a guide for estimation
Copper Cu purposes at the beginning of the selection process. The actual resistance values may vary outside the
Silver Ag shown range depending on additional parameters, such as actual dimensions of the heat sink, type of
0.05 Cusil 72% Ag 28% Cu the heat sink, flow configuration, orientation, surface finish, altitude, etc. The lower values shown
correspond to a heat sink volume from approximately 100 to 200cm3 up to about 1000cm3.

0 The tabulated ranges in Table 5.12 assume that the design has been optimized for a given flow
0 4 8 12 16 condition. Although there are many parameters to be considered in optimizing a heat sink, one of the
most critical parameters is the fin density. In a planar fin heat sink, optimum fin spacing is mainly related
Thickness t mm to two parameters: flow velocity and fin length H in the direction of the flow, as shown in Table 5.5.
Figure 5.21. Example of effect of thickness of heat spreader for various: It is beneficial to increase the width of a heat sink rather than its length. Also, the effect of radiation heat
(a) heat source areas, material thermal conductivities, and heat transfer coefficients and transfer is important in natural convection, as it can be responsible for up to 30% of the total heat
(b) material and resultant relative heat spreading resistance. dissipation. As the ambient temperature rises the heatsink temperature increases for a constant thermal
power loading, and the heatsink thermal resistance decreases due to the increased significance of
thermal radiation (T4 dependence in equation (5.1)) in the heat removal process. As the ambient
5.10.1 Required heat-sink thermal resistance temperature decreases the heatsink thermal resistance decreases slightly.
Unless the component is facing a hotter surface nearby, it is imperative to have the heat sink surfaces
To begin the heat sink selection, the heat sink thermal resistance required to satisfy the thermal criteria suitably ‘black’ painted or anodized to enhance radiation.
of the component is determined. By rearranging equation (5.27) into terms of the ambient temperature
Ta, the heat sink resistance is obtained as 5.10.3 Heat sink types
T -T
Rθ s −a = j a - Rθ j −c - Rθc −s (5.60) Heat sinks separate into three broad categories:
Pd • Plate-fin – suitable for general straight airflow;
where Tj, Pd and Rθj-c are provided by the device manufacturer, and • Pin-fin – suitable for omi-directional airflow; and
Ta and Rθc-s are user-defined parameters. • Foam-fin – suitable for ducted airflow with a high pressure drop.
The ambient air temperature Ta for cooling electronic equipment depends on the operating environment Heat sinks can be classified in terms of manufacturing methods and their final form shapes.
in which the component is used. Typically, it ranges from 35 to 45°C, if external air is used, and from 50 Conventional heat sink manufacturing methods involve extruding and die-casting.
to 60°C, if the component is enclosed or in the wake of another heat-generating component.
The most common types of air-cooled heat sinks are summarised in Table 5.13 and include:
Extruded fins:
Extrusion is a process in which a solid block is converted into a continuous length of uniform cross-
section by forcing it to flow under high pressure through a die orifice, which is so shaped, as to impart
the required form to the product. Typically, billets of aluminium are placed within a strong walled
enclosure and are caused to extrude through the die under a high pressure exerted by a ram, actuated
hydraulically or mechanically. Extrusion is the most widely used method for heat sink manufacture.
169 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 170

This process allows the formation of elaborate two-dimensional shapes capable of dissipating large heat Alternatively, corrugated sheet metal of either aluminium or copper increases surface area, hence, the
loads. They may be cut, machined, and coated. Cross-cutting will produce omni-directional, rectangular volumetric performance. The heat sink is then attached either to a base plate or directly to the heating
pin fin heat sinks, and incorporating serrated fins improves the performance by approximately 10 to surface via epoxying or brazing. It is not suitable for high profile heat sinks because of the availability
20%, but with a slower extrusion rate. Extrusion limits, such as the fin height-to-gap fin thickness, L/s, and fin efficiency.
usually dictate the flexibility in design options (see figure 5.4). Typical fin height-to-gap aspect ratio L/s
of up to 6 and a minimum fin thickness t of 1.3mm, are attainable with a standard extrusion. A 10 to 1 Skived fins:
aspect ratio W/L and a fin thickness t of 20mm can be achieved with special die design features. In the skiving process, fins are machined using special tooling, whereby precisely sliced layers from an
However, as the aspect ratio increases, the extrusion tolerance is compromised. extruded metal block are bent at the base of the slice to form slender curved fins. Since the fins and
base are an integral unit, the interface resistance found in folded and bonded heat sinks is absent.
Casted fins: Aluminium 6063 is the preferred material because of its superior machinability and strength, but copper
In the die-casting method, molten metal is forced under pressure into metal dies or moulds to produce arrays are also available. The depth of cut determines the fin thickness and can result in extremely thin
accurately dimensioned parts. It is the fastest of all casting processes and is often employed where fin structures, yielding light and competitive heat sink designs.
rapidity and economy in production are essential. The thermal conductivity of cast heat sinks may be
worsened by porosity caused by gases evolving during solidification. Machined fins:
Heat sinks are machined out of a metal block by material removal to create the inter-fin spaces. Most
Sand, lost core, and die casting processes are available with or without vacuum assistance, in commonly, they are manufactured by gang saw cutting on a computer numerical control machine. The
aluminium or copper/bronze. This technology is used in high-density pin fin heat sinks which provide gang saw consists of multiple saw cutters on an arbour with precise spacing, which depends on the heat
maximum performance when using impingement cooling. sink geometry to be machined. Fins damaged and distorted during processing require extensive
secondary operations. Material is also consumed in an unproductive manner by the generation of scrap
Modified die-casted fins: swarf metal.
The modified die-casting process involves the extension of basic die-casting principles, whereby the
base of the heat sink is die-cast around a fixtured array of extremely thin stamped fins. The fins are Swaged fins:
separated by spacers, which prevent the die-cast material from flowing into the fin-to-fin spacing. Individual fins are placed in a pre-grooved base, and then rollers swage the sides of the fins to maintain
Aluminium is the most commonly used material for this technique. The absence of a so-called interface them in place.
between the fins and the base eliminates the impact of an interface resistance. This process allows
much higher aspect ratios while fulfilling the requirement of small inter-fin spacing. These heat sinks are
usually combined with a heat pipe to provide an effective thermal solution. Table 5.13: Feature of different types of heatsinks
Bonded/fabricated fins:
Thermal
Bonded heat sinks are often built-up extrusions, typically manufactured by assembling extruded plates Heatsink fin type Applications Advantages Disadvantages
resistance
into slots or grooves on an extruded or machined heat sink base plate, and held in place by an interface,
extruded most applications varies versatile limited size
usually a two part thermosetting thermally conductive aluminium-filled epoxy or a solder. However, the
bonding agent presents a thermal barrier. These heat sinks are often costlier to manufacture, and the low thermal conductivity,
die-casted low power high expensive
base typically requires special machining. Process limitations are usually related to the strength of the expensive die charge
bonding agent and dimensional constraints for the slot in the heat sink base. Hybrid heat sinks utilizing bonded large applications high close tolerances expensive
different materials for the fins and the base are possible. Bonded fin arrays are most commonly light weight and
rectangular plate fin arrays. single-fin fabricated
all application very low low profile with high expensive
assembly
degree of flow
Most air-cooled heat sinks are convection limited, and the overall thermal performance of an air-cooled limited in design and
heat sink can be improved if more surface area can be exposed to the air stream. The bonding process forged most applications moderate inexpensive
flow management
allows for a much greater fin height-to-gap aspect ratio L/s of 20 to 40, greatly increasing the cooling
stamped low power high inexpensive low performance
capacity without increasing volume requirements.
convoluted high at low flows high heat flux expensive,
ducted air
(folded) fin low at high flows density needs ducting
Forged/stamped fins:
thick base, high weight,
In forged heat sinks, the fin arrays are formed by forcing raw material into a moulding die using a punch. skived most applications moderate close tolerances
orientation sensitive
A common problem in forging is the choking of material in the moulding die cavity, which leads to fins of high aspect ratio fins
uneven height. Aluminium and magnesium alloys are readily forged, and an important economic machined prototypes design dependant quickly produced difficult to machine –
advantage is a typically low rejection rate for the process. inconsistent fin geometry
heavy and bulky,
Some of the attractive benefits of forging include high strength, superior surface finish, structural rigidity, suitable for power
limited availability for
swaged high power medium
close tolerance capabilities, continuity of shape, and high uniformity of material. devices
flow management
Copper or aluminium sheet metals are stamped into desired shapes. They are used in traditional air-
cooling of electronic components and offer a low cost solution to low density thermal problems. They are
suitable for high volume production, because advanced tooling with high-speed stamping lowers costs. 5.10.4 Heatsink fin geometry
Additional labour-saving options, such as taps, clips, and interface materials, can be factory applied to
help to reduce the assembly costs. Figure 5.22, showing the nomenclature of the array geometry, including the fin height, L, fin thickness, t,
inter-fin spacing, s, width of base, W, and the length of the heat sink base, H.
Folded/convoluted fins:
Folded heat sinks are built-up sheet metal, manufactured by folding sheet metal into a serpentine fin Figure 5.23 shows the results of thermal resistance calculations for an aluminium heat sink (0.1x0.1m
array. The folded metal sheets are attached to the base of the heat sink by soldering or brazing, which heat sink base; fin height, L=0.05m), in forced air convection for flow conditions of 20Pa and 0.01m3/s.
results in additional thermal resistance at that interface. This contact resistance is small, due to the fact The thermal resistance is defined as the ratio of the excess temperature difference in °C to the heat
that the ‘bends’ of folded fins are typically flattened while bonding or brazing, thus increasing the contact dissipation rate in W, where the excess temperature difference is between the bottom surface of the
surface area. Difficulty in achieving smaller fin pitches required to construct dense arrays is a common heat sink base and the incoming air at the heat sink inlet. It is assumed that such a flow condition is
issue. Similar to the bonding process, this manufacturing method allows flexibility in designing hybrid defined by a pressure drop and flow rate representing a single point on a fan characteristic curve.
heat sinks made up of a combination of different materials.
171 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 172

L Table 5.14: Process capabilities for different manufacturing techniques

intake air
Conventional Processes Modern Processes

W
Parameter Modified
exit air Extruded Die-Casting Bonding Folding Forging Skiving Machining
(see figure 5.4) Die-Casting

min t, mm 1 1.75 0.75 0.25 0.2 0.4 0.3 0.5

H
max L/s 8:1 6:1 60:1 40:1 20:1 50:1 25:1 50:1

min s, mm 6.6 8.3 0.8 1.25 0.2 1 2 1

Material Aℓ Aℓ, Zn-Alloy Aℓ, Cu, Mg Aℓ, Cu Aℓ, Zn-Alloy Aℓ Aℓ Aℓ, Cu, Mg
s
t

For the conditions illustrated in Figure 5.23, a minimum value of 0.135°C/W occurs at 4 fins/cm, yielding
Figure 5.22. Geometry of the plate fin heat sink analyzed. a design that will dissipate 186W at a 25°C excess temperature from the base to the inlet air. The fin
thickness and spacing for this aluminium structure are 0.57mm and 1.93mm, respectively with a heat
sink mass of 0.308kg. In Table 5.14, not all the processes are capable of creating a heat sink with such
dimensions with a gap aspect ratio of 26 and a fin aspect ratio of 86.
For each value of fin density, there is a corresponding fin-to-fin spacing, s, and fin thickness, t, which
meets the pressure drop requirement at the specified flow rate. Increasing fin number decreases both Figure 5.24 shows the typical range of cost functions for different types of heat sinks in terms of required
the fin thickness and the spacing. For a given thermal operating condition, described by pressure drop thermal resistance Rθ.
and volumetric flow rate, designs with particularly small and relatively large fin densities yield high
values of thermal resistance, the former due to limited surface area and the latter caused by the highly
inefficient thin fins. An intermediate geometry, which minimizes the thermal resistance, can thus be 102
identified, where the thermal optimization illustrated in Figure 5.23 uses a heat sink volume as the
heat pipes

US $
driving thermal constraint and not an actual application specification.
liquid
systems bonded fins
°C / W

1 101
3
Aluminium, 20Pa, 0.01m /s

Cost per heatsink @ 5000 quantity

10cm × 10cm × 5cm folded fins
Heat sink thermal resistance , Rθhs-a

0.8 die-casting fabricated

extrusions

100
0.6

skiving
stampings
0.4 bonded fin, folded fin,
Modified die-cast
machine 10
-1

forging
extrusion
0.2 10
-2
10
-1
10
0 -2
10 10
2

swaging Thermal resistance Rθ C/W (junction to heatsink)

Figure 5.24. Cost versus required thermal resistance.

0 1 2 3 4 5 6 7
Fin Density, fins/cm
The performance of different heat sink types varies dramatically with the airflow across the surface area
4.6 2.8 2.2 1.9 1.75 1.6 of the heat sink. To quantify the effectiveness of different types of heat sinks, the volumetric heat
Fin Spacing, s, mm transfer efficiency is defined as
PD
ηv = (5.61)
5.4 2.2 1.1 0.6 0.25 0.04 mf c p ∆T sa
Fin Thickness, t, mm where mf is the mass flow rate through the heat sink, cp is the heat capacity of the fluid, and ∆Tsa is the
average temperature difference between the heat sink and the ambient air. The heat transfer efficiencies
for a wide range of heat sink configurations, and their ranges are listed in Table 5.15. The improved
Figure 5.23. Typical thermal characteristic plot for a heat sink:
thermal performance is associated with additional costs in either material or manufacturing, or both.
Thermal Resistance versus Fin Density, 0.1mx0.1m, 0.05m, Aluminium, 20Pa, 0.01m3/s.
173 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 174

Table 5.15: Range of heat transfer efficiencies 5.11 Heatsink cooling enhancements

Heat sink type

ηv range Other than natural convection cooling of heatsinks, other possible active thermal management
% technologies for power electronics applications include:
Stamping & flat plates 10-18 Heatsink air-cooling with fans and blowers
Enhanced air-cooling
Finned extrusions 15-22 Piezo fans
Impingement flow 'Synthetic' jet cooling
25-32
Fan heat sinks 'Nanolightning'
Fully ducted extrusions 45-58 Indirect liquid cooling
Ducted pin fin, Heat pipes
78-90 Cold plates
Bonded and folded fins
Direct liquid cooling
Immersion cooling
5.10.5 Thermal performance graph Liquid jet impingement
Spray cooling
Typical heat sink performance graphs are shown in figure 5.25. It is assumed that the device to be Microchannels and minichannels
cooled is correctly mounted, and the heat sink is in its normal mounting orientation with respect to the Electrohydrodynamic and electrowetting cooling
direction of airflow. The solid plot is the natural convection curve of heat sink temperature rise, ∆Tsa, Liquid metal cooling
versus PD, which assumes that the heat sink is appropriately painted or anodized black. The dashed Solid-State cooling
curve is the forced convection curve of thermal resistance versus air velocity. In forced convection, ∆Tsa Thermoelectric (Peltier devices)
is linearly proportional to PD, hence Rθsa is independent of PD and becomes a function only of the flow Superlattice and heterostructure cooling
velocity. However, the natural convection phenomenon is non-linear, making it necessary to present Thermionic and thermotunnelling cooling
∆Tsa as a function of PD. Phase change materials and heat accumulators
The performance graphs can be used to identify the heat sink and, for forced convection applications, to
determine the minimum flow velocity that satisfy the thermal requirements. For example, if the required
thermal resistance in a force convection application is 8°C/W, the thermal resistance versus flow velocity 5.12 Heatsink fan and blower air-cooling
curve indicates a velocity of at least 2.4m/s (480lfm). For natural convection applications, the required
thermal resistance Rsa can be multiplied by PD to yield the maximum allowable ∆Tsa. The temperature Fans are low-pressure air pumps that utilize power from an electric motor to output a volumetric flow of
rise of a chosen heat sink must be equal to or less than the maximum allowable ∆Tsa at the same PD. air at a given pressure. A propeller converts torque from the motor to increase static pressure across the
fan rotor and to increase the kinetic energy of the air molecules. Electronic cooling fans draw or flow air
from outside an electronic enclosure into the electronics area, expel heated air from inside the
Heat dissipated PD W
enclosure, or move air across a heat sink or electronic device to accelerate the removal of heat energy
0 1 2 3 4 5 from the device.
100 20

°C/W
°C

Fan motors are typically permanent split capacitor ac induction or brushless dc.
80 16 Rθ
∆T

Air moving devices are generally either a type of axial (and/or propeller) fan, figure 5.26a, or a
60 12 centrifugal blower, figure 5.26b. The main difference between fans and blowers is their flow and
Thermal resistance

pressure characteristics. The specific ratio - the ratio of the discharge pressure over the suction
Temperature rise

40 8 pressure - is used for defining fans (<1.11), blowers (1.11 to 1.20), and compressors (>1.20). Axial fans
take and deliver air in an overall direction that is parallel to the fan blade axis (no change in direction)
20 4 and can be designed to deliver a high flow rate, but work against low pressure. Radial (centrifugal)
blowers tend to deliver air in a direction that is perpendicular to the blower axis at a relatively low flow
0 0 rate, but against high pressure. The air changes direction twice (on entering and on leaving).
0 1 2 3 4 5
The most common axial fans are propeller, tube-axial and vane-axial styles.
Air velocity m/s • Propeller fans are the simplest type of fan, consisting of a motor and propeller. One
problem with propeller fans is that tip vortices are produced by the pressure differential
Figure 5.25. Typical finned heatsink thermal performance graphs.
across the airfoil section. The required pressure is low.
• A tube-axial fan (the most common type in electronic cooling systems) is similar to a
The natural convection curves assume an optimal orientation of the heat sink with respect to gravity. propeller fan, but also has a Venturi around the propeller to reduce the vortices. It
Also, the flow velocity in the forced convection graph represents the approach flow velocity without develops high pressure but has a low efficiency, with peak efficiency generally occurring
accounting for the effect of flow bypass. Flow bypass reduces the performance of a heat sink by as just before the stall dip.
much as 50% for the same upstream flow velocity. • The vane-axial fan has vanes that trail behind the propeller in the airflow to straighten
When a device is substantially smaller than the base plate of a heat sink, the thermal spreading the swirling flow created as the air is accelerated.
resistance needs to be considered in the selection process. Performance graphs generally assume that
Impeller types, known as flat packs, have a small aspect ratio, with good flow rate and pressure drop.
the heat is evenly distributed over the entire base area of the heat sink, and therefore, do not account for
the additional temperature rise caused by a concentrated heat source. This spreading resistance could Centrifugal blowers may have a forward curved wheel, a backward curved wheel, airfoil-tubular, or be of
typically be 5 to 30% of the total heat sink resistance. the squirrel cage (radial) variety. They tend to be quiet: noise decreases with increased number of
Metal heat sinks may act as electromagnetic radiators, even when earthed. When EMC is an issue, blades and have excellent pressure drop characteristics.
thermal conducting plastic resins (thermoplastics) may be a viable alternative to a metal heat sink.
Thermal conductivities from 5 up to 50W/mK (similar to stainless steel 15W/mK and ceramic aluminium Mixed flow fans combine the characteristics of both the axial fan the radial blower. The air flows in both
oxide 25W/mK) are a result of additives. Such composite plastic heat sinks (moulded base, folded fin) axial and radial directions relative to the shaft. Mixed flow fans develop higher pressures than axial fans.
are 50% lighter than aluminium equivalents and advantageously are available in electrically insulative In a cross flow fan the airflows in an inward direction and then in an outward radial direction.
and electrical conductive grades.
 Table 5.16 Design, characteristics, and applications of axial fans and blowers

Performance
type Impeller design Housing design applications
characteristics
• Low efficiency
• Simple circular ring, • High flow rate but low efficiency •Low pressure, high volume air moving
• Limited to low pressure applications
orifice plate or Venturi and pressure capabilities applications such as air circulation in a space
• Low cost impellers have at least two
• Optimum design is • Maximum efficiency near free or ventilation through a wall without ductwork
blades of constant thickness
small blade tip gap delivery (zero static pressure) •Some exhaust applications
connected to a small hub
and forms smooth • Discharge pattern circular and •Used for makeup air applications
• Primary energy transfer is by airfoil into wheel airstream swirls

PROPELLER
velocity pressure

• High flow rate, medium pressure

capabilities •Low and medium pressure ducted HVAC*
• More efficient and develops higher
applications where air disturbance
static pressure than propeller fans • Higher efficiency than propeller
• Cylindrical tube with downstream is not critical
• Usually 4 to 8 blades with airfoil or types
close clearance to •Some industrial applications such as drying
single thickness cross section • Performance curve dips before
blade tips ovens, paint spray booths, and fume exhausts
• Hub less than transfer by velocity peak pressure
pressure

TUBE-AXIAL
• Discharge pattern circular and air *HVAC Heating Ventilation and Air Conditioning
stream rotates or swirls

AXIAL FANS
• High pressure characteristics with
• Good blade design give medium to •General HVAC* systems with low, medium, to
• Cylindrical tube with medium volume flow capabilities
high pressure capability at good high pressure applications where straight
close clearance to • Performance curve dips before
efficiency through flow and compact installation are
blade tips peak pressure due to
• Most efficient versions have airfoil required
• Guide vanes up or aerodynamic stall. Avoid
blades •Good down-stream air distribution
down stream from operation in this region
• Blade may be fixed, adjustable or •Replaces tube-axial fans in industrial
controllable pitch impeller increase • Guide vanes correct circular applications
pressure capability and motion imparted by wheel and

VANE-AXIAL
• Hub is usually less than half fan tip •More compact than centrifugal fans for same
efficiency improve fan pressure
diameter duty
characteristics and efficiency

BWW

Chapter 5 Cooling of Power Switching Semiconductor Devices 176

type Impeller design Housing design Performance characteristics

PROPELLER
EFFICIENCY

PRESSURE POWER

VOLUME FLOW RATE

Axial fans
TUBE-AXIAL
EFFICIENCY

PRESSURE POWER

VOLUME FLOW RATE

VANE-AXIAL
PRESSURE POWER
EFFICIENCY

VOLUME FLOW RATE

177 Power Electronics

Performance
type Impeller design Housing design applications
characteristics
• Highest efficiency of all centrifugal fan designs
• Scroll-type design for
• 10 to 16 blades of airfoil contour curve
efficient conversion of • Highest efficiencies at 50% to 60%
backwards from rotation direction. Deep blade • General HVAC large
velocity pressure to static of wide open volume. This volume
allow for efficient expansion within blade systems
pressure has good pressure characteristics
passage • Large clean-air operations
• Maximum efficiency requires • Power reaches maximum near peak
• Air leaves passage at velocity less than tip for significant energy
close clearance and efficiency and decreases or self-

AIRFOIL
speed savings
alignment between wheel limits toward free delivery
• For a given duty, has highest speed of
and inlet
centrifugal designs

• HVAC, as for airfoil types

• Efficiency slightly less than airfoil fan
• Uses same housing • Industrial applications
• 10 to 16 single thickness blades curved or • Similar to airfoil fan, except peak
configuration as airfoil where airfoil blade may
inclined away from rotation direction efficiency slightly lower
design corrode or erode due to
• Efficient for same reasons as airfoil
environment

BACKWARD CURVED
BACKWARD INCLINED
• Scroll. Usually narrowest of • Material handling in
• Higher pressure characteristics than airfoil and • Higher pressure characteristics than
all centrifugal designs industrial plants
backward blade types airfoil and backward blade fans
• Because wheel design is less • Rugged wheel is easy to
• Curve has a break before pea pressure and • Pressure drop before peak pressure
efficient, housing dimensions repair. Wheel may be

CENTRIFUGAL FANS
fan should not be operated in this area but this usually causes no

RADIAL
are not as critical as for airfoil specially coated
• Power rises continually to free air delivery problems
and backward blade types • Usual not for HVAC

• Pressure curve less steep than

backward curved fans. Curve dips
• Scroll similar to possibly
• Flatter pressure curve and lower efficiency before reaching peak pressure
identical to other centrifugal
than airfoil and backward blade types • Highest efficiency after peak • Primarily for low-pressure
designs
• Do not rate fan in pressure dip before peak pressure at 40 to 50% of wide HVAC, such as
• Fit between wheel and inlet
pressure open volume residential furnaces,
not as critical as for airfoil
• Power rises continually to free air delivery. • Rate fan to volume greater peak central station units and
and backward blade types
Motor selection must account for this pressure packaged air-conditioners
airfoil and backward blade
• Account for power curve which rises
types
continually towards free delivery,

FORWARD CURVED
when selecting motor

Chapter 5 Cooling of Power Switching Semiconductor Devices 178

type Impeller design Housing design Performance characteristics

AIRFOIL
EFFICIENCY

VOLUME FLOW RATE

PRESSURE POWER

INCLINED
BACKWARD
EFFICIENCY

VOLUME FLOW RATE

PRESSURE POWER

BACKWARD CURVED

CENTRIFUGAL FANS
RADIAL

VOLUME FLOW RATE

EFFICIENCY

PRESSURE POWER

CURVED
FORWARD
EFFICIENCY

VOLUME FLOW RATE

PRESSURE POWER
 Pressure increase and flow rate in conjunction with the rotational velocity ω at the perimeter, the outer
diameter Do of the impeller, and the density ρℓ of the fluid medium result in the dimensionless
parameters pressure figure Φf and volume figure, θf.
∆p
φf = (5.62)
½ρ Aω 2
V
θf = (5.63)
¼π Do2ω

Axial flow Radial flow Mixed flow These parameters allow comparison of different designs, dimensions and speed, with one another.
Figure 5.27 illustrates this comparison for typical characteristic curves of the various designs, making
the advantages apparent, in particularly:
• Radial fans give a large increase in pressure and low flow rates
• Mixed-flow fans give medium pressure and medium flow rates
• Axial fans give high volume flow rates and low increase in pressure

The fan curve

The aerodynamic aspects of a fan are exhibited in a fan curve such as is shown in figure 5.28. Healthy
aerodynamic flow is on the x-axis and rotates anticlockwise through to aerodynamic stall (the vertical
axis). A stalled fan continues to deliver air, but at an increased static pressure and a decreased
volumetric flow rate, and also at the cost of an increase in noise. If noise is not a consideration, the fan
can be utilized in this condition.
Axial flow Radial flow Cross flow
with Meridional acceleration with inducer section The fan performance curve can be understood in terms of energy. At the shut-off or no-flow point, A, the
fan is in a condition of maximum potential energy, hence produces a maximum possible pressure. At
free delivery, point D, the fan is in the maximum kinetic energy condition. Although neither of these
extreme conditions are likely to occur in practice, they are useful parameters when comparing fans. The
fan stall region, B, is unstable, and should be avoided. The region C to D is the fan stable low-pressure
area, where a preferred operating point near D improves efficiency and compensates for filter clogging.

static,
no delivery
pu speed
th
centrifugal 4 sector
1 A Blowers, which provide
medium flow at high pressure
system
centrifugal impedance
forward blades curve nd
3 sector

Φ
0.8
Mixed Flow fans, which provide
B high flow at increased pressure
stall

Static pressure
0.6 region
4th nd
2 sector
C sector Mixed Flow fans, which

Pressure figure
axial provide high flow at
mixed flow recommended 3rd increased pressure
0.4 sector
selection range
2nd st
1 sector
maximum sector
0.2 Tube-axial fans, which
flow rate
1st provide very high flow
centrifugal at low pressure
radial sector
D
0
0 0.1 0.2 0.3 0.4
Airflow G cfm
Volume figure θ equation 5.63

Figure 5.27. Fan curves: (a) comparison of normalized curves for various fan designs and
(b) typical fan characteristic regions.

The governing principle in fan selection is that any given fan can only deliver one flow at one pressure in
a particular system. This ‘operating point’ is determined by the intersection of the fan static pressure
curve and the system pressure curve. Figure 5.28 illustrates the operating points of both high and low
resistance systems. It is better to select a fan that gives an operating point toward the high flow, low-
(a) (b) pressure end of the performance curve to maintain propeller efficiency and to avoid propeller stall. Each
Forward Curved Backward Curved Radial
particular power electronic system should be analyzed for possible reduction in the overall resistance to
airflow. Considerations, such as available space and power, noise, reliability, and operating environment
Figure 5.26. Typical (a) impeller axial fan and (b) radial blower with backward curved blades. are other deciding factors.

BWW BWW
 181 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 182
Shut off

Normal operating range For axial designs, a sharp increase in noise is particularly noticeable when the flow rate is excessively
restricted. The axial fan enters an operating range in which the airflow no longer follows the contour of
Pa
η η
operating the impeller hub, resulting in additional noise.
point
For a specific operating point, an ideal blade geometry achieves the highest aerodynamic efficiency,
m
syste
W W

Fa
which coincides with minimal noise generation. For this reason, the lowest noise generation can only be
Static pressure

n
cu
achieved for the given point of operation.
tance

rv
operating Pa Pa

e
point Sources of fan noise include:
resis

• Vortex shedding - This is a broad-band noise source generated by air separation from the
high

e sy
s te m
Backward curve blade surface and trailing edge. It can be controlled by the blade profile design, proper pitch
ta n c operating axial
Lo w
re s is point Maximum centrifugal angle, and notched or serrated trailing blade edges.
free • Turbulence - Turbulence is created in the airflow stream itself. It contributes to broad-band
delivery
noise. Inlet and outlet disturbances, sharp edges and bends will cause increased turbulence
3 3
Airflow G m3/s m /s m /s and noise.
Figure 5.28. Fan curve: (a) fan (backward curved centrifugal) system interaction; • Speed - Speed is a major contributor to fan noise and its effect can be seen through the fan
input power and efficiency (b) backward curved centrifugal and (c) axial fans. laws in Table 5.17.
• Fan load - Noise varies as the system load varies. This variation is unpredictable and fan
dependent. However, fans are generally quieter when operated near their peak efficiency.
Acoustic noise • Structure vibration - This can be caused by the components and mechanism within the fan,
Sound is propagated in air by pressure waves. The effective value of pressure change is expressed such as residual unbalance, bearings, rotor to stator eccentricity and motor mounting. Motor
relatively as sound pressure level, SPL, in decibels, dB. The so-called A weighting curve is normally mounting noise is difficult to define. Cooling fans are motors and should be treated as such
used and the sound pressure level obtained is expressed in dB(A). when mounted.
P
SPL = 20 log (5.64) The following points will aid in the minimization of fan noise.
PRef
• System impedance - This should be reduced so that the least noise for the greatest airflow
where P is pressure and is obtained. The inlet and outlet ports of a cabinet can make up to between 60 and 80% of
Pref is the reference pressure the total system impedance, which is too high for a low-noise result. Also, if a large part of
the fan's flow potential is used up by the impedance of the inlet and outlet, a larger, faster
Since sound pressure level varies with distance to and direction of a device, it is not suitable as a fan and noisier fan will be required to provide the necessary cooling.
comparison basis. By contrast, the sound power level, PWL, determined from sound pressure
• Flow disturbance - Obstructions to the airflow must be avoided, especially in the critical
measurements, comprises all sound emissions, and is unaffected by distance to the fan noise source.
inlet and outlet areas. When turbulent air enters the fan, noise is generated, usually in
W discrete tone form, adding up to 10 dB. Obstructions placed near the fan intake raise the
PWL = 10 log (5.65)
WRef noise level more than obstruction on the exhaust side of the fan.
where W is the acoustic power of the source and • Fan speed and size - Most fans have several low speed versions. These should be
Wref is the acoustic reference power assessed and used if possible. Various fan sizes should also be explored; quite often a
larger, slower fan will be quieter than a smaller, faster fan delivering the same airflow.
• Temperature rise - Airflow is inversely proportional to allowable temperature rise within the
system. Therefore, the ∆T limit placed on the equipment will dictate the required flow, and
52 52 10
Maximum static
0 therefore, noise. If the temperature limit can be relaxed slightly, a noise reduction results.
Radial

Pa
pressure
• Vibration isolation - Fan isolation from the cabinet will avoid vibration transmission.
Pa

dBA

Because fans operate at a low frequency, and are light in weight, vibration isolators must
50 50 be soft and flexible. Since noise transmission is system dependent, experimentation is the
80

∆p
best approach to identify quiet system/fan interaction. In systems that require 20 CFM or
Mixed flow
∆p

less, noise radiated by the cabinet is the predominant noise, and isolation of the fan is the
Lp

Bigger radial component

48 48 Radial fan only practical solution. Mount the fan on the enclosure interior surface rather than on an
60 exterior surface. Use structural reinforcements to control enclosure resonant frequencies.
Diagonal
Pressure increase
Pressure increase

Bigger radial
Sound pressure

46 46
component 40
5.12.1 Fan selection
Axial Axial fan Estimate the required airflow
44 44 Before selecting a fan, it is necessary to estimate accurately the heat to be dissipated, because the
20 overall system air temperature differential above the inlet ambient is directly proportional to the heat
Maximum dissipated. Then the amount of required cooling air can be estimated.
fluid volume
42 42
The basic heat transfer equation is:
0 10 20 30 40 50 0 10 20 30 40 50 60
PD = mf × c p × ∆T (5.66)
Flow rate G x10-3 m3/s Flow rate G x10-3 m3/s where PD = amount of heat dissipated in the enclosure and transferred to the cooling system, W
cp = specific heat of air, 1.021 kJ/kg.K
Figure 5.29. Sound pressure level characteristics at 1m; comparison between axial and radial fans. mf = mass flow rate of air through enclosure, kg/s
∆T = desired air temperature differential (enclosure inlet to discharge ambient outside air), K
The sound radiation of a fan changes with its operating state, so sound power level is only conditionally
indicative for applications in which the fan does not operate under optimum conditions. A ‘characteristic The relationship between mass flow rate and volumetric flow rate is
acoustical curve’ of the fan results when the sound pressure level is measured in relation to a pressure mf = ρ A × G (5.67)
increase or flow rate. Figure 5.29 depicts this characteristic for axial and radial fans, wherein the sound where G = volumetric flow rate, m3/s [1 cfm = 1.7m³/h]
pressure level measured at a 1m distance from the fan intake side is depicted as a function of flow rate. ρℓ = air density, kg/m3, which is altitude dependant, see Table 5.19
183 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 184

From equations (5.66) and (5.67), the required volumetric flow rate is calculated as If this should fail to provide a solution, a different fan or perhaps even multiple fans should be
PD considered. The consideration of multiple fans is more complex. An additional fan doubles the cost,
G = (5.68) doubles the noise, doubles the heat generated by fans, and may provide only a minimal improvement to
ρ A × c p × ∆T
the cooling, but redundant fans may increase system reliability.
This equation yields an estimate of the airflow needed to dissipate a given amount of heat at sea level.
Note that it is the mass flow rate of air, not its volumetric flow rate that governs the amount of cooling. Impact of varying system impedance
To demonstrate the impact of system resistance on fan performance, figure 5.30 shows three typical
Estimate the actual system airflow fans, where A is a 120 cfm fan, B is a 100 cfm fan and C is a 70cfm fan. Line D represents a system
The actual operating airflow is determined by the intersection of the fan curve and the system resistance impedance within a given designed system. If 50 cfm of air are needed, fan A will meet the need.
curve, as shown in figure 5.27b. There are three options for estimating the operating point: However, fan A is a high performance, high noise fan that draws more power and is more costly. If the
• experimental measurement using a thermal/mechanical mock-up of the system, system impedance could be improved to curve E, then fan B would meet the 50 cfm requirement, with
• calculation of the operating point using airflow network methods, or reduced cost, noise and power draw. If the system impedance could be optimized to where curve F
• calculation of the system airflow using computational fluid dynamics software. were representative, then fan C meets the airflow requirement, at a dramatically lower power, noise and
cost level. This is a well-designed system from a forced convection cooling viewpoint, noting that a given
The experimental procedure can be used to measure the total airflow for specific fans or several fan can only deliver a single airflow into a given system impedance.
pressure-airflow data pairs can be measured to develop a complete system resistance curve. The latter
experimental method requires superimposition of the selected fan pressure versus airflow curve and Multiple fans - series and parallel operation
system resistance curve to obtain the operating airflow. Combining fans in series or parallel can achieve the desired airflow without greatly increasing the
The airflow network procedure provides adequate results when the geometry is simple and the flow path system enclosure size or fan diameter.
within the enclosure is known or an estimate can be made. • Parallel operation is two or more fans blowing together side by side. The performance of two
fans in parallel will increase the volume flow rate (by ∆G and double at maximum delivery).
Determining system impedance The best results for parallel fans are achieved in systems with low resistance. A fan curve
To estimate the system airflow, all enclosures are characterized by a system resistance curve of the simulating multiple, identical fans in parallel may be constructed by scaling the fan curve
type shown in Figure 5.30. System resistance curves are expressed as a non-linear expression of airflow axis data in direct proportion to the number of fans. As figure 5.31 shows, when a
pressure versus airflow: system curve is overlaid on the parallel performance curves, the higher the system resistance,
n the lower the flow gain with parallel fan operation. Thus, this type of application should only be
∆P = k exp × ρ A × G q (5.69)
used when the fans can operate in a low impedance near free delivery.
where ∆P = system static pressure loss [1Pascal = 1N/m² and velocity pressure Pv = ½ρℓv2] • In series operation, the fans are stacked one upon the other, resulting in an increase of static
kexp = a load factor specific to the system (determined experimentally) pressure, ∆P, doubling at shut-off, but less elsewhere, as seen in figure 5.31. The best results
ρℓ = density of fluid, air for series fans are achieved in systems with high resistance. A fan curve simulating multiple,
G = airflow rate identical fans in series, may be constructed by scaling the fan curve pressure axis data in
nq = airflow quality constant, which varies between 1 and 2 depending on whether the flow is direct proportion to the number of fans.
completely laminar (nq = 1) or completely turbulent (nq = 2)
In both series and parallel fan operation, certain areas of the combined performance curve will be
unstable and should be avoided. This instability is unpredictable and is a function of the fan and motor
construction and the operating point.
0.4 A
120 cfm High impedance design
fan
D
Static pressure

0.3 B E
100 cfm optimal design
fan high
0.2 low impedance design impedance
system
70 cfm F
0.1
fan
C low
Pressure increase impedance

Static pressure
Pressure loss system
2 fans
in series
0 20 40 60 80 100 120 ∆P
Air flow G cfm
Figure 5.30. Fan characteristics of three different fans.
single
fan 2 fans in
System flow parallel
Once the volume of air and the static pressure of the system to be cooled are known, their intersection
specifies the fan, specifically its airflow. To find the most effective fan for a system the typical airflow ∆G
curve is divided into four sectors, as in Figure 5.27b. In general, each type of air mover will be best
suited to one area, from high-flow/low-pressure to high-pressure/low-flow, as shown. Any given fan can Air flow G
only deliver one flow at one pressure in a given system.
Figure 5.27a shows a typical fan pressure versus flow curve along with what is considered the normal
operating range of the fan. The fan, in any given system, can only deliver as much air as the system will Figure 5.31. The effects of multiple fans (series versus parallel fan operation)
pass for a given pressure. If the estimated value of the actual airflow is significantly less than the on system pressure and flow rate.
required value, before increasing the number of fans in a systems, or attempting to increase the air
volume using a larger fan, the system should be analyzed for possible reduction in the overall resistance 5.12.2 The fan (affinity) laws
to airflow. Other considerations, such as available space and power, noise, reliability, and operating
environment should also be considered. It may be necessary to determine the output of a given fan under other operating conditions of speed or
fluid density, or to convert the known performance of an air mover of one size to that of another
185 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 186

geometrically similar unit of a different size. The fan laws permit this and geometrically similar fans can Solution
be characterized by the following five equations:
The new flow rate, from equation (5.68), is
Volumetric Flow-rate: G = Kq ND 3 P 0.05 × P 0.05 × 1200W
G = = = = 4m3 /s
Mass Flow Rate: mf = Km ρ AND 3 ρ × c p × ∆T ∆T 15K

Pressure: P = K p ρ AN 2D 2 (5.70) This is a flow increase of 1m3/s.

The volumetric flow rate G is given by equation (5.70)
Power: HP = KHP ρ AN 3D 5 G = K q ND 3
sound: Lw = 3.71log10 Vt + 0.96 log10 Q − 10.8 (imperial units ) that is
G2
rpm2 = × rpm1
where: Kq, Km, Kp, KHP = constants for geometrically and dynamically similar operation G1
G = volumetric flow rate, m3/s
4m3 /s
mf = mass flow rate, kg/s = × 3, 000rpm = 4, 000rpm
N = fan impeller speed, rps 3m3 /s
D = fan diameter, m The fan power requirements are given by
HP = impeller input power to rotate HP = K HP ρ N 3D 5
ρℓ = air density, kg/m3 that is
Lw = sound pressure level, dB 3
Vt = tip speed of impeller  rpm2 
Power 2 = Power1 ×  
 rpm1 
These five fan laws apply where the fan airflow rate and pressure are independent of Reynolds’s 3
number, Re, specifically when  4, 000 
= 8W ×   = 19.0W
πρ AND2
 3, 000 
Re = > 2 × 106 (5.71) The pressure increase produced is given by
CR ν
where CR = correction factor P = K p ρ N 2D 2
ν = absolute viscosity, Ns/m2 that is
2
 rpm2 
From these relationships, it is possible to calculate fan performance at a different condition. Table 5.17 Pressure2 = Pressure1 ×  
is a summary of the fan law equations in a dimensionless form, useful for fan analysis.  rpm1 
2
 4, 000 
= Pressure1 ×   = Pressure1 × 1.78
Table 5.17: Basic fan laws, scaling  3, 000 
The pressure is increased by 77.8%. The expected noise increase is 55×log10(4000/3000) = 6.87dBA.
♣
Variable Constants Fan Laws

Diameter (D) G2 = G1 (N2 / N1)

Speed (N) P2 = P1 (N2 /vN1)2
Density (ρℓ) 3
HP2 = HP1 (N2 / N1) Fan Efficiency
3
Speed (N) G2 = G1 (D2 / D1)
Diameter (D) P2 = P1 (D2 / D1)
2 The overall fan efficiency, ηf, is the ratio between power transferred to the airflow (power output from the
Density (ρℓ)
HP2 = HP1 (D2 / D1)5 impeller) and the electrical input power used by the fan. The fan efficiency is in general independent of
Diameter (D) 2 air density and can be expressed as:
P2 = P1 (p2 / p1)
Density (ρℓ) Speed (N) 5 P ∆p × G
Volumetric Flow Rate (G)
HP2 = HP1 (p2 / p1) ηf = f =
Pe Pe
G2 = G1 (p2 / p1) where
Diameter (D) P2 = P1 (p2 / p1)
Density (ρℓ)
Mass Flow Rate (m) N2 = N1 (p2 / p1)
Pe is the electrical power input to the fan motor (W)
HP2 = HP1 (N2 / N1)
2 Pf is the fan power output calculated from volume flow (m³/s) and pressure developed (Pa).
G = air volume delivered by the fan (m3/s)
dp = static/total pressure (Pa)
The pressure could be total (mechanical) or static pressure and there can be a significant difference.
N  D  ρ 
sound: Lw 2 = Lw 1 + 55 log10  2  + 55 log10  2  + 20 log10  2 
 N1   D1   ρ1  The equation for determining mechanical efficiency is:
∆p (total pressure ) × G
ηf −mechanical = (5.72)
Pe

Example 5.12: Fan laws The static efficiency equation is the same except that the outlet velocity pressure is not added to the fan
static pressure
A chassis uses a single 120mm fan for cooling. The maximum acceptable temperature rise in the
enclosure is 15°C, when it dissipates 900W. A redesign results in the power dissipation increasing to ∆p ( static pressure ) × G
ηf −static = (5.73)
1200W. At 900W dissipation, the 120mm fan produces a 3m3/s flow rate at 3000rpm using 8W of power. Pe
What are the fan requirements at 1200W enclosure dissipation?
187 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 188

Table 5.18 Fan peak efficiencies Table 5.19: Air density ρℓ and thermal resistance change with altitude, z

Fan type Peak efficiency range, %

Altitude z Density ρℓ Heatsink thermal resistance multiplier, p.u.
Airfoil backward curved/inclined 79-83 Naturally Cooled
Fan-Cooled Fan-Cooled R (z )
m kg / m3 1

Centrifugal fan
Modified radial 72-79 (General) (High Power) =
θ

R O metres 1 − 5 × 10 z
−5

Radial 69-75 Sea Level 1.19 1.00 1.00 1.00

Pressure blower 58-68 1,500 1.06 1.20 1.16 1.10
3,000 0.904 1.45 1.35 1.21
Forward curved 60-65
4,500 0.771 1.77 1.58 1.33
Vane axial 78-85
6,000 0.652 2.18 1.86 1.48
axial fan

Tube axial 67-72 7,600 0.549

Propeller 45-50 9,100 0.458

Enclosure cooling
Density (altitude) effects on fan performance In addition to selecting a fan, it is important to consider fan placement in or on the enclosure.
Since a radial fan is a constant volume machine, it will move the same volume flow of air G independent Pressurizing (as opposed the evacuating) the enclosure is the preferred cooling method, since incoming
of the density of the air, as shown in figure 5.32. A fan is not a constant mass flow machine, therefore air can be readily filtered. In addition, a pressurized enclosure will prevent dust entering through cracks
mass flow m changes as the density changes. This is important when equipment must operate at or crevices. The fan is also transferring cooler, denser air, and therefore has a slightly higher-pressure
various altitudes. The mass flow m is directly proportional to density change ∆ρℓ, while the volume flow capability, which may be a slight advantage for low heat dissipating systems. An important feature of a
G remains constant. As air density decreased, mass flow decreases and the effective cooling diminishes pressurized system is that the fan life and reliability are increased due to the fan ambient temperature
proportionately. Therefore, equivalent mass flow is needed for equivalent cooling, or the volume flow being lower. The disadvantage of pressurization is that heat generated by the fan is dissipated into the
required at altitude (low-density air) will be greater than that required at sea level to obtain the equivalent enclosure.
heat dissipation.
When locating the fan or fans, the enclosure layouts illustration in Figure 5.33 highlight some desirable
cooling aspects. The airflow path will always take the path of least resistance, so use baffles to eliminate
impedance recirculation of the same air and to direct the airflow. Importantly, air is forced in at a lower level than the
performance at outlet vents in order to benefit from the chimney effect, where less dense (hot) air rises.
at sea level • Locate components with highest heat dissipation near the enclosure air exits.
sea level • Size the enclosure air inlet and exit vents at least as large as the Venturi opening
static pressure

of the fan used.

• Allow enough free area for air to pass with a velocity less than 7m/s.
performance • Avoid hot spots by spot cooling with a small fan.
at altitude impedance
at altitude • Locate components with the most critical temperature sensitivity nearest to inlet air
to provide the coolest airflow.
• Blow air into cabinet to keep dust out, that is, pressurize the cabinet.
• Use the largest area filter possible, in order to:
o increase dust capacity
o reduce pressure drop.
air flow

Figure 5.32. Density effects on fan performance. 5.12.3 Estimating fan life

Convection air-cooling is the most commonly used method of cooling power electronics. In order to
G is volumetric flow rate; a measurement of volume over time. It pertains to no particular gas or gas deliver air-cooled equipment with higher reliability, life expectancy of the air moving devices must be
density. G is strictly a rate of volume measurement. But within that volume of gas, and in this case - air, considered.
the quality of the air and its ability to transfer heat can be calculated. Every molecule of air has a mass,
and this mass has the ability to absorb or emit energy; also known as transferring heat. The number of Definition of Fan Failure
molecules for a given volume gives the density of the air (mass/volume). If more molecules of air are Fan parametric failures typically include excessive vibration, noise (+ 3dBA), rubbing or hitting of the
packed into a given volume, increasing the density, mass per volume increases and the ability to propeller, reduction in rotational speed (< 0.8×Nnom), increased running current (> 1.2×Inom), etc. Non-
transfer heat increases; and vice versa. functional failure include locked rotor and failure to start.
At sea level, the density of air is 1.19kg/m3, as seen Table 5.19. This value is created by all of the other Increased noise is a result of a bearing failure, which is usually caused by a loss of lubricant, which
molecules in the atmosphere weighing down on the molecules at sea level. As the elevation increases, leads to bearing wear.
there are fewer higher up molecules weighing down and the density of the air decreases. At non-sea
level altitudes, the heat transfer equation is recalculated using the appropriate density for the altitude The capacitor may fail in ac fans and the electronics may contribute to early failures in dc fans. Failure
that the fan is operating at. In equation (5.68), the specific heat, cp, which is mass dependant, is a criteria in fan life tests can also include a change in coast-down time or start time to reach full speed.
constant for a given molecule, viz. 1.021kJ/kg.K for air. Problems with winding insulation breakdown or similar, are classified as workmanship problems or an
out-of-control manufacturing process.
 189 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 190

ambient
air out Sometimes the mean time to failure (mttf) is also quote. For the Weibull distribution
return  1
air mttf = α × Γ  1 +  (5.77)
 ß
where Γ denotes the Gamma function.

am bient side
electronics

enclosure
side
Locate component with Size outlet openings at The mttf is often confused with the mean time between failures (mtbf). The mtbf should only be used in a
highest dissipation near least as large as Venturi repairable system setting. If a system uses ten fans, and any failed fan is promptly replaced, then the
air outlet opening of fan
supply mtbf may be used to understand the system's maintenance needs and service cost. But since the
air underlying hazard rate of the fans is not constant, computing the mtbf of a multiple-fan system is quite
ambient difficult. Instead, system reliability studies use a one-number hazard rate for the individual fans, in which
air in
case the average hazard rate may be appropriate.
Avoid hot spots
with small fan ambient Fan life estimation
cooling air out
The life of most fans is limited by the bearings. Electronics, even in dc fans, play a secondary role.
When temperatures range from 25 to 60°C, ball bearing fans on average outlasted sleeve-bearing fans
Allow enough free return air by 50%. When temperatures exceeded 70°C, ball bearing fans ran for 45,000 hours, while sleeve-
Use vertical baffles

am bient air in
area to pass with bearing fans became inoperable. Yet, when the ambient temperatures are relatively low, sleeve bearing

am bient side
to direct air flow

enclosure
velocity <7m/s
fans lasted as long as ball bearing fans. Therefore, if an application generates high levels of heat, a ball

side
bearing fan is used. If the equipment generates low heat intensities, or if the equipment has a short life
supply air span, a sleeve-bearing fan can be used.
Locate components
with most critical
Bearing life is generally limited by the grease life, which is primarily a function of temperature. Grease
temperature sensitivity
near the inlet air
ambient life is affected by the type of grease, percentage of grease fill, operating environment, load, and bearing
air out
design. The Booser grease life equation is based on grease life tests on electric motor bearings, and is
ambient valid for rolling-element bearings. The equation for the bearing grease life in the application is
air out kT
log L10 = -2.6 + - 0.301 × S ½ (5.78)
supply air T brg
where

am bient side

am bient air in
S ½ = SG + S N + S P

enclosure
(5.79)

side
return air
Use largest filter possible Blow air into the enclosure D ×N
i. increase dust capacity to keep dust out S N = 0.86 × H
ii. reduce pressure drop DN L
where
supply air DH × N × P
S P = 2.95 ×
ambient C r2
air out
P equivalent dynamic bearing load, kg
Figure 5.33. Cabinet cooling, external mount versus through-mount, with vertical orientation. N speed, rpm
Cr basic dynamic load Capacity, kg
DH bore diameter, mm
Reliability concepts DNL speed limit, rpm-mm
Experimentation and model fitting have shown that the Weibull distribution provides a good fit to fan life S½ half-life subtraction factor; for S½ = 1, the life falls 50%
data, because it accurately represents wear-out phenomena. For the Weibull distribution, the cumulative SG grease half-life subtraction factor, typically 0 for many greases
distribution function, a function of age t, is given by SN speed half-life subtraction factor
ß SP load half-life subtraction factor
- t
 

F (t ) = 1 - e α 
(5.74) kT grease temperature factor = 2450 for acceleration factor of 1.5 for each 10°C
Tbrg bearing temperature, K
where αs is the characteristic life (for example, 9780 hours) and
ßs is the shape parameter (for example, 4.9). This equation, however, does not account for the effect of grease quantity and may not cover all
available greases, particularly modern synthetic oils. For these new greases and depending upon the
Shape parameters for Weibull models fit to fan life are generally greater than 1, which means that a operating conditions, the results from the Booser equation may be conservative. Therefore, unless
fan's failure tendency increases with age (wear-out). The reliability function is 1 - F(t), which at any age t adjustment factors are available for a certain fan type, it is better to use the Booser equation to obtain a
represents the proportion of survivors from the original population. The Weibull hazard rate (also known qualitative comparison of two fan designs rather than an absolute life estimate.
as the failure rate or hazard function) is given by
βs -1
βs  t  Example 5.13: Fan lifetime
H (t ) = ×  (5.75)
αs  αs 
Calculate fan tenth percentile lifetime, L10, using the following fan data sheet information.
P = 960g,
Two metrics of fan reliability are the L2 life and L10 life, which are the second and tenth percentiles under Cr = 57kg,
some assumed fan life distribution, such as the Weibull distribution. L10 refers to the time at which 90% D = 3 mm
of a large population of fans continue to operate. Since F(t) = 0.1 at L10 and 0.02 at L2 in equation (5.74): N = 2200 rpm,
DNL = 270,000 rpm-mm,
L10 = αs × 0.105361 / ßs L2 = αs × 0.020201 / ßs (5.76)
Tbrg = 42°C when Tamb = 25°C

For example, given αs = 100k POH (power on hours) and ßs = 1.5, L2 = 7, 418 hours represents the age Solution
at which 98% of the population is expected to still be operating. The advantage of specifying an L2 life in
place of L10 life, is that the desired early life failure distribution is more tightly specified. The half-life subtraction factor, from equation (5.79), is calculated as
191 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 192

D ×N 3mm×2200rpm Although Booser's nominal temperature acceleration factor applies specifically at a bearing temperature
S N = 0.86 × = 0.86× = 0.021 of 100°C, no model exists for a room temperature of 25°C.
DN L 270,000rpm-mm
D ×N ×P 3mm×2200rpm×0.96kg A reliable fan will maintain system cooling and protect against system meltdown. A quality axial fan can
S P = 2.95 × = 0.28× = 0.540
C r2 57kg2 fulfil these requirements. Debate exists over which bearing (ball or sleeve) to use in the axial fan. The
thus bearing type is a crucial factor in determining an axial fan’s reliability. Table 5.20 outlines the relative
S = SG + SN + SP = 0 + 0.021 + 0.540 = 0.561 features of the two bearing types when used in axial fans.
kT =2450/1.5(25-25)/10 = 2450/1.50 = 2450, that is, assume no temperature derating.
From equation (5.78)
Table 5.20: Fan ball and sleeve bearing comparison
2450
log L10 = -2.6 + - 0.301 × 0.561
273.2°C + 42°C Criteria Ball Bearing Sleeve Bearing
= -2.6 + 7.78 - 0.169 = 5.01
Fan Longevity Longer life Shorter life
The resulting life estimate is L10 = 102,000 hours.
♣ Heat Endurance Higher Lower
Fan Mounting Options Vertical, shaft centre line parallel, perpendicular Vertical
In situations where fan reliability is critical, limit the bearing temperature rise to 10°C, particularly when a
Noise Emission Quieter at high speeds Quieter at low speeds in early life
single fan failure results in a system shutdown. The Booser life estimate can also be significantly
affected by the bearing load and the bearing size. Installing a fan with the shaft mounted vertically will Parts Precision Non-precision
result in a lower bearing load and a longer fan life. Using a larger bearing will also yield a longer fan life.
Lubricant Less evaporation More evaporation
Fan life experiments Contact Point Line
Because of economic and time constraints, a zero failure test strategy and/or accelerated testing
techniques are adopted. A zero failure test strategy may be used to estimate the test time required to Cost More expensive per unit Less expensive per unit
verify a life expectancy criterion such as a minimum L10 life. The precision of this approach depends on
the accuracy of the shape parameter assumption.
In summary, if the system has a short life span, or will not generate high levels of heat, a sleeve-bearing
fan can be used. However, if the application is a densely packed or a compact electronic system, a ball
Example 5.14: Fan testing bearing fan will endure hotter temperatures, have a greater life span, and ultimately provide a better
long-term investment.
How long should a sample size of 30 fans be tested to determine with 90% confidence that L10 is greater
than or equal to 80,000 hours, at 30°C?

Solution 5.13 Enhanced air cooling

Assuming a Weibull distribution, each of n fans should be tested t1 hours, with With ever increasing gravitational and volumetric power density demands, traditional air-cooling
1
 β techniques have reached their limit for cooling of high-power applications. With standard fans and
t1 = α χ 2 C  blowers, a maximum heat transfer coefficient of h = 150W/m2K can be reached with acceptable noise
 2; 2n  levels, which is about 1W/cm2 for a 60°C temperature difference. Using 'macrojet' impingement,
where χ 22 is the C-th percentile of the Chi Square distribution with two degrees of freedom; C is theoretically 900W/m2K may be reached, but with unacceptable noise levels. Dedicated non-standard
determined by the desired confidence level. From a Chi Square table, χ 2;2 0.90 = 4.60. fans - heat sink combinations for cooling have a maximum of about 50W/cm2. Advanced methods to
Assuming ß = 2, solving for α in equation (5.76) gives extend the useful range of air-cooling are piezo fans, 'synthetic' jet cooling, and 'nanolightning'.
α = L10 ( 0.10536 )
−½
= 246, 460 hours
Piezo fans:
Substituting α, with n = 30 into the Weibull distribution equation, gives t1 = 68,280 hours of test time for
Piezoelectric fans are low power, small, relatively low noise, solid-state devices that are viable thermal
each fan. If all 30 fans operate t1 hours, at 30°C, without failure, then it can be asserted with 90%
management solutions for a variety of portable power electronics applications in laptop computers and
confidence that L10 is at least 80,000 hours.
cellular phones. Piezoelectric fans utilize piezoceramic patches bonded onto thin, low frequency flexible
♣ blades to drive the fan at its resonance frequency. The resonating low frequency blade creates a
streaming airflow directed at the electronic components.
Accelerated Life Testing
Since life test durations are lengthy, even when a zero failure test strategy is used, accelerated testing 'Synthetic' jet cooling:
techniques are essential to complete component evaluation within a reasonable time and cost. Due to the periodic pulsating flow nature, synthetic jets introduce a stronger entrainment than
The first acceleration factor is on/off cycles. These cycles stress the fan by accelerating the bearing from conventional-steady jets of the same Reynolds number and more vigorous mixing between the wall
zero speed to normal speed. An on/off cycle every 8 hours would be representative of a personal boundary layers and the rest of the flow. A synthetic jet draws cool air from ambient, impinges on the top
computer application. Even if this degree of stress is not appropriate, some on/off cycles are required to hot surface and circulates the heated air back to the ambient through the edges of the plate. A radial
detect fan problems such as failure to start, changes in rotational speed, coast down time or start time, counter air current flow is created in the gap between the plates with hot air dispersed along the top and
and increased noise. ambient air entering along the bottom surface.
Elevated temperature is generally the primary acceleration factor. The range of acceleration factors
typically used in fan reliability calculations is 1.3 to 2 per 10°C increase. For fan failures caused by 'Nanolightning':
lubricant breakdown, it is reasonable to use the acceleration factor of 1.5 per 10°C increase as in 'Nanolightning' increases the heat transfer coefficient with 'micro-scale ion-driven airflow' using high
Booser's equation. For example, to extrapolate the results of a life test run at 80°C down to 40°C, use a electric fields created by nanotubes. The ionized air molecules (tiny wind currents) are moved by
decreasing acceleration factor of 1.5(80-40)/10 = 5.1. another electric field, thereby inducing secondary airflow. Cooling at a heat flux level of 40W/cm2 (similar
At the accelerated life test temperature, there should not be a significant change in grease structure. to water) is possible.
The performance of the grease is degraded mainly due to evaporation loss and oxidation. Accelerated
life testing should therefore be conducted at air temperatures below 85°C.
193 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 194

5.14 Liquid coolants for power electronics cooling 5.14.2 Dielectric liquid coolants

Although air-cooling continues to be the most widely used method for cooling electronic packages, While the food industry might be more likely to select propylene glycol PG over ethylene glycol EG for
significantly higher heat fluxes (W/m2), due to the higher heat transfer coefficient are achievable with heat transfer, the power electronics, laser, and semiconductor industries might be more likely to choose
liquid cooling. Coolants are used in both single phase and two-phase applications. dielectric fluids over water. A dielectric fluid is non-conductive and therefore preferred over water when
• A single-phase cooling loop consists of a pump, a heat exchanger (cold plate/mini- or working with sensitive electronics. Perfluorinated carbons, such as dielectric fluid Fluorinert, are non-
micro-channels), and a heat sink (radiator with a fan or a liquid-to-liquid heat exchanger flammable, non-explosive, and thermally stable over a wide range of operating temperatures. Deionised
with chilled water-cooling). The heat source in the power electronics system is attached to water is also non-conductive, Fluorinert is less corrosive than deionised water and therefore may be a
the heat exchanger. The fluid does not change state: water does not change to steam. better choice for some applications. However, water has a thermal conductivity of approximately
• Liquid coolants are also used in a change of state or two-phase systems, such as heat 0.59W/m°C, while Fluorinert FC-77 has a thermal conductivity of only about 0.063W/m°C. Fluorinert is
pipes, thermo-siphons, sub-cooled boiling, spray cooling, and direct immersion systems. also much more expensive than deionised water.
A heat flux of 2kW/cm2 can be removed through boiling water, based on water molecules turning into
vapour without influencing each other, using high velocities and high pressures. Available microcoolers Table 5.21: Properties of different liquid coolant chemistries at 20°C
can handle about 1kW/cm2. Liquid cooling for power electronics applications is generally divided into the
two main categories of indirect and direct liquid cooling. freezing flash thermal specific
viscosity density
• Indirect liquid cooling is one in which the liquid does not directly contact the components cooling
point point conductivity heat
to be cooled. ν λ cp ρℓ
chemistry
• Direct liquid cooling brings the liquid coolant into direct contact with the components to °C °C
-1
×10 kg.m .s
-1 -1 -1
W.m .s
-1
[Link].K
-1
kg.m
-3

be cooled.
The following sections discuss indirect liquid cooling in the form of heat pipes and cold plates and direct Aromatic (DEB) < -80 57 1 0.14 1700 860
liquid cooling in the form of immersion cooling and jet impingement. Silicate-ester
< -50 > 175 9 0.132 1750 900
(coolanol 25R)
Liquid cooling can reduce the effective thermal resistance to as low as 0.01K/W. Both oil and water Aliphatic
< -50 > 175 9 0.137 2150 770
(which has 4 times the thermal capacity and 770 times the density of air) are used as the coolant and (PAO)
the heat-sink arrangement can either be immersed in the fluid (direct), or the fluid is pumped through a Silicone
< -110 46 1.4 0.11 1600 850
(syltherm XLT)
hollow heat sink (indirect). The fluid heat is dissipated remotely. Fluid boiling with direct cooling should
Fluorocarbon
be avoided, since the creation of air bubbles can cause local hot spots. Water has the advantage of low (FC-77)
< -100 none 1.1 0.06 1100 1800
viscosity, so can be pumped faster than mineral oil. While oil may be inflammable, water corrodes thus
EG/water
requiring the use of de-ionised water with an oxide inhibitor, like antifreeze (ethylene glycol). Oil -37.8 none 3.8 0.37 3285 1087
50:50 vol
emersion has the added advantage of offering possibilities of increasing the breakdown and corona PG/water
voltage levels, particularly with devices rated and operated at voltages above a few kilovolts. -35 none 6.4 0.36 3400 1062
50:50 vol
Methanol/water
-40 29 2 0.4 3560 935
5.14.1 Requirements of a liquid coolant 40:60 wt.
Ethanol/water
-32 27 3 0.38 3500 927
There are many requirements for a liquid coolant for power electronics applications and vary depending 44:56 wt
on the type of application. Some of the general requirements are: Potassium
• Good thermo-physical properties (high thermal conductivity and specific heat; low viscosity; high Formate/water -35 none 2.2 0.53 3200 1250
latent heat of evaporation for two-phase application) 40:60 wt
• Low freezing point and burst point. The burst point (or solidification temperature) is the Ga-In-Sn -10 none 2.2 39 365 6363
temperature associated with expansion of a freezing coolant.
• High atmospheric boiling point (or low vapour pressure at the operating temperature) for a single
phase system; a narrow desired boiling point for a two-phase system Polyalphaolefin, PAO, is a synthetic hydrocarbon used frequently in military and aerospace applications
• Good chemical and thermal stability for the life of the power electronics system for its dielectric properties and wide range of operating temperatures. PAO compatible recirculating
• High flash point and auto-ignition temperature (sometimes non-combustibility is a requirement) chillers are available for cold plates and heat exchangers that use PAO as the heat transfer fluid. PAO
• Non-corrosive to the construction materials (metals as well as polymers and other non-metals) has a thermal conductivity of 0.14 W/m°C. Although dielectric fluids provide low risk liquid cooling for
• No or minimal regulatory constraints (environmentally friendly, non-toxic, and biodegradable) electronics, they generally have a much lower thermal conductivity than water and most water-based
• Economical solutions.
The best electronics coolant is an inexpensive and non-toxic liquid with excellent thermo-physical • Aromatics: Synthetic hydrocarbons of aromatic chemistry (that is, diethyl benzene DEB,
properties and a long service life. A high flash point and auto-ignition temperature are desired so that the dibenzyl toluene, diaryl alkyl, partially hydrogenated terphenyl) are common heating and
fluid is less susceptible to ignition. Good thermo-physical properties are required to obtain the high heat cooling fluids used in a variety of applications. However, these compounds cannot be
transfer coefficients and low pumping power needed for the fluid to flow through a tube or a channel. classified as non-toxic. Also, some of these fluids (namely, alkylated benzene) have strong
odours, which can be irritating to the personnel handling them.
Electrical conductivity (not mentioned in the list) of a coolant becomes important if the fluid comes in • Silicate-ester: This chemistry (that is, Coolanol 25R) was widely used as a dielectric coolant
direct contact with the power electronics (such as in direct immersion cooling), or if it leaks out of a in airborne radar and missile systems. These fluids have caused significant and sometimes
cooling loop or is spilled during maintenance and comes in contact with the electrical circuits. In certain catastrophic problems due to their hygroscopic nature and subsequent formation of flammable
applications, a dielectric coolant is necessary, whereas in many other applications it is not a requirement alcohols and silica gel. Therefore, these fluids have been replaced by more stable and
because of the remote chance of coolant leakage (or in case of a leak, the coolant does not come in dielectric aliphatic chemistry (polyalphaolefins or PAO).
contact with the power electronics). • Aliphatics: Aliphatic hydrocarbons of paraffinic and iso-paraffinic type (including mineral oils)
The various liquid coolant chemistries are divided into dielectric and non-dielectric fluids and their are used in a variety of direct cooling of electronics parts as well as in cooling transformers.
properties: refer to Table 5.21 and Table 5.32X. Water, deionised water, glycol/water solutions, and Many petroleum based aliphatic compounds meet the criteria for incidental food contact.
dielectric fluids such as fluorocarbons and polyalphaolefins PAO are the heat transfer fluids commonly These petroleum-based fluids do not form hazardous degradation by-products. Most of these
used in high performance liquid cooling applications. The heat transfer fluid must be compatible with the fluids have a non-discernible odour and are non-toxic in case of contact with skin or ingestion.
fluid path, offers corrosion protection or minimal risk of corrosion, and meet the application’s specific As mentioned, aliphatic PAO-based fluids have replaced silicate-ester fluids in a variety of
requirements. military electronics (and avionics) cooling applications in the last decade.
195 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 196

• Silicones: Another class of coolant chemistry is dimethyl- and methyl phenyl-poly (siloxane) or everything to which it is exposed. Therefore, all materials in the cooling loop must be
commonly known as silicone oil. Since this is a synthetic polymeric compound, the molecular corrosion-resistant. Copper and many other common materials are not compatible with DI
weight and the thermo-physical properties (freezing point and viscosity) can be adjusted by water and will contaminate it.
varying the chain length. Silicone fluids are used at temperatures down to -100°C and as high • Ethylene Glycol (EG): Commonly used as antifreeze in automotive engine cooling, EG also
as 400°C. These fluids have excellent service life in closed systems in the absence of oxygen. has found use in many industrial cooling applications, at lower temperatures. Ethylene glycol is
Also, with essentially no odour, the non-toxic silicone fluids are workplace friendly. However, colourless and practically odourless and is completely miscible with water. Ethylene glycol has
low surface tension gives these fluids the tendency to leak around pipe-fittings, although the desirable thermal properties, including a high boiling point, low freezing point, stability over a
low surface tension improves the wetting property. Similar to the aliphatics, high molecular wide range of temperatures, and high specific heat and thermal conductivity. It also has a low
weight silicone oils are used in cooling transformers. viscosity and, therefore, reduced pumping requirements. When properly inhibited, it has a
• Fluorocarbons: Fluorinated compounds such as perfluorocarbons (that is, FC-72, FC-77) relatively low corrosivity. However, this coolant is classified as toxic and should be handled
hydrofluoroethers (HFE) and perfluorocarbon ethers (PFE) have certain unique properties and and disposed of with care. Typically, water with low chloride and sulphate ion concentration
can be used in contact with the electronics. First, these fluids are non-combustible and non- (<25ppm) is recommended. Also, a monitoring schedule should be maintained to assure that
toxic. Some fluorinated compounds have zero ozone depleting potential and other inhibitor depletion is avoided and the pH of the solution is consistent. Once the inhibitor has
environmental properties. Second, some of these fluids have low freezing points and low been depleted, the old glycol should be removed from the system and a new charge installed.
viscosities at low temperatures. However, these fluids are expensive, have poor thermal Even though EG’s thermal conductivity is not as high as water’s, EG provides freeze protection
properties, some have global warming potential (greenhouse effect), and, due to the extremely that can be beneficial during use or during shipping. Although ethylene glycol is the chemical
low surface tension, leaks can develop around fittings. used in automotive antifreeze, it should not be used in a cooling system or heat exchanger
because it contains silicate-based rust inhibitors. These inhibitors can gel and foul, coating
5.14.3 Non-dielectric liquid coolants heat exchanger surfaces, reducing their efficiency. Silicates have also been shown to reduce
Non-dielectric liquid coolants are often used for cooling electronics because of their superior thermal significantly the lifespan of pump seals. While the wrong inhibitors can cause significant
properties, as compared with the dielectric coolants. Non-dielectric coolants are normally water-based problems, the right inhibitors can prevent corrosion and significantly prolong the life of a liquid
solutions. Therefore, they possess a high specific heat and thermal conductivity. De-ionized water is an cooling loop. Inhibited glycols are recommended over non-inhibited glycols.
example of a widely used coolant. As the concentration of glycol in the solution increases, the thermal performance of the heat
transfer fluid decreases. Therefore, it is best to use the lowest possible concentration of
• Water: Water is a suitable choice for liquid cooling applications due to its high heat capacity inhibited glycol necessary to meet corrosion and freeze protection needs. Dow Chemical
and thermal conductivity. It is also compatible with copper, which is one of the best heat recommends a minimum concentration of 25 to 30% EGW4. At this minimum concentration,
transfer materials to use in the fluid path. Water used for cooling comes from different sources. the ethylene glycol also serves as a bactericide and fungicide. With recirculating chillers, a
The benefit of using facility or tap water is that it is readily available and inexpensive. However, solution of 30% ethylene glycol results in only a 3% drop in thermal performance over using
facility water or tap water is likely to contain impurities, which cause corrosion in the liquid water alone but will provide corrosion protection as well as freeze protection down to -15°C.
cooling loop and/or clog fluid channels. Therefore, use good quality water in order to minimize The quality of the water used in the glycol solution is also important. Even with an inhibited
corrosion and optimize thermal performance. glycol, water ions can cause inhibitor precipitation, resulting in fouling and corrosion.
Water’s ability to corrode metal can vary considerably depending on its chemical composition. • Propylene Glycol (PG): In its inhibited form, PG has the same advantages of low corrosivity
Chloride, for example, is commonly found in tap water and can be corrosive. Facility or tap shown by ethylene glycol. In addition, propylene glycol is considered non-toxic. Other than lack
water should not be used in liquid cooling loops if it contains more than 25ppm of chloride. The of toxicity, it has no advantages over ethylene glycol, being higher in cost and more viscous.
levels of calcium and magnesium in the water also need to be considered, since calcium and Although EG has more desirable physical properties than PG, PG is used in applications
magnesium form scale on metal surfaces and reduce the thermal performance. where toxicity might be a concern. PG is generally recognized as safe for use in food or food
If the facility water or tap water contains a large percent of minerals, salts, or other impurities, processing applications, and can also be used in enclosed spaces.
the water can be purchased filtered or deionised. If the tap water is relatively pure and meets • Methanol/Water: This is a low cost antifreeze solution, finding use in refrigeration services
recommended limits, a corrosion inhibitor should be added for additional protection. Phosphate and ground source heat pumps. Similar to glycols, it can be inhibited to stop corrosion. This
is an effective corrosion inhibitor for stainless steel and most aluminium components. It is also fluid can be used down to -40°C owing to its relatively high rate of heat transfer in this
effective for pH control. One disadvantage of phosphate, however, is that it precipitates with temperature range. Its main disadvantages as a heat transfer fluid are its toxicological
calcium in hard water. For copper and brass, tolyltriazole is an effective corrosion inhibitor. For considerations. It is considered more harmful than ethylene glycol and consequently has found
aluminium, organic acids such as 2-ethyl hexanoic or sebacic acid offer protection. use only for process applications located outdoors. Also, methanol is a flammable liquid and,
• Deionised Water: Tap water meets the needs of most liquid-cooling applications. However, as such, introduces a potential fire hazard where it is stored, handled, or used.
deionised (DI) water has chemical and electrical properties that make it the optimal choice for • Ethanol/Water: This is an aqueous solution of denatured grain alcohol. Its main advantage is
cooling when the liquid circuit contains micro-channels or when sensitive electronics are non-toxicity. Therefore, it has found application in breweries, wineries, chemical plants, food
involved. Deionised water is water that has an extremely low concentration of ions, including freezing plants, and ground source heat pumps. As a flammable liquid, it requires certain
sodium, calcium, iron, copper, chloride, and bromide. The lack of ions in DI water eliminates precautions for handling and storage.
the following problems. • Calcium Chloride Solution: Aqueous solutions of calcium chloride are used as circulating
First, it eliminates mineral, salts and impurities that can cause corrosion or scale formation and coolants in food plants. It is non-flammable, non-toxic and thermally more efficient than the
block the coolant flow. This will degrade cooling efficiency and system operating performance. glycol solutions. A 29% (by weight) calcium chloride solution has a freezing point below -40°C.
Second, it eliminates the risk of electrical arcing due to static charge build up from the The main disadvantage of this coolant is that it is highly corrosive, even in the presence of
circulating coolant. The arcing can damage sensitive control electronics in the equipment corrosion inhibitors.
being cooled. Compared to tap water and most fluids, deionised water has a high resistivity, • Potassium Formate/Acetate Solution: Aqueous solutions of potassium formate and acetate
and as an excellent insulator, is used in the manufacturing of electrical components salts are non-flammable and non-toxic as well as much less corrosive and thermally more
where parts must be electrically isolated. However, as water’s resistivity increases, its efficient than calcium chloride solution. Therefore, even a with higher price than calcium
corrosivity increases. Deionised water has a pH of approximately 7.0 but quickly becomes chloride, they have found a large number of applications, in the food, beverage,
acidic when exposed to air. The carbon dioxide in air dissolves in the water, introducing ions, pharmaceuticals, chemical and climatic chamber applications, and single-phase convection
giving an acidic pH of around 5.0. Therefore, when using water that is virtually pure, it is cooling of microprocessors.
necessary to use a corrosion inhibitor. When using deionised water in a recirculating chiller, • Liquid Metals: Liquid metals of Ga-In-Sn chemistry are utilized with a magneto-fluid-dynamic
special high purity plumbing is needed. The fittings should be nickel-plated and the (MFD) pump. It utilizes the high thermal conductivity and density of the metal alloy to remove
evaporators should be nickel-brazed. When using deionised water in cold plates or heat high heat flux from the heat source.
exchangers, stainless steel tubing is recommended.
The lack of ions makes this coolant unusually corrosive. Called the ‘universal solvent’, DI water
is one of the most aggressive solvents known. In fact, to a varying degree, it will dissolve
197 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 198

5.15 Direct and indirect liquid cooling Heat in Metal wick structure Heat out
Saturation Liquid flow
Application of liquid cooling for power electronics may be categorized as either indirect or direct. curve
Indirect liquid cooling is one in which the liquid does not contact the power electronic chips, nor the Adiabatic
substrate upon which the chips are mounted. In such cases, a good thermal conduction path is section

Temperature
provided from the power electronic heat sources to a liquid cooled cold-plate attached to the 2
module surface, as shown in Figure 5.34. Since there is no contact with the electronics, water can 2 3
be used as the liquid coolant, taking advantage of its superior thermo-physical properties.
evaporator 2′
Direct liquid cooling involves the components to be cooled being immersed in and in direct contact 3
with the cooling fluid, as with oil-immersed transformers. 4
4 condenser Isobar shell
1
Thermal path Heat in Heat out
to cap die wick 1 4
Saturated Saturated
Liquid out liquid state vapour state

fins/pins Entropy v J/kg.K

Figure 5.35. Heat pipe thermodynamic T-v operational vapour power cycle.

Liquid in Heat pipes are an efficient, reliable, passive, silent, high thermal conducting, evacuated, sealed
cylindrical device for quickly extracting, transporting, and remotely dissipating heat, in any orientation.
Liquid cooled Although there is virtually no limit to the size of a heat pipe, the effectiveness of a heat pipe decreases
substrate cap
Cold-plate
with decreasing lengths. For heat pipes with a length less than about 1 cm the performance of a solid
piece of metal, copper, is comparable. They are effective as efficient heat conductors to transport heat
Figure 5.34. Example of indirect and direct liquid immersion cooling for a disk package. to locations were more area is available. 2D heat spreaders (otherwise known as vapour chambers)
based on the heat pipe principle can achieve much higher effective thermal conductivities than copper.
A thin planar heat spreader has a thermal performance greater than diamond.
Loop heat pipes (LHP) have the advantage over conventional heat pipes that the vapour and liquid
paths are separated enabling much better performance of the liquid return loop, accommodate a heat
5.16 Indirect liquid cooling flux of 625W/cm2.
5.16.1 Heat pipes – indirect cooling

Heat pipes provide an indirect and passive (no moving parts) means of applying liquid cooling. It is a
two-phase device. A heat pipe can be used in situations when a heat source and a heat sink need to be
placed apart. They are vacuum pumped (evacuated) and sealed vessels that are partially filled with a
liquid, usually high purity water or alcohol in a saturated vapour form. The internal walls of the pipes are
lined with a porous medium (the wick) that acts as a passive capillary pump. When heat is applied to
one end of the pipe (evaporator), the liquid starts evaporating. A pressure gradient exists causing the
vapour to flow to the fractionally cooler regions. The vapour condenses (with the latent heat of
vaporisation transferred to the condenser) back to the liquid state at the cooler regions and is
transported back by the capillary wick structure (or gravity), thereby closing the fluid two-state
thermodynamic loop.
cooling
This thermodynamic cycle can be summarised, with the aid of figure 5.35, as follows: fins

Stage 1-2: Heat applied to the evaporator through the external source vaporizes the working fluid to a Tamb
adiabatic section
saturated (2′) or superheated (2) vapour.
Stage 2-3: Vapour pressure drives vapour through the adiabatic section to the condenser. evaporator section
evaporation
Stage 3-4: Vapour condenses, releasing heat to a heat sink.
Tj
Stage 4-1: Capillary pressure created by wick menisci pumps condensed fluid into evaporator section.
condenser section
Then the continuously circulating process starts over. condensation
heat source

Heat pipes provide an enhanced means of transporting heat, much better than copper, from a source to
a heat sink where it is rejected to the cooling medium by natural or forced convection. The effective Figure 5.36. Example of heat pipes used in a notebook application.
thermal conductivity of a heat pipe can range from 50kW/mK to 200kW/mK, but is often lower in practice
due to additional interface thermal resistances. The performance of heat pipes scales from 10 W/cm2 to
over 300 W/cm2. A simple water-copper heat pipe has an average heat transfer capacity of 100 W/cm2, A heat pipe is designed for a certain temperature range. Apart from the vapour temperature range,
a thermal conductivity that is in excess of 300 times better than that of an equivalently sized pure copper factors like thermal stability and thermal conductivity influence the choice of working fluid. Suitable
component. A heat pipe provides efficient transport of concentrated heat. An example of a typical working fluids, which desirably have a high surface tension, include (see figure 5.42):
application of a heat pipe for an electronics cooling application is given in Figure 5.36. • For ultra low temperatures: inert gases (helium), nitrogen, ammonia
• For usual temperatures to meet power electronics cooling requirements: distilled water
with various additions, organic fluids like acetone, methanol, ethanol, and toluene.
• For high temperatures: metals like mercury, sodium, silver.
199 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 200

The component to be cooled is mounted on the evaporator end (the hot end), where the heat boils and An advantage of the sintered powder wick is its ability to handle high heat fluxes. Since sintered powder
expands the liquid to the vapour phase, increasing the pressure. Boiling occurs because energy, in the wicks are generally 50% porous, there is accordingly a large surface area available for evaporation.
form of heat, is taken from the surrounding area, which cools the heat source. This vapour rises through Typical sinter powder wicks handle 50 W/cm2, and up to 250 W/cm2. In comparison, a groove wick
the adiabatic tube section (low-pressure drop, hence low temperature change) to the remote condenser nominally handles 5 W/cm2 and a screen wick will nominally handle 10 W/cm2. Since a sintered powder
end of the tube (the cold end), taking the heat within it. Effectively the heat is transported at the rate P: wick is integral with the heat pipe envelope, and the fluid charge is only enough to saturate the wick, the
dm heat pipe can be subjected to freeze/thaw cycles with no degradation in performance.
P =L
dt
where L is the heat of vaporization per unit mass and dm/dt is the mass evaporation rate. Working fluid or coolant
A first consideration in the identification of a suitable working fluid is the operating vapour temperature
The vapour condenses back to the liquid phase, releasing its latent heat of vaporisation, and creates a range. Within the approximate temperature band, several possible working fluids may exist, and a
pressure gradient, which helps draw more vapour towards the condenser. The temperature difference variety of characteristics must be examined in order to determine the most acceptable of these fluids for
between the ends may only be a couple of degrees. The remotely situated condenser end is connected the application considered. The prime requirements are:
to an external heatsink or a radiator type grill, for cooling. The condensed working fluid runs back to the • compatibility with wick and wall materials
evaporator end due to gravity, or due to capillary pressure action, along porous capillaries that form a • good thermal stability
wick, depending on the physical application orientation design for the heat pump. • wettability of wick and wall materials
• vapour pressure not too high or low over the operating temperature range
The quality and type of wick usually determines the performance of the heat pipe. • high latent heat
The wick is a porous structure made of materials like steel, aluminium, nickel or copper in various • high thermal conductivity
ranges of pore sizes. They are fabricated using metal foams, and more commonly, felts. By varying the • low liquid and vapour viscosities
pressure on the felt during assembly, various pore sizes can be produced. By incorporating removable • high surface tension
metal mandrels, an arterial structure can be moulded in the felt. • acceptable freezing or pour point
The two most important properties of a wick are the pore radius and the permeability. The pore radius
determines the pumping pressure (the maximum capillary head) the wick can develop. The wick The typical temperature operating range is within the bounds -55°C to over 200°C, depending on the
permeability increases with increasing pore size. The permeability determines the frictional losses of the coolant, as shown in Table 5.23 (see figure 5.42). Heat pipes can be designed to operate over a broad
fluid as it flows through the wick. The heat transport capability of the heat pipe is raised by increasing range of temperatures from cryogenic (< -243°C) applications utilizing titanium alloy/nitrogen heat pipes,
the wick thickness. The overall thermal resistance at the evaporator also depends on the conductivity of to high temperature applications (>2000°C) using tungsten/silver heat pipes. In power electronic cooling
the working fluid in the wick. applications where junction temperatures below 125 to 150°C are desired, copper/water heat pipes are
used. Copper/methanol heat pipes are used if the application requires heat pipe operation (and
There are several types of wick (capillary) structures available including (decreasing permeability and importantly, start up) below 0°C. Water heat pipes, with a temperature range from 5 to 230°C, are less
decreasing pore radius): grooves, screen, cables/fibred, and sintered powder metal, shown in figure sensitive than methanol to orientation, and are most effective for power electronics cooling applications,
5.37 and summarized in Table 5.22. Specifically: plus copper vessels are compatible with water.
Sintered powder: This wick provides high power handling, low temperature gradients, and high
capillary forces for anti-gravity applications. The complex sintered wick has several vapour Table 5.23: Heat pipe fluids, in increasing operating temperature range (see Table 5.29)
channels and small arteries to increase the liquid flow rate. Tight bends in the heat pipe can
be achieved with this type of structure. Melting Boiling pt @ Measured Measured Operating
Medium Vessel material
Grooved tube: The small capillary driving force generated by the axial grooves is adequate for low point atm pressure axial heat flux surface heat flux range
2 2
power heat pipes when operated horizontally, or with gravity assistance. The tube can be °C °C kW/cm W/cm °C
readily bent. When used in conjunction with screen mesh the performance can be enhanced. Helium (ℓ) - 271 - 261 -271 to -269
Screen mesh: This type of wick is the most common and provides readily variable characteristics in
0.067 1.01
terms of power transport and orientation sensitivity, according to the number of layers and Nitrogen (ℓ) - 210 - 196 Stainless steel -200 to -80
@ -163°C @ -163°C
mesh counts used.
Ni, Al, stainless
Fibre/spring: Fibrous materials, like ceramics, generally have smaller pores. The main Ammonia (ℓ) - 78 - 33 0.295 2.95 -70 to 100
steel
disadvantage of ceramic fibres is minimal stiffness and requires support by a metal mesh.
Carbon fibre filaments have many fine longitudinal grooves on their surface, have high Cu, Ni, Al,
Acetone - 95 57 0 to 120
capillary pressures, are chemically stable, and show a greater heat transport capability. stainless steel
0.45 75.5 Cu, Ni, stainless
Methanol - 98 64 -45 to 120
@ 100°C @ 100°C steel
Table 5.22: Wick material properties Flutec PP2 - 50 76 10 to 160
Ethanol - 112 78 0 to 130
Conductivity Overcome Thermal Conductivity lost
Wicking Material Stability 0.67 146
(Straight) Gravity Resistance (bended and flatten) Water 0 100 Cu, Ni 5 to 230
@ 200°C @ 170°C
Axial Groove Good Poor Low Good Low/Average
Toluene - 95 110 50 to 200
Screen Mesh Average Average Average Average Low/Average 25.1 181
Mercury/Mg - 39 361 Stainless steel 190 to 600
Fine Fibre Poor Good High Poor Average @ 360°C @ 360°C
5.6 181 Stainless steel,
Sintering (powder) Average Excellent High Average High Potassium 63 759 400 to 800
@ 750°C @ 750°C Ni
9.3 224 Stainless steel,
Sodium 98 892 500 to 1000
Since grooved wicks have a large pore radius and a high permeability, the pressure losses are low but @ 850°C @ 760°C Ni, Nb
the pumping head is also low. Grooved wicks can transfer high heat loads in a horizontal or gravity 2 207 W, Ti, Niobium
Lithium 179 1340 900 to 1600
aided position, but cannot transfer large loads against gravity. The powder metals have small pore radii @ 1250°C @ 1250°C +1% Zirconium
and relatively low permeability. Powder metal wicks are limited by pressure drops in the horizontal Tantalum
Silver 960 2212 4.1 413 1500 to 2200
position but can transfer large loads against gravity. +5% Tungsten
201 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 202

The capillary or lifting height H is given by A solid rod of copper conducts heat by diffusion, and a constant, geometry-independent thermal
2γ cos θ conductivity can be defined for the material. A heat pipe, however, conducts heat by transport of the
H = (5.80) vapour and therefore is more correctly thought of as a heat carrying device rather than a simple thermal
rc g f ρ A
resistor. A heat pipe carries heat with a certain temperature drop ∆T, but ∆T is virtually independent of
where γ is the surface tension, N/m, unique for each working fluid at certain temperatures, the length of the heat pipe. Thus, heat pipes not only have low thermal resistance R (small ∆T) but also
θ is the contact angle, rad, differs for each working fluid, have the property that R is roughly independent of length. The advantage of using a heat pipe, rather
rc the effective capillary radius, m, is a unique characteristic of the wick type used, than a rod of Cu metal, therefore increases with increasing length. The heat transfer or transport
gf is the gravitational acceleration, m/s2, assumed constant, and capacity of a heat pipe is specified by its Axial Power Rating APR. This is the energy moving axially
ρℓ is the working fluid density, kg/m3, dependant on the application conditions. along the pipe. As in equation (5.81), the larger the heat pipe diameter, the higher the APR, while the
The heat power transfer capabilities (axial power rating, APR) of a heat pipe are related to its cross- longer the heat pipe, the lower the APR.
sectional area Ax and length ℓ according to equation (5.3) The fact that the heat pipe functions at all in an ‘upside-down’ orientation, as shown in figure 5.40, is due
Ax to the wicking of the condensed fluid against gravity.
PD = k (W) (5.81)
A
There are five primary heat-pipe heat transport limitations, summarized in Table 5.24, which are a
while the temperature difference ∆T between the hot and cold ends is
function of the heat pipe operating temperature, as shown in figure 5.38:
 1 1  • Viscous limit – Heat pipes will not function when the pipe temperature is lower than the freezing
∆T = k ′ PD  +  (K) (5.82)
 Ae Ac  point (zero viscosity) of the working fluid. At low temperatures (just above freezing), the vapour
where Ae and Ac are the effective evaporator and condenser surface areas. pressure difference between the condenser and the evaporator may not be enough to
overcome viscous forces. The vapour from the evaporator cannot move to the condenser and
As a heat pipe has no moving parts, its mttf is estimated to be over 100,000 hours of use. Improper thus the thermodynamic cycle does not occur. Freezing and thawing may destroy the sealed
bending and flattening of the pipe may cause leakage at the pipe seal. There are some external factors joint of a heat pipe when place vertically.
that may also shorten the life of a heat pipe such as shock, vibration, force impact, thermal shock, and • Sonic limit – The rate that vapour travels from the evaporator to the condenser. The limit occurs
corrosive environment. when the vapour velocity reaches sonic speed at the evaporator and any increase in pressure
difference cannot speed up the flow. This usually occurs during heat pipe start-up.
Unlike Peltier elements (see 5.21.1): • Capillary pumping limit – the rate at which the working fluid travels from the condenser to the
• a heat pipe does not consume energy or produce heat itself and evaporator through the wick. This limit occurs when the capillary pressure is too low to provide
• it is not possible to cool a device below ambient temperature using a heat pipe. enough liquid to the evaporator from the condenser. This leads to evaporator dry-out, which
prevents continuing the thermodynamic cycle and the heat pipe no longer functions properly.
• Entrainment or flooding limit – Friction between the working fluid and vapour that travel in
opposite directions. At high vapour velocities, droplets of liquid in the wick are extracted from
the wick and deposited into the vapour. This results in dry-out.
• Nucleated boiling limit – the rate at which the working fluid vaporizes from the added heat. This
limit occurs when the radial heat flux into the heat pipe causes the liquid in the wick to boil and
evaporate causing dry-out.

Each limit has its own particular range of importance. However, in practice, the capillary and boiling
limits are the most important. Figure 5.38 illustrates the five limitation boundaries.
entrainment
Capillary limit

Sonic
limit

Power
Boiling
Viscous limit
limit

Temperature

Figure 5.38. Heat transport limitation boundaries of a heat pipe.

Axial Groove Fine Fibre Screen Mesh Sintering For a heat pipe to function, the net capillary pressure difference between the evaporator (heat source)
and condenser (heat sink) must be greater than the sum of all pressure losses occurring along the liquid
and vapour flow paths. This relationship, termed the capillary limitation, and for correct operation:
∆Pc max ≥ ∆Pliquid + ∆Pvapour + ∆Pg (5.83)
where ∆Pcmax is the maximum capillary pressure difference generated within the capillary wicking
structure between the evaporator and condenser,
∆Pliquid is the viscous pressure drops occurring in the liquid, to return the liquid from the
condenser to the evaporator,
∆Pvapour is the viscous pressure drops occurring in the vapour, to cause the vapour to flow from
Figure 5.37. Examples of heat pipes wicks. the evaporator to the condenser,
 203 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 204

∆Pg represents the hydrostatic pressure drop due to the gravitational head, ρℓgℓsinφ.
When the maximum capillary pressure is equal to or greater than the sum of these pressure drops, the
capillary structure returns an adequate amount of working fluid (priming or repriming of the heat pipe) to
prevent the evaporator wicking structure from drying out. When the sum of all pressure drops exceeds 500 Cu/H2O

W
the maximum capillary pumping pressure, the working fluid is not supplied rapidly enough to the Screen wick
1cm diameter 30.5cm long
Horizontal operation at 100°C
evaporator to compensate for the liquid loss through vaporization, and the wicking structure becomes 400
starved of liquid and dries out (depriming of the heat pipe). This condition, referred to as capillary
limitation, varies according to the wicking structure, working fluid, evaporator heat flux, operating 300

Maximum power
temperature, and body forces.
200 Powder metal wick
Table 5.24: Heat-pipe mechanisms and limitations
100
Heat Transport Limit Description Cause Potential Solution
Heat pipe operating below Increase heat pipe operating 0
Viscous forces prevent vapour
Viscous recommended operating temperature or find
flow in the heat pipe 90 45 0 -45 -90
temperature alternative working fluid
Vapour flow reaches sonic This is typically only a Angle θ from horizontal degrees
velocity when exiting heat pipe Power/temperature problem at start-up. The heat
evaporator resulting in a combination, too much pipe will carry a set power
Sonic Figure 5.40. Dependence of maximum stable heat flow on orientation, specifically capillary limits
constant heat pipe transport power at low operating and the large ∆T will self
power and large temperature temperature correct as the heat pipe versus operating angle. Arrows indicate the direction of the heat flow.
gradients warms up
Sum of gravitational, liquid and
Heat pipe input power Heat pipe effective thermal resistance
vapour flow pressure drops Modify heat pipe wick
exceeds the design heat
Capillary exceed the capillary pumping structure design or reduce A primary heat pipe design consideration is effective pipe thermal resistance or overall heat pipe ∆T at a
transport capacity of the
head of the heat pipe wick
heat pipe
power input given design power. As the heat pipe is a two-phase heat transfer device, a constant effective thermal
structure resistance value cannot be assigned. The effective thermal resistance is not constant but a function of
Heat pipe operating above many variables, such as heat pipe geometry, evaporator length, condenser length, wick structure, and
High velocity vapour flow Increase vapour space
designed power input or at working fluid.
Entrainment/Flooding prevents condensate from diameter or operating
too low an operating
returning to evaporator temperature
temperature
Film boiling in heat pipe
The total thermal resistance of a heat pipe is the sum of the resistances due to
High radial heat flux causes • conduction through the wall,
evaporator typically initiates at 5- Use a wick with a higher heat
film boiling resulting in heat • conduction through the wick,
Boiling 10 W/cm2 for screen wicks and flux capacity or spread out
2 pipe dry-out and large
20-30 W/cm for powder metal
thermal resistances
the heat load • evaporation or boiling, axial vapour flow,
wicks • condensation, and
• conduction losses back through the condenser section wick and wall.
Cu/H2O Cu/H2O
viscosity 1cm diameter 30.5cm long 1cm diameter 30.5cm long Because heat pipes are two-phase heat transfer devices that do not have relatively constant thermal
250 250
limit Horizontal operation Horizontal operation conductivities like solid materials, an effective thermal conductivity is used. The equation used to
W
W

Powder metal wick Screen wick calculate the effective thermal conductivity for a heat pipe is:
sonic P × Leff 1 Leff
200 200 limit Boiling λeff = T = × (5.84)
sonic limit Ax × ∆T Rθ Ax
flood
150 limit
flood Boiling 150 limit where: Leff = ½(Levaporator + Lcondenser) + Ladiabatic
limit limit λ eff = thermal conductivity
Ax = the cross-sectional area of the heat pipe
100 100 PT = power transported by the heat pipe
Capillary limit Capillary limit ∆T = the measured temperature difference across the heat pipe.
Power
Power

50 50 A few rules of thumb can be used for first pass design considerations. A rough guide for a copper/water
heat pipe with a powder metal wick structure is to use 0.2°C/W/cm2 for thermal resistance at the
evaporator and condenser, and 0.02°C/W/cm2 for axial resistance.
0 0
The evaporator and condenser thermal resistances are based on the outer surface area of the heat
10 25 50 75 100 125 150 200 250 300 10 25 50 75 100 125 150 200 250 300
pipe. The axial thermal resistance is based on the cross-sectional area of the vapour space. This design
Temperature °C Temperature °C guide is only useful for powers at or below the design power for the given heat pipe.
These equations for thermal performance are only rule of thumb guidelines and should only be used to
Figure 5.39. Predicted heat pipe limitations, where the capillary limit is usually the limiting factor:
help determine if heat pipes will meet your cooling requirements, not as final design criteria.
(a) powder metal wick and (b) screen wick (viscosity limit on y-axis off-scale).
The groove heat pipe has the lowest capillary limit among the four but functions best under gravity-
Note temperature scale entrainment.
assisted conditions.

Figure 5.39 shows graphs of the axial heat transport limits as a function of operating temperature for Example 5.15: Heat-pipe
typical powder metal and screen wicked heat pipes.
To calculate the effective thermal resistance for a 1.27 cm diameter copper/water heat pipe 30.5 cm
The capillary limit is usually the limiting factor in a heat pipe design, when used in its optimal long with a 1 cm diameter vapour space, the following assumptions are made.
temperature range, and this limit is set by the pumping capacity of the wick structure. As shown in The heat pipe is dissipating 75W with 5cm evaporator and 5cm condenser lengths.
Figure 5.40, the capillary limit is a function of the operating orientation and the type of wick structure.
 205 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 206

Solution Length and diameter affect on heat pipe performance

The rate of vapour travelling from the evaporator to the condenser is governed by the difference in
The evaporator heat flux p equals the power PT divided by the heat input area vapour pressure between them. It is also affected by the diameter and the length of the heat pipe. In a
P PT 75W large diameter heat pipe, the large cross sectional area allows a higher vapour volume to be transported
pevap = T = = = 3.8W/cm2
Aevap π Do Levap π × 1.27cm × 5cm from the evaporator to the condenser, than in a small diameter pipe. The cross sectional area of a heat
pipe is a direct function of both the sonic and entrainment limits of the heat pipe. However, the
PT PT 75W operational temperature of the heat pipe also affects the sonic limit of the heat pipe. Figure 5.41
pcond = = = = 3.8W/cm2
Acond π Do Lcond π × 1.27cm × 5cm compares the heat transport for heat pipes with different diameters and shows that the heat pipes
transport more heat at higher operational temperatures.
The axial heat flux equals the power divided by the cross sectional area of the vapour space The rate at which the working fluid returns from the condenser to the evaporator is governed by the
P PT 75W capillary limit and is a reciprocal function of the heat pipe’s length. A longer heat pipe transports less
p axial = T = = = 95.5W/cm2 heat than shorter heat pipes. In figure 5.41, QmaxLeff (W.m) ,that is, Y-axis, represents the amount of heat
Aaxial ¼π Di2 ¼π × 1.02
a pipe will carry per meter length. Therefore, if the pipe is half a meter, it can carry twice the wattage a
The temperature gradient equals the heat flux times the thermal resistance. metre long heat pipe would carry.

Orientation affect on heat pipe performance

∆T = pevap × Revap + p axial × Raxial + pcond × Rcond A wick structure with a higher capillary limit can transport more working fluid from the condenser to the
∆T = 3.8 W/cm ×0.2°C/W/cm + 95.5 W/cm ×0.02°C/W/cm + 3.8 W/cm ×0.2°C/W/cm
2 2 2 2 2 2
evaporator against gravity. But as mentioned, the groove heat pipe, with the lowest capillary limit, works
best under gravity-assisted conditions where the evaporator is located below the condenser. Figure 5.41
∆T = 3.4°C
shows the effect of gravity on groove wick heat pipes.
♣
Performance affect of heat pipe flattening or bending
If a heat pipe is flattened or bent, the sonic and entrainment limits will be reduced in relation to the
Maximum heat transport QmaxLeff
Rθ

5 horizontal length 6 horizontal orientation flattened thickness, the number of bends, and the angle of each bend. Therefore, any flattening or
Cu/H2O, 6mm diameter Cu/H2O 6mm
heat transport 10W 5 groove wick
bending to a heat pipe will reduce the amount of heat that can be transported. Figure 5.41 shows the
4
groove
effect of flattening on a heat pipe.
Thermal resistance

4 High temperature
K/W

3
Heat Pipes
3
Wm

Fibre + spira
Intermediate temperature
2
mesh 2 4mm Heat Pipes
powder metal
1 Ag
1 3mm

0 Li
0
100 125 150 175 200 225 250 35 40 45 50 55 60 65 Low temperature Na
Heat Pipes
length L mm Heat pipe operational temperature °C Hg
K
P
Rθ

4
5 vertical length 10 vertical length orientation Cs
Cu/H2O

y
Cu/H2O, 6mm diameter

ilit
heat transport 10W Fibre + spira groove wick
22.2mm H2O

ab
4
Maximum heat transport

ap
15.9mm
NH3
Thermal resistance

rc
10 12.7mm

sfe
mesh
K/W

3 9.5mm (CH3)CH
Cryogenic Heat Pipes

an
powder metal
W

6.4mm

Tr
groove (CH3)CO
2

at
2
10

He
C6H6

g
F-11

sin
1 3mm

ea
F-21

cr
in
0 10
1
CH4
100 125 150 175 200 225 250 0 25 40 60 80 100 120 140
length L mm Heat pipe operational temperature °C O2
N2
QmaxLeff

QmaxLeff

80 4 Ne
Heat out 250mm length D4 round 0°C
Cu/H2O groove wick H2
60 θ 9.5mm Cu/H2O 3 working temp 50°C
groove wick
Maximum heat transport

Maximum heat transport

Heat in 10 50 100 500 1000 5000

T3 flat
Wm

Wm

40 2 Temperature T K
Figure 5.42. Operating temperature ranges for various heat pipe fluids.
6.4mm Cu/H2O
20 groove wick 1
T2 flat
Operating temperature range
0 0 The operating temperature ranges of heat pipes are referred to as Cryogenic (0 to 100K), ‘Low
0 10 20 30 40 50 60 70 80 90 -150 -10 -5 0 5 10 15 20 Temperature’ (100 to 250K), ‘Intermediate Temperature’ (250 to 600K), and ‘High Temperature” (600 to
inclination θ degrees inclination θ degrees 3000K). Working fluids are elemental gases in the cryogenic range, mainly polar molecules or
halocarbons in the low temperature range, simple organic molecules in the intermediate temperature
Figure 5.41. Heat pipe parameter performance characteristics. range, and liquid metals in the high temperature range.
207 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 208

The approximate useful heat pipe fluid range of some working fluids is indicated in Figure 5.42. Also

W / m2
107 iodine 1012
indicated are the limits of the four defined temperature regimes. The range limits are approximate since TiCl4 toluene
phenol
water
Dowtherm A
some of the fluids overlap into an adjacent temperature range.

Pa
6 water
10 naphthalene
mercury
cesium
Low temperature heat pipe technology

FoM
1011
105
Low order hydrocarbons are common low temperature heat pipe fluids, as specifically shown for ethane sulphur phenol

Vapour pressure,
(in stainless steel) and ammonia (in an aluminium alloy) based constant conductance heat pipes in 104 Dowtherm A

Liquid transport factor,

cesium 1010
figure 5.43. TiCl4 iodine
103

102 Sulphur/
109 iodine

W
200 naphthalene
101
toluene
100 108
150

P
ammonia 400 500 600 700 800 400 500 600 700 800
Temperature, K Temperature, K
Maximum power 100
ethane Hg

W / m2
50 107 1012 Na
SiCl4 water TiCl4

Pa
TiBr4 SiI4
6
10 water
0

FoM
-120 -80 -40 0 40 105 TiI4
1011 NH3

Vapour pressure,
Operating temperature °C 104

Liquid transport factor,

cesium SnBr3
Figure 5.43. Low temperature constant conductance heat pipe performance for Al/NH3 and Al/C2H6 103 N2
with slight adverse elevation, 12mm internal diameter, 60cm adiabatic section and 15cm 1010 iodine
102 SnCl4
evaporator/condenser. GaBr3 TiCl4

101 AlBr3 BiCl3

Intermediate temperature heat pipe technology 100 109

The intermediate temperature range extends from 450K to 750K. The alkali metals, such as caesium, 400 500 600 700 800 200 300 400 500 600 700 800
potassium, and sodium, are suitable working fluids at temperatures above this range. In the intermediate Temperature, K Temperature, K
temperature range, the alkali metal vapour density is so low that the vapour sonic velocity limits the heat
transfer, and the heat pipe vapour space becomes too large to be practical for alkali metals. Figure 5.44. Vapour pressure and figure of merit for intermediate temperature heat-pipe fluids,
Water is commonly used at temperatures up to about 425K. Higher temperature water heat pipes can be at a constant latent heat and normal boiling temperature.
used with titanium or Monel envelopes at temperatures up to 500K. Their effectiveness starts to fall off
after 500K, due to the decrease in the surface tension of water. Potential intermediate temperature
working fluids including Sulphur/Iodine mixtures, Iodine, Naphthalene, Phenol, Toluene, Mercury, and There are many candidate fluids in the intermediate temperature range. While some of the fluids have
several halides. Sulphur has a temperature dependent polymerization property at 475K, which increases sufficient physical property data to allow their use, none of the fluids has adequate life test data at
its liquid viscosity to approximately three orders of magnitude higher than the maximum level for appropriate conditions.
effective heat pipe operation. The addition of 3-10% of iodine reduces the viscosity of sulphur to a level
sufficient for effective heat pipe operation. A disadvantage of iodine is its low liquid thermal conductivity.
SbBr3 Cesium
Another set of potential working fluids is the halide salts of titanium, aluminium, boron, phosphorus, and Iodine Quad artery dual artery

W
silicon. Salts such as TiCℓ2F2 working fluids are polar, which increases the latent heat and the liquid Quad artery
2000
transport factor.
Vapour Pressure and Merit Number are two parameters used to screen potential working fluids. Vapour

P
pressures for some intermediate temperature working fluids are shown in the parts of Figure 5.44. Note 1500
that the vapour pressure for water is too high, and the vapour pressure for caesium is too low in this

Maximum power
temperature range, so a vapour pressure between these two extremes is desirable. Most of the fluids
1000
discussed have a suitable vapour pressure. Water
hybrid
The Merit Number (liquid transport factor) Mℓ is a means of ranking heat pipe fluids, with a higher merit
500
number being more desirable: BiCl3
ρ Aσλ Quad artery
MA = (5.85) 0
ν 400 500 600 700 800
where: Mℓ = Merit Number, W/m2
Operating temperature K
ρℓ = liquid density, kg/m3
σ = surface tension, N/m Figure 5.45. Theoretical heat pipe powers for different fluids.
λ = latent heat, J/kg Heat pipe dimensions: 2.54mm diameter, 220mm length.
ν = liquid viscosity, Pa
Figure 5.44 shows the Merit Number as a function of temperature for a number of fluids. Whilst water
and caesium have good merit numbers, their vapour pressures eliminate them from contention.
Evaporator cooling
Figure 5.45 compares the theoretical heat transfer capability (power) for heat pipes with five working
fluids: water, iodine, BiCl3, SbBr3, and caesium. Water is the best fluid at the low temperature end and Heat removal at the heat pipe remote evaporator can be affected with conventional aluminium finned
caesium at the upper end. Iodine and SbBr3 offer good performance in the mid-region. However, Iodine heatsinks, with fan assistance if necessary. Any alternative cooling method in this chapter may be
has two potential problems: low liquid thermal conductivity and high corrosiveness. applicable.
209 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 210

e se e at Cold plates are used in closed systems where pumped coolants continuously cycle, conveying excess
vic ca ratur er te he ng er t
de ction pe ad pla a ien heat away from the devices being cooled. This heat is then dumped into the ambient air via a radiator
u n em sp
re ld xch am
b
j t co e heat exchanger or recirculating chiller, or to facility water via a liquid-to-liquid heat exchanger or liquid-
cooled recirculating chiller, as shown in figure 5.46.
Depending on the specific cold-plate design, the components may be mounted on one or both sides of
the plate, which is usually aluminium, as shown in figure 5.47.
H 2O

Si
In transferring the conducted (dissipated) heat PD to the coolant, a cold plate in steady-state must satisfy
Qflow = PD

H 2O
Al mc p (T out − T in ) = PD = hA (T s − T m )

exchanger

ambient
heat

air
where T m = ½ (T out + T in )

H 2O
Al

The heat transfer coefficient, from the heated surface, area A, at temperature Ts, is
pump fan ∆T Nu × k
h= = (5.86)
PD DH
Cu

where the Nusselt number is Nu = 4.44 for fully developed flow conditions.
Figure 5.46. Improved cooling with compact indirect cold-plate spreader water-cooling.
fL ρ V 2
pressure drop : ∆P = ½ (Pa)
DH
pump power : Pw = ∆P × G (W)
5.16.2 Cold plates – indirect cooling (5.87)
m c ∆T
heat exchanger length : L = − f p A n out ( m)
Liquid-cooled cold plates perform a function analogous to air-cooled heat sinks by providing an effective π hDH ∆T in
means to transfer heat from a component to a liquid coolant. Unlike heat pipes, they are active devices −
1

outlet temperature : T out = T s − (T s − T in ) e (K )

mf c p Rt
in that liquid is usually forced through them by the action of a mechanical fluid pump. Vacuum-brazed
fin-stock cold-plates and copper-based superalloy structures are used and a liquid-cooled microchannel
where L = length of the cold plate
heat sink can remove 790W/cm2 with a temperature increase of 71°C for a 600ml/min flow rate with a
mf = mass flow rate
pressure drop of 207kPa.
f = friction factor
Liquid cooling can reduce effective thermal resistance to as low as 0.01K/W and may provide a much cp = specific heat
more compact clamp cold-plate spreader heat-sink arrangement, as shown in figure 5.46. Equipment DH = diameter (hydraulic)
compact form factors require efficient yet cost-effective modes of removing excess heat. Rt = thermal resistance of one channel
G = volumetric flow rate, m3/s
Liquid cooling via cold-plate technology combines a high capacity for heat rejection with an ability to See section 5.18 on microchannels.
move heat remote from the power electronics to the ambient room air or facility water.
A number of cold plate technologies are available. Generally, the cold plate's cost increases with
improving performance. The technologies to follow are listed in order of what is typically increasing cold
plate efficiency and cost.
Single piece, • Tubed cold plates
bent copper tube Tubed cold plates are the most common designs, giving dependable high performance at a
comparatively low cost. A typical device is a flat metal plate, 15 to 30 mm long, with a series of
channels on one or both sides. Into these channels is secured a length of serpentine metal
Fluid inlet/outlet
Stainless steel or tubing through which the liquid coolant flows. Fittings at the inlet and outlet of the coolant tubing
copper tubing for connect to the user's coolant source.
fluid compatibility Usually aluminium, a tubed cold plate may either be cut to size (with channels for coolant tubing
later machined into it), or extruded to size (with the channels formed at the same time the plate
Tin is extruded). Some manufacturers use a channel geometry that holds the coolant tubing in
place, ensuring good metal-to-metal contact and optimizing thermal transfer. Others secure the
tubing to the channels with a performance-limiting thermal epoxy.
Tubing material is predominantly either 5 mm or 10 mm out-side diameter copper or stainless
steel. The larger diameter offers a lower pressure-drop. Smaller tubing increases fluid
Tout turbulence and can be formed into tighter radii for tighter tube packing and more-compact lower-
profile cold plates. If the coolant is tap water, copper will suffice. If the coolant is deionised water
or another corrosive fluid, stainless steel is preferable because it is non-reactive.
Larger diameter for
low fluid pressure drop
• Gun-drilled cold plates
A gun-drilled cold plate is usually fabricated by drilling a series of holes through the length of an
aluminium plate, inserting copper or stainless-steel tubing, and then expanding the tubing to
Machinable ensure a secure metal-to-metal contact with efficient thermal transfer properties. If aluminium
mounting surface corrosion is not a risk, it is possible to flow coolant with an ethylene glycol additive directly
through the gun-drilled holes without inserting the tubing.
For the return fluid path, the holes are drilled perpendicular to the main fluid path and then
partially plugged to create a continuous coolant path. Before the development of two-sided
Metal to metal pressed interface for tubed cold plates, gun drilling was the preferred approach for two-sided applications. Tubed cold
optimal thermal performance and reliability plates are preferred because the inserted copper tubing eliminates the concern of aluminium
corrosion. One additional benefit of gun-drilled cold plates is that they can have tighter
Figure 5.47. Compact indirect cold-plate spreader water-cooler. tolerances than tubed cold plates, specifically in meeting flatness requirements.
 211 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 212

• Vacuum-brazed inner finned cold plates Heat transfer fluids for cold plate liquid cooling
Vacuum-brazed aluminium cold plates are reserved predominantly for high-performance custom One of the most important factors when choosing a liquid cooling technology for an application is the
designs that provide low thermal resistance, and superior leak-free reliability. They afford the compatibility of the heat transfer fluid with the wetted surfaces of the cooling components or system and
greatest flexibility in specifying thermal resistance, thermal flow, pressure drop, fluid path, size, the application. Heat transfer fluid compatibility is critical in ensuring long-term system reliability. Some
shape, material hardness, surface geometry, and dual-sided component mounting. other requirements for a heat transfer fluid may include high thermal conductivity and specific heat, low
To build these cold plates, high-performance corrugated aluminium-fin material is brazed into viscosity, low freezing point, high flash point, low corrosivity, low toxicity, and thermal stability. Based on
the liquid cavity below the mounting surface, which is also made of aluminium. The internal fins these criteria, the most commonly used coolants for indirect liquid cooling applications are:
add heat transfer surface area and create turbulence in the coolant to minimize the fluid • water
boundary layer and reduce thermal resistance. • deionised water
Depending on requirements, these cold plates can be surface machined to optimize heat-load • inhibited glycol and water solutions
and/or cold-plate contact. The size range of vacuum-brazed cold plates is from about 20 cm2 to • dielectric Fluids
about 4000 cm2.
• Extruded cold plates By selecting a compatible pairing of heat transfer fluid and wetted materials, the risk of corrosion as well
Extruded cold plates represent the highest performance designs with the best thermal as the thermal performance are optimised. Copper is compatible with water and glycol/water solutions
performance characteristics. A dimensionally compact structure (typically 5 cm wide x 4 mm. and aluminium is compatible with glycol/water solutions, dielectric fluids, and oils. When using deionised
thick), this design features an aluminium or copper extrusion with many internal, discrete, water or other corrosive fluids, however, stainless steel is used since it is more corrosion resistant than
parallel microchannels. other metals. (See Tables 5.21 and 5.25.) Most cooling systems are compatible with water or
The open ends of the extrusion are welded to connector tubes. The microchannel design glycol/water solutions but require special plumbing for compatibility with deionised water or a dielectric
provides a large internal surface area and a thin mounting surface that minimizes thermal fluid such as polyalphaolefin, PAO.
resistance (see section 5.18).
Table 5.25: Cold plate metals and compatible fluids
The channels create turbulence, which minimizes the fluid boundary layer and reduces thermal
resistance. This approach also yields excellent thermal uniformity because coolant flows below Materials and Transfer Glycols Deionised Dielectric Fluids Polyalphaolefin,
the entire cold-plate surface. Depending on the direction the header manifolds are mounted, the Water Oil
Fluid Compatibility EGW Water (Fluorinert) (PAO)
extruded cold plates may have a ‘U’ or ‘Z’ fluid inlet/outlet configuration. Multiple extruded cold Copper tubing X X
plates can be assembled into a single unit or pressed into an aluminium plate.
Stainless steel tubing X X X
Cold-plate performance comparison Aluminium flat tube or plate fin X X X X
Cold-plate performance is normally expressed as thermal resistance, in °C/W. The lower the thermal Copper flat tubing, copper etched
X X X X X
resistance, the better the cold plate performance and the cooler the surface. Figure 5.48 compares or copper brazed
normalized results with respect to area, allowing the various cold-plate technologies to be compared Nickel brazed X
independent of individual part geometries.

Fluid pressure drop increases exponentially with fluid flow rate and viscosity. Selecting a cold plate
To determine the maximum allowable thermal resistance of the cold plate hence to specify a cold plate,
it is necessary to know the cooling fluid flow rate, the fluid inlet temperature, the heat load of the devices
attached to the cold plate, and the maximum desired cold plate surface temperature, Tmax.
10 First, calculate the maximum temperature of the fluid when it leaves the cold plate, Tout. This is important
because if Tout is greater than Tmax, there is no solution to the problem.
2 pass, 9.5mm Cu tube,
Pd Pd
[Link] / W

Plate (thick) side single side, T out = T in + = T in + (5.88)

2

13mm thick Aℓ plate m ×cp ρ Av × c p

5 Tube (thin) side
where Tout = temperature of fluid leaving cold plate, K
Tin = inlet temperature of fluid, K
Pd = heat load of power devices, W
ρℓ = density of the fluid, kg/m3
v = cooling fluid flow rate, m3/s
Vacuum 6.4mm tube (thick) 4 pass, 9.5mm Cu tube,
single side cp = specific heat of the cooling fluid, J/kgK
brazed both sides
Normalise thermal resistance

8mm thick 13mm thick Aℓ plate Alternatively, heat capacity graphs describe the change in temperature, ∆T, that occurs along the fluid
6.4mm tube (thin) both sides path. To find Tout, add ∆T to the inlet temperature, Tin.
1 8mm thick Assuming Tout is less than Tmax, the next step is to determine the required normalized thermal resistance
Rθ for the cold plate using:
Aℓ, extruded µ-channel
A
(flat tube) 3.3mm thick Fin technologies Rθ = (T max − T out ) (5.89)
Pd
0.5
where Rθ = thermal resistance, K/W per unit area
Tmax = maximum desired cold plate surface temperature, K
Cu, extruded µ-channel A = area being cooled, m2
(flat tube) 3mm thick
Any cold plate technology that provides a normalized thermal resistance less than or equal to the
0.2 calculated value will be a suitable solution.
0 1 2 3 4 5 6 7
The graph in figure 5.48 compares normalized thermal resistance expressed in °C.cm2 / W of cold plate
Water flow rate LPM technologies. Normalization of the data allows thermal performance comparison across technologies
without regard to physical size. Technologies with lower thermal resistance values offer the highest
Fig. 5.48. Performance comparison of various types of cold plates. performance in the application.
213 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 214

Example 5.16: Cold plate design 5.17 Direct liquid cooling

A cold plate is used to cool a 5cm x 10cm copper base IGBT that generates 500W of heat. It is cooled 5.17.1 Immersion cooling – direct cooling
with 20°C water at a 2 lpm flow rate. The surface of the cold plate must not exceed 55°C. Calculate the
plate outlet header temperature, thence necessary plate thermal resistance. Direct liquid or immersion cooling is a well-established method for accommodating high heat flux. With
Use cp = 4,184 J/kgK and ρℓ = 998 kg/m3 natural convection two-phase flow, generally termed nucleate pool boiling, the critical heat flux using FC-
72 is in the range of 5 to 20W/cm2. However, much higher heat fluxes up to 100W/cm2 can be
Solution accommodated with surface enhancement of the heat source. With a device submerged in a pool of
dielectric liquid (electrical non-conducting), the heat dissipated in the device produces vapour bubbles
Given: Tin: 20°C, Tmax: 55°C, P: 500 W, Area: 5x10=50 cm2 that are driven by buoyancy forces into the upper region of the container, where the vapour condenses
and drips back into the liquid pool. A disadvantage of this technique is the need for a liquid compatible
First calculate Tout. Using the heat capacity equation (5.88), the temperature change for 500W at a 2 lpm with the device. Often, water cannot be used because of its chemical and electrical characteristics.
flow rate is less than 4°C. That is Direct liquid cooling may also be termed direct liquid immersion cooling, since there are no physical
P walls separating devices and its substrate surface from the liquid coolant. This form of cooling offers the
T out = T in +
ρv × c p opportunity to remove heat directly from the package with no intervening thermal conduction resistance,
500W other than that between the device heat sources and the package surfaces in contact with the liquid.
= 20°C + = 20°C + 3.6°C = 23.6°C Direct liquid immersion cooling offers a high heat transfer coefficient which reduces the temperature rise
2
998kg/m3 ×
l/s × 10-3m3 /l × 4,184J/kgK of the package surface above the liquid coolant temperature. As shown in Figure 5.50, the relative
60 magnitude of a heat transfer coefficient is affected by both the coolant and the mode of convective heat
Tout is less than Tmax so the required thermal resistance is determined by equation (5.89): transfer (that is, natural convection, forced convection, or boiling). Water is the most effective coolant
A and the boiling mode offers the highest heat transfer coefficient. Direct liquid immersion cooling also
Rθ = (T max − T out )
P offers greater uniformity of package temperatures than is provided by air-cooling.
50cm2
= (55°C - 23.6°C ) ×
500W Air 1 to 3 atm
= 3.14°C.cm2 /W at 2 lpm Air compressed

This point is plotted on the normalized thermal resistance graph in figure 5.48 where any technology Fluoro vapour
below this point will meet the thermal requirement. But because the cooling fluid is water, aluminium Silicone oil
tube or plate fin, or nickel-brazed technologies should not be employed. natural
♣ Transformer oil convection
Fluoro liquids
Air 1 to 3 atm
Al extruded µ-channel
0.08 flat tube, 3.3 mm thick 0.4 Fluoro vapour
Transformer oil forced
convection
°C/W

Fluoro liquids
Rθ = 0.063 °C/W single-phase
bar

0.06 0.3 Water

30% GaInSn
Rθ

EGW
∆p

Fluoro liquids pool

Thermal resistance

Water boiling
0.04 0.2
pressure drop

Fluorochemical spray
Water cooling
water
0.02 0.1

1 10 100 1000 10000 100000

Heat transfer coefficient h W/m2K

0 1 2 3 4 Figure 5.50. Relative magnitude of heat transfer coefficients

for various coolants and modes of convection. (see figure 5.19)
flow rate lpm
Figure 5.49. Effects of flow rate and coolant variation on thermal resistance.
Coolant considerations
Immersion cooling involves more than just the selection of a direct immersion liquid based on heat
transfer characteristics alone. Chemical compatibility of the coolant with the mounting and packaging
In example 5.16 the cold plate thermal resistance requirement is 3.14 Kcm2/W / 50cm2 = 0.063K/W. This materials exposed to the liquid are the primary consideration.
requirement can be met by an aluminium µ-channel cold plate with the characteristics shown in figure There may be several coolants which can provide adequate cooling, but a limited few will be chemically
5.49. This cold plate readily fulfils the thermal resistance requirements with de-ionised water. The compatible. Water is an example of a liquid which has desirable heat transfer characteristics, high
aggressive nature of deionised water can be avoided by the using 30% ethylene glycol - water but at the thermal conductivity for example, but is generally unsuitable for direct immersion cooling because of
expense of a decreased thermal conductivity, as seen in Table 5.21. From figure 5.49, the necessary other chemical characteristics, such dielectric constant. The fluorocarbon liquids listed in Table 5.26 are
flow rate is 1.5 lpm, with a cold plate input to output fluid pressure drop of 0.08 bar. (0.6 lpm and 0.024 generally considered the most suitable liquids for direct immersion cooling, despite their poorer thermo-
bar if de-ionised water could be used). physical properties.
215 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 216

As shown in Table 5.26, the thermal conductivity, specific heat, and heat of vaporization of fluorocarbon C
coolants are lower than water. These coolants are clear, colourless per-fluorinated liquids with a Nucleate
10 E
boiling

2
relatively high density and low viscosity. They also exhibit a high dielectric strength and a high volume

W / cm
D
resistivity. The liquid therefore serves to both cool and insulate the components. The boiling points for
available ‘fluorinert’ liquids range from 30 to 253°C. F
G
All high power and high voltage power electronics can be immersed in the fluorinert. As a component Film
1 velocity boiling
heats up, the fluorinert in contact with it vaporizes and it is this liquid to vapour phase transition which Forced
effectively removes the excess heat from the components (heat of vaporization). The fluorinert vapour is convection
cooled by a heat exchanger located in the area above the fluid. This cooling technique allows high- Thermal hysteresis
temperature overshoot
power electronics to operate continuously and reliably in a small volume.

Heat flux
Why use a dielectric liquid, specifically FC-72 instead of water? 0.1 Natural
• Same weight of water will extract more than three times the heat for the same B convection
temperature rise (higher specific heat). Typical forced air
• Water is eleven times more conductive. A convection
• Same weight of water upon evaporating will extract 27 times more heat (higher heat of 0.01
vaporization). 0.1 1 10 100 1000
• However, FC-72 has a dielectric constant that is 2% that of water, that is, much less
Temperature difference ∆T = Twall - Tsat °C
unwanted dielectric current will flow.

Table 5.26: Comparison of thermo-physical properties of some fluorocarbon coolants and water Figure 5.51: Typical heat transfer regimes for immersion cooling with a fluorocarbon liquid.

Property units FC-87 FC-72 FC-77 H2O Boiling: Boiling is a complex convective heat transfer process depending upon liquid-to-vapour phase
Boiling Point @ 1 Atm, °C 30 56 97 100 change with the formation of vapour bubbles at the heated surface. It is commonly characterized as
3 3 either pool boiling (occurring in a stagnant liquid) or flow boiling. The pool boiling heat transfer rate Q
Density ρℓ ×10 kg/m 1.633 1.680 1.780 0.997
usually follows a relationship of the form
3
Specific Heat ×10 W.s/kg.K 1.088 1.088 1.172 4.179
Thermal Conductivity λ W/m.K 0.0551 0.0545 0.057 0.613
Q = C sf' × A × (Twall - T sat )n (5.90)
-4
Dynamic Viscosity ν ×10 kg/m.s 4.20 4.50 4.50 8.55 where C sf is a constant depending on each fluid-surface combination,
'

Heat of Vaporization L ×10

4
W.s/kg 8.79 8.79 8.37 243.8 A is the heat transfer surface area,
Twall is the temperature of the heated surface,
Surface Tension ×10-3 N/m 8.90 8.50 8.00 58.9
Tsat is the saturation temperature (that is, boiling point) of the liquid, and
Thermal Coefficient of Expansion ×10-3 K-1 1.60 1.60 1.40 0.20 n is an exponent, typically about 3.
Dielectric Constant =1 for a vacuum 1.71 1.72 1.75 78.0
The boiling curve for a particular surface and fluid of interest (for example, silicon and FC-72) is usually
obtained experimentally. An example of a boiling curve depicting the cooling path from natural
convection to film boiling is shown in Figure 5.51. If die power is gradually increased in small steps,
These liquids should not be confused with the ‘Freon’ coolants which are chlorofluorocarbons (CFCs). cooling occurs first by natural convection (A - B). Eventually a power level is reached at which sufficient
Although some of the ‘Freons’ exhibit similar cooling characteristics, concern over their environmental superheat is available to initiate the growth of vapour bubbles on the surface and boiling starts (B). As
effect preclude their use. power is increased, more nucleation sites become active and the frequency of bubble departure
Modes of heat transfer increases. The region between B and C is termed the nucleate boiling regime. Vigorous agitation of the
The convective heat transfer processes upon which liquid immersion cooling depends may be classified hot boundary along the heated surface, and gross fluid circulation caused by the motion of the vapour
as natural convection, forced convection, or boiling modes. The relative magnitude of heat fluxes which bubbles, provide the ability to accommodate substantial increases in heat flux with minimal increases in
can be accommodated by each mode is shown in Figure 5.51, as a function of ‘wall superheat’ or surface temperature. As power is increased to point C, the critical heat flux condition is reached. So
surface-to-liquid temperature difference for a typical fluorocarbon coolant. many bubbles are generated at this point that they begin to form a vapour blanket inhibiting fresh liquid
from reaching the surface. Further increases in power will result in a transition to film boiling (D - E). In
Natural convection: As in the case of air-cooling, natural convection is a heat transfer process in which this regime, heat transfer from the surface to the liquid is dependent on thermal conduction through the
mixing and fluid motion is induced by coolant density differences caused by the heat transferred to the vapour, which is poor. In most power electronic cooling applications, transition to film boiling will result in
coolant. As depicted in Figure 5.51, this mode of heat transfer offers the lowest heat flux or cooling failure due to high junction temperatures. To take advantage of boiling to cool electronic devices, it is
capability for a given ‘wall superheat’. Nonetheless, the heat transfer rates attainable with liquid natural desirable to operate in the nucleate boiling regime (B - C).
convection can exceed those attainable with forced convection of air. Natural convection would typically A problem often associated with pool boiling of fluorocarbon liquids is that of temperature overshoot.
be employed within a closed container to transfer heat from die or modules to liquid, and then from the This behaviour is characterized by a delay in the inception of nucleate boiling (that is, beyond point B),
liquid to the walls of the container. Heat could then be transferred from the walls to outside air ambient such that the heated surface continues to be cooled by natural convection; with increased surface
by natural or forced convection. temperatures unless a sufficient superheat is reached for boiling to occur. This behaviour is a result of
the good wetting characteristics of the fluorocarbon liquids and the smooth nature of silicon die and their
Forced convection: Higher heat transfer rates may be attained by utilizing a pump to provide forced metallisation. There is minimal temperature overshoot associated with flow boiling cooling applications.
circulation of the liquid coolant over the die or module surfaces. This process is termed forced The typical critical heat fluxes encountered in saturated (that is, liquid temperature saturation
convection; and as with air-cooling, the allowable heat flux for a given surface-to-liquid temperature temperature) pool boiling of fluorocarbon liquids range from about 10 to 15 W/cm2, depending upon the
difference can be increased by increasing the velocity of the liquid over the heated surface. Depending nature of the surface (that is, material, finish, geometry). The allowable critical heat flux may be
upon the surface geometry and the nature of the flow (that is, laminar or turbulent), the heat transfer extended by subcooling the liquid below its saturation temperature. For example, the critical heat flux
coefficient will be proportional to the velocity to a power between 0.5 and 0.8. The penalty for increasing can be increased to as much as 25 W/cm2 by reducing the liquid temperature to -25°C.
cooling performance in this way, is a higher-pressure drop. This can mean a larger pump and higher Higher critical heat fluxes may be achieved using flow boiling. For example, heat fluxes from 25 to over
system operating pressures. Although forced convection requires the use of a pump and the associated 30 W/cm2 have been reported for liquid velocities of 0.5 to 2.5 m/s over the heated surface. Heat fluxes
piping, it allows removal of heat from high power modules in a confined space; and then transports the in excess of 100 W/cm2 have been obtained with a FC-72 liquid jet impinging upon a 6.5 mm x 6.5 mm
heat via the liquid coolant to a remote heat exchanger to reject the heat to air or water. chip at a flow rate of 2.2 cm3/s.
217 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 218

Other considerations
Since fluorocarbon liquids are expensive, they should only be used in closed systems. Whether the
application is in a self-contained module or a forced flow scheme, the seal materials must be compatible
with the liquid. If boiling is to take place, then the design must incorporate a means to condense the Jet
resulting vapours. A finned surface may be designed for this purpose, or a remote finned condenser stream
surface cooled by air or water might be used. In flow systems, care must be taken in selecting a pump.
The relatively high vapour pressure of the low boiling point fluorocarbons generally require that a higher
suction head be provided to prevent cavitation in the pump. In forced circulating liquid systems, it may
be desirable to add a particulate and a chemical filter to ensure the long-term purity of the coolant. Bubbles

5.17.2 Liquid jet impingement – direct cooling

Jet impingement cooling, as shown in figure 5.52, is similar to spray cooling but is performed with a
lower nozzle pressure drop and a higher fluid flow rate. This lower pressure requirement, in conjunction
with a high flow volume, reduces nozzle clogging and reduces noise levels. Cooling of 90W/cm2 with a
100°C temperature rise using a flow rate of only 8ml/min is possible. A closed loop impingement jet
gives cooling of 180W/cm2 using water and a flow of 0.3 l/min at 300kPa. The micropump used 7W to Heat area Heat area
drive the fluid flow. h
(a) Heat transfer coefficient (b)

Jet
Spray

(c)
Pressurised
cooling water x
o

Figure 5.53. Illustration of:

Returned
(a) spray cooling; (b) jet impingement cooling; and (c) resultant heat transfer coefficient, h.
heated water

5.18 Microchannels and minichannels

The concept of microchannels applied to the thermal management of high heat-flux power electronics is
simple. Because heat-transfer coefficients generally increase with decreasing size, the passage size
(microchannels) should be made as small as possible. This results in a dense package with higher heat
transfer and a larger surface area-to-volume ratio than a conventional cooling device.
Figure 5.52. Available multiple jet impingement liquid cooling. The term 'micro' is applied to devices having hydraulic diameters of ten to several hundred micro-
metres, while 'mini' refers to diameters of the order of one to a few millimetres. A low flow rate within
micro-channels produces laminar flow resulting in a heat transfer coefficient inversely proportional to the
5.17.3 Spray Cooling – direct cooling hydraulic diameter. In other words, the smaller the channel, the higher the heat transfer coefficient.
Unfortunately, the pressure drop increases with the inverse squared of the channel width, whilst
Spray cooling breaks up the liquid into fine droplets that impinge individually on the heated wall surface. maintaining the mass flow constant. Stacking of microchannel layers is used to decrease the pressure
Cooling of the surface is achieved through a combination of thermal conduction through the liquid in drop, but flow non-uniformity across the channel remains a problem, producing non-uniform cooling.
contact with the surface and evaporation at the liquid-vapour interface (latent heat of vaporization). The A 50µm wide, 300µm deep, 1cm long microchannel passage, experiences a 30-psi drop with a 0.66 lpm
droplet impingement both enhances the spatial uniformity of heat removal and delays the liquid water flow, and can dissipate 790 W/cm2 with a 71°C temperature rise. With 5 to100um channel sizes,
separation on the wall during vigorous boiling. The hot vapour is recovered by removing the waste heat the heat transfer coefficient can reach 80kW/m2K.
to the ambient in a heat exchanger where the vapour condenses back to a liquid. The fluid is continually
recycled within a closed system. Using vapour for heat transport eliminates all resistance between the Integrated microchannel heatsink
heat source and the ultimate heat destination. Figure 5.54 displays the fractions of the thermal resistivity attributed to different individual layers for a
Spray cooling technology enables all surfaces exposed to the liquid-vapour environment to remain close typical power module. In the case of microchannel cooling, the largest contributor to the thermal
to the saturation temperature of the fluid. The result is an isothermal environment around the power resistivity is the thermal grease layer between the base-plate and the heatsink.
devices which effectively reduces hot spots and thermal cycling – the primary cause of power device Because of the increased importance of the conductive resistance of the stack, a better solution would
failure. Spraying reduces the flow rate requirements but requires a high nozzle pressure drop than jet be to eliminate some of the layers from the structure. An integrated heatsink with a series of
impingement. Spray pattern types include an array of swirling turbulent hollow-cone sprays, with a wide microchannels fabricated directly into the bottom copper layer of the active metal braze (AMB) substrate
range of droplet sizes. Drawbacks include likelihood of nozzle clogging, repeatability of impact patterns, is shown schematically in figure 5.55, and has advantages in reducing both the convective and
high sensitivity to nozzle to surface distance, and the need for filters and pumps. conductive resistances of the module.
Spray cooling and jet impingement (as shown in Figure 5.53) are often considered competing options for
electronic cooling. Spray evaporative cooling with a Fluorinert coolant can maintain junction
temperatures between 70 and 85°C for heat fluxes from 15 to 55 W/cm2. Spray cooling improves
thermal management, increases system-packaging density, and reduces weight. Die-level spray cooling
allows a maximum heat flux of over 160 W/cm2.
219 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 220

L
∆P = ½ρ v 2f (5.92)

Thermal resistance of layer

Thermal resistance of layer

heatsink heatsink DH
5 5
grease
with
4 4 64

K cm / W

K cm / W
f =
Cu Re
2

2
3 Cu AMB AMB 3 AMB AMB
base
base Cu Cu Cu Cu where ρℓ is the fluid density and
2 2 f is the friction factor (loss coefficient), in laminar, fully developed flow in a circular passage
grease ceramic die ceramic die
1 solder solder 1 solder solder
Re is the Reynolds number
0 0 ρ A × v × DH
Re = (5.93)
layer layer υ
(a) (b) where v is the fluid velocity (volumetric flow rate) and
υ is the fluid kinematic viscosity.
Figure 5.54. Thermal resistivities of a typical power module with:
(a) a conventional heatsink and (b) a microchannel heatsink, indicate that layers other than the These basic expressions for Rθ and ∆P are used to select the optimal channel sizing. The passage
heatsink can dominate the performance of microchannel designs. dimensions are chosen to minimize the thermal resistivity Rθ subject to pumping constraints on
maximum pressure loss and flow rate. Typical performance characteristic are as shown in figure 5.49.
By using substrate level microchannels, the convective thermal resistivity is reduced dramatically. In
The primary variables to be optimized are the channel width and pitch. These values vary beneath a
addition, this stack removes the base-plate solder, the copper base-plate and, most importantly, the
heat source of 2×2 cm size between 10 to 200 cooling passages and a ratio of wall thickness to channel
thermal grease from the conductive path. As seen in figure 5.54, this results in the elimination of the two
width of 0.1 to 2. These dimensions result in a range of flow conditions, including laminar, turbulent,
largest resistances in the structure. A thermal grease or epoxy layer for attachment, is a thin 75µm layer
developing and fully developed regimes. In addition, the channel height in the AMB substrate is varied
with a relatively high thermal conductivity of 9 W/m/K. The overall result is a reduction of the total stack
from 0.05mm to 0.3mm, which is the maximum depth allowed due to the thickness of the bottom copper
resistivity by a factor of two when compared to a microchannel heatsink.
layer. The typical coolant is water at room temperature. The pump constraints are specified as 4 lpm
Power die maximum flow rate and 25 psi maximum pressure loss, which are representative values for power
AMB electronics cooling applications.
substrate The channel width is at least 100µm, due to the difficulty of manufacturing narrower passages in copper.
Copper base
For the integrated microchannel heatsink, the preferred channel shape has a width of 100µm, a depth of
plate 300µm and a wall thickness between channels of 100µm. The narrow channel width and small pitch
results in a high surface area-to-volume ratio, while the tall channel height abates the pressure loss
through the passage. The calculated thermal resistivity for this design is 0.042K.cm2/W.
Si
The performance is effective and superior to existing heatsinks since no thermal grease layer is needed.
The overall thermal resistivity of a power module equipped with this heatsink is less than 0.15K.cm2/W,
Cu resulting in less than a 75°C junction-to-coolant temperature rise for a heat flux of 500W/cm2. This
thermal performance is better than any existing heatsink using a comparable material stack.
ceramic
Microchannel cooling is more effective for areas smaller than 7 x 7 cm. Integrated single and two-phase
microchannels Si
micro heat sinks are able to cool about 450 W/cm2 using both single and two-phase heat transfer. For
two-phase flow, the pumping power is about ten times lower and the required flow rate is considerably
(b) (a) lower. By using off-set strip fins and a split-flow arrangement, cooling of over 300 W/cm2 at 24 kPa is
possible with a flow of 1.5 lpm. A silicon microchannel cooler can be used for high-power chips. A
Figure 5.55. Integrated microchannel heatsink design, where microchannels are: separate microchannel cold plate is bonded to the back of the chip. This requires a low interface thermal
(a) fabricated directly into the bottom copper layer of the AMB substrate or resistance. If the microcooler is based on silicon, a rigid bonding means that silver-filled epoxies or
(b) created between at least two thin (0.3mm) copper etched mesh layers, bonded by a melting solder should be used, giving power densities in excess of 400W/cm2, for a flow of 1.2 lpm at 30kPa.
process. It is possible to push microchannel heat transfer even further by utilizing boiling. In addition to offering
higher heat transfer coefficients, boiling convection in microchannels requires less pumping power than
single-phase liquid convection to achieve a given heat sink thermal resistance. For the same heat flux,
Analytical optimization the pressure drops by a factor of 20. A 1000W/cm2 cooling system based on boiling heat transfer in
First-order heat transfer is governed by the thermal resistivity Rθ, which is defined as the temperature microchannel heat sinks using a flow rate of ½ lpm is possible. Local heat transfer coefficients may
rise divided by the heat flux. For convective heat transfer in channels having hydraulic diameter DH, the change appreciably over time leading to local temperature changes of 10 to 15°C. Also backflow of
thermal resistivity is calculated as: already heated flow due to expansion of bubbles is observed.
∆T DH If fluid impinges on the surface to be cooled, performance can reach 1000 W/cm2K, 14 to 21kPa and
Rθ = = (5.91)
Pd k × Nu 0.05K/W/cm2. Pumped liquid (both single and two-phase) cooling technologies in addition to loop heat
where k is the working fluid thermal conductivity and pipes for space applications, in a single-phase solution, incorporates an oscillating flow heat transfer
Nu is the Nusselt number for the appropriate flow condition ( Nu = A × Re m Pr n ). mechanism, are capable of cooling over 1300W/cm2.
A different way of making microchannels is to use metal foams or metal made porous, which
h ×L
Nu = accommodated a heat flux of 500W/cm2 for a 50K difference at a pressure drop of 115kPa, using water.
k
2
where h is the convective heat transfer coefficient, W/m K
L is the characteristic passage length of the microchannel 5.19 Electrohydrodynamic and electrowetting cooling
Pr is the Prandtl number
For example, for laminar, fully developed flow in a circular passage with constant heat flux, NuDH = 4.36. As an alternative to a continuous flow created either by temperature differences or by mechanical
[NuDH = 3.66 for convection with a constant uniform surface temperature.] means, liquid can be formed and moved in droplets of nano-to-millilitre size by means of electric fields.
In addition, the pressure loss ∆P is calculated using the friction factor f as: Electrowetting on a dielectric film, in which the surface property of a dielectric film can be modified
between hydrophobic and hydrophilic states using an electric field, can be used to provide the basis for
221 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 222

a direct micropumping system. Electrowetting involves control of the surface tension of a liquid and can This alloy is associated with material and handling problems. Liquid sodium has nevertheless been used
cause a droplet of liquid to bead or spread out on the surface depending upon its surface state. as a coolant for nuclear reactors, which shows that these drawbacks can be managed.
With 0.4 mlpm it is theoretically possible to cool 90 W/cm2. The application of electrowetting to liquid Any form of liquid cooling is associated with leakage hence reliability problems. Consideration of
metals has the advantage, besides a better heat transfer capability, of necessity a much lower voltage, emersion cooling, etc., should be restricted to applications requiring precise temperature control and
2V instead of 50V, to produce the electric field. heat dissipation rates that are too high for effective removal by conduction and air cooling.

5.21 Solid state cooling

5.20 Liquid metal cooling
5.21.1 Thermoelectric coolers
Table 5.27 shows properties for various liquid metals, where the viscosity is given at room temperature A thermoelectric module is a highly reliable, small, light, solid-state, active device that can operate as a
and water is included for comparison. Advantages of liquid metals are lower thermal expansion heat pump or as an electrical power generator. When used to generate electricity, the module is called a
coefficient compared to water and freezing introduces fewer problems. Ideally, the metal should be non- thermoelectric generator (TEG). When used as a heat pump, a refrigerator, the module utilizes the
poisonous, non-caustic, low viscosity, high thermal conductivity, and high heat capacity. Peltier effect to move heat and is called a thermoelectric cooler (TEC).
Pure Bismuth, Bi, melts at 271°C but some of it alloys have considerable lower melting points. Woods When current passes through the junction of two different types of conductors, a temperature change
metal (Lipowitz's alloy, eutectic alloy 50% Bi, 26.7% Pb, 13.3% Sn, and 10% Cd by weight) is probably results at the junction. However, the practical application of this concept requires semiconductors that
the most well known, melting at 70°C. By alloying with metals such as lead, tin, cadmium and indium, it are good conductors of electricity but poor conductors of heat. Anisotropic orientated polycrystalline
is possible to get a lower limit of 47°C. Such alloys have the disadvantage of high melting points. bismuth telluride is mainly used as the semiconductor material, heavily doped to create either an excess
(n-type) or a deficient (p-type) of electrons, as shown in figure 5.56a.
Pure gallium, Ga, melts at 29.7°C and has a latent heat of 80.1 J/g, but several of its alloys have much If the current is reversed, the heat is moved in the opposite direction. In other words, what was the hot
lower melting points. Although non-toxic and relatively cheap, the main drawback is its aggressiveness face will become the cold face and vice-versa.
towards most metals. All gallium alloys must therefore be enclosed within ceramic walls, which is difficult
A thermoelectric device consists of a number of p and n type pairs (couples) connected electrically in
to realise. Because its surface tension is much higher than water, liquid Gallium is immune to the
series and sandwiched between two ceramic plates, as shown in figure 5.56. When connected to a DC
presence of small cracks or channels in imperfect seals that would cause leaks if water were the
power source, dc current causes heat to move from one side of the TEC to the other. This creates a hot
coolant.
side and a cold side on the TEC. The device to be cooled is mounted on the cold side of the TEC and
the hot side is thermally connected to another TEC or a heatsink which dissipates the heat into the
environment. A heat exchanger with forced air or liquid may be required. A thermoelectric cooler does
Table 5.27: Approximate thermal properties for liquid metals. Water is included for comparison
not absorb heat, it only transfers or moves it.
Melting Boiling Specific Conductivity Viscosity Kinematic Prandtl Heat absorbed
Metal Density
point point heat λ ν viscosity number Pcold
2
3 Ns/m m/s cold side
°C °C kg/m J/kgK W/mK -3 cp µ / k
×10 ×10-8
bi-alloys 47-271 9,800 142 8.4 ceramic
- +
electrons holes
Ga 29.7 2205 5,900 334 28 2.04 32 0.0261
ceramic
n-type copper p-type
Hg -38.8 356.8 13,546 140 7.8 0.15 11.4 0.0278 substrate
- conductors +

FC-72 56 1680 1088 0.0545 0.45

H2O 0 100 998 4184 0.613 0.86 85.5 6.62

Ite negative (-)
NaK alloy
-11.1 783.8 892 1058 25.3 0.94 53.8 0.0213 heatsink
(22/78%)
+
Vte positive (+)
Heat rejected
The heat transfer coefficient h is the rate that thermal energy is removed from a surface per unit surface Phot
area per temperature differential, and is related to Nusselt number Nu, by (a) (b)
Nu × λ
h=
kD
where kD is the characteristics dimension of the geometry and Figure 5.56. The thermoelectric cooler: (a) principle and (b) module.
λ is thermal conductivity, shown in tables 5.26 and 5.27.
In a circular tube characterised by convection with a uniform surface temperature and laminar fully Design for power electronic cooling involves the initial specification of three parameters, the hot and cold
developed conditions, Nusselt number is constant, Nu = 4.36, and 3.66 for a uniform heat flux condition. side temperatures, Thot and Tcold, (or Th and Tc) hence the temperature gradient or difference ∆T = Th - Tc,
Apart from heat pipes based on liquid metals, mainly for the high-temperature range, Ga-Sn-In eutectics (∆T > 0) and the amount of heat, or the thermal load, in Watts, to be absorbed or removed at the cold
that remain liquid down to minus 19°C are possible. The thermal heat transfer coefficient h is ten times surface of the TEC, Pcold.
larger than for water. High-performance liquid metal cooling loops use magnetofluiddynamic MFD The cold surface temperature is the desired temperature of the power electronic component to be
pumps, with over 200 W/cm2 cooling capacity, using a flow of 0.3 lpm at 15 kPa. cooled, directly in contact with the TEC.
Mercury, Hg, has attractive thermal properties and has been used as a working fluid for power
generating purposes. It has environmental drawbacks. The hot surface temperature is defined by two major parameters:
The best liquid metal candidate is an eutectic solution of sodium and potassium, NaK. The melting point • The temperature of the ambient environment to which the heat is being rejected.
is as low as -12°C. Its density and viscosity are similar to water but has a lower specific heat and a much • The efficiency of the heat exchanger that is between the hot surface of the TEC and the ambient.
higher thermal conductivity. It can be used with nickel, chrome and steel but is aggressive to cadmium, The thermal load includes the active I2R type losses of the device to be cooled, as well as parasitic
antimony, bismuth, copper, lead, silicon, tin, and magnesium. It also reacts violently with air and water. loads such as conduction through any mechanical object in contact with both the cold surface and any
 223 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 224

warmer environment, like conduction through mounting bolts and plates (and the radiation from the Example 5.17: Thermoelectric cooler design
plates). Figure 5.57 shows the thermal resistance components and system model.
The coefficient of performance, CoP, is useful in selecting a module (the larger CoP the better), and is A semiconductor junction dissipating 100W has a steady-state thermal resistance to its case mounting
defined by of Rθ = 0.15 K/W. If the TEC cold-side temperature is not to exceed 5°C in a 25°C external ambient,
heat absorbed at the cold junction Pcold P determine the heatsink requirement to dissipate the transferred TEC heat, if the TEC has the
CoP = = = cold (5.94) characteristics shown in figure 5.58, where Th = 35°C. Assuming Ploss = 0W, determine:
electrical dc input power Vte × I te Ptec
i. the semiconductor junction temperature;
where Ite is the current drawn by the TE module ii. the coefficient of performance CoP for the TE module; and
Vte is the voltage applied to the TE module iii. heatsink requirement.
A maximum CoP represents the minimum input dc power Ptec, therefore minimum total heat to be
Solution
rejected by the heat exchanger on the hot side Phot, that is Phot = Pcold + Pe.
Performance characteristic charts, as in figure 5.58, are usually provided. These allow the terminal dc i. The junction temperature is Tcold + Power dissipated × Rθ j-c = 5°C + 100W×0.15K/W = 20°C.
voltage and dc current requirements to be determined from the temperature difference ∆T and heat to
be absorbed on the cold side, Pcold. The maximum ∆T is about 67°C for a single TEC, higher than this ii. The temperature differential across the TE module is
requires cascading (stacking) of TECs. The negative quadratic shape in the lower plot, represents the ∆T = T hot - T cold = 35°C - 5°C = 30°C
optimal operating curve. Further TE effect and module technical details can be found in Chapter 22.10.
Only one stage (no TEC cascading is required) should be necessary since the maximum ∆T < 62°C.
Tobject Since thermal leakage losses are assumed zero, the cold-side power losses to be transferred are 100W,
Pactive + Ppassive which is shown as the horizontal plotted line in the performance graph in figure 5.58.
The vertical intersection of the 100W horizontal line and the ∆T = 30°C curve gives the necessary TE
R1 R1 cold-sink thermal resistance
module input current 18.2A, and associated terminal voltage, 9.6V, from the upper curve intersection.
Ploss
The TE module electrical input power Pe is therefore 9.6V×18.A = 74.7W
From equation (5.94), the TE module coefficient of performance is
R2 R2 cold-sink spreading resistance P P 100
CoP = cold = cold = = 1.39
Pcold = Pactive + Ppassive + Ploss Vte × I te Ptec 74.7
R3 R3 module substrate and interface
Tcold
- thermal resistance
16
Ptec = Vte x Ite Σ R6 R6 bolts, gasket, internal air and
radiation thermal resistance Ptotal = 174.7W

Volts
Thot
R3
+ R3 module substrate and interface
12 60
Tamb=25°C
50
Phot = Pactive + Ppassive + Ploss + Pe thermal resistance
40
R4 R4 heat-sink spreading resistance 8 30 Rθfins=0.036°CW 10°C
20
10 Thot=35°C
Th/s
0

Ve
R5 R5 heat-sink thermal resistance 4 °C
Pactive + Ppassive + Pe
Tjunction ∆T
Ptec = 74.7W +
Tambient 0 0 TEC -30°C
250 CoP=1.39
Rθ j-TEC
10 -
Tcold

Watts
200

PTEC
TEC
model
- 20 Tcold=5°C
150
∆T 30 15°C
Rθcase=0.15°CW
+ 100
Thot 40
Pload Tj=20°C
Rθ TEC-amb 50

Pcold
Ptotal 50
Pload = 100W
Tambient
°C

Heat transfer without TEC 0 60

0 4 8 12 16 20 24 28
0K
Ite Amperes
T

Heat transfer with TEC

Figure 5.58. Thermoelectric module characteristics,
Temperature

Pmax = 270W, with 127 couples for a 35°C hot-side temperature Thot, ∆T = 62°C, and Imax = 30A.
Tj TTIM1 Tspreader Tcase Theat sink Tambient
iii. The heatsink thermal resistance requirement is
Rθj-TIM1 RθTIM1-spreader Rθspreader-case Rθcase-sink Rθsink-ambient T − T amb
Rθ hs = hot
Power
Power dissipated PD + Pe 35°C − 25°C 10°C
PD TEC Power dissipated Ptec = = = 0.036°C/W
100W + 18.2A × 9.6V 100W + 18.2A × 9.6V
Figure 5.57. The thermal resistance TEC cooling model.
♣
225 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 226

Example 5.18: Thermoelectrically enhanced heat sink • For a CoP greater than 3.5, sufficient ∆T is not generated by the TEC to offset the additional
interface and heat sink rise and still provide measurable performance improvement.
A switching semiconductor device dissipates 120W. Figure 5.59 shows the thermal resistor equivalent • For a CoP less than 2, considerable additional input electrical power must be supplied and
network and associated resistances for cooling system designs based on: consequently dissipated by the heat sink.
Tamb = 40°C
Rθhs-a = 0.18 K/W For the 120W switch heat dissipation, from equation (5.94), a CoP of 3 equates to a TEC input power of
Rθs-hs = 0.08 K/W 40W and a optimal TEC ∆T of about -14.5°C from figure 5.60. This 40W must be dissipated by the heat
RθTIM2 = R θTIM3= R θTIM4= 0.2 cm2 K/W sink in addition to the switch 120W heat load, thus raising the heat sink temperature by
ATE = 64cm2, representing the area of four TECs, each 4x4 cm 40Wx0.18°C/W=7.2°C. Multiple TECs cover the heat sink area (total TEC area of 64cm2). Thus the heat
Aswitch = 9cm2 sink has a much larger heat input area compared to the no-TEC case thereby eliminating the need for a
i. Using the TEC characteristics in figure 5.60, calculate the switch case temperature with and thick copper base plate (0.08°C/W assumed) or embedded multiple heat pipes to spread the heat. The
without a TEC, and the effective thermal resistance in each case. need for spreading is moved from the heat sink to the cold side of the TEC. This cold side spreader
ii. What is the maximum heatsink thermal resistance which when exceeded does not justify using could be a heat pipe assembly, vapour chamber, or solid copper spreader, with are cost, performance,
TECs. and weight tradeoffs for each case.
T c = T amb + Pd × (RθTIM 2 + Rθ spread + RθTIM 3 ) − TTEC + (PD + PTEC ) × (RθTIM 4 + Rθ fins )
Solution
= 40°C + 120W × ( 0.2 / 9 + 0.08 + 0.2 / 64 ) − 14.5°C + (120W + 40W ) × ( 0.2 / 64 + 0.18 )
i. For the no-TEC case, the switch heat load is 120W from a 30 x 30 mm package. The package is
mounted to a heat sink using thermal grease with a TIM2 thermal resistance of 0.2 m2.°C/W. The sink to = 40°C + 12.7°C − 14.5°C+29.3°C
ambient thermal resistance comprises two components; a spreading component due to the small heat T c = 67.5°C
input area, and the fin to ambient component as shown in Figure 5.59.
With the TEC, the case temperature decreases from 73.9°C to 67.5°C, a reduction of approximately
From an equivalent Ohms law calculation, the case temperature without a TEC is
6.4°C, resulting in a 19% reduction in Rθc-amb for the 120W case. Note that the TEC actually operated
T c = T amb + PD × (Rθ fins + Rθ spread + Rθ'TIM 2 / Aswitch ) with a 14.5°C negative ∆T, yet the case temperature was reduced by only 6.4°C. More than half of the
T c =40°C + 120W × (0.18K/W + 0.08K/W + 0.2cm2K/W / 9cm2 ) = 73.9°C. negative ∆T from the TEC must be used to offset the additional interface losses and additional rise in
heat sink temperature due to the TEC 40W input power.
This equates to a case to ambient thermal resistance of
∆T net = ∆TTE + ∆T interfaces + ∆T heatsink
(T -T ) = ( 73.9°C - 40°C ) = 0.283°C/W.
Rθc -amb = case amb = −14.5 + 0.5 + 0.4 + 7.2 = −6.4°C
P 120W Also, it should be noted that for the baseline system without the TEC, the 0.283°C/W thermal resistance
is a constant with respect to heat load. For the TEC, the 0.229°C/W is not a constant with respect to
No TEC with TEC, CoP=3 heat load and accurately represents the thermal resistance for the 120W heat load only.

Tamb 40°C Tamb 40°C No gain in using a TEC approach results when the TEC thermal drop of 14.5°C is the same as the
temperature rise – when the junction operating temperature is the same by both cooling methods.
Rθfins 21.6°C Rθfins 28.8°C
21.6°C+7.2°C
Tfins 61.6°C Tfins 68.8°C ii. With a poor heat sink with a fin to ambient thermal resistance of 0.362°C/W instead of 0.18°C/W (with
all other parameters maintained), the additional heat sink temperature rise due to the 40W TEC power
Rθspread 9.6°C RθTIM4 0.5°C increases from 7.2 to 14.5°C (0.362°C/W x 40W). The TEC gains become null and void.
Tsink Thot ♣
71.2°C + 69.3°C
Q / Qmax
RθTIM2 2.7°C Ptec = 40W TEC -14.5°C
1 5
- 0

∆T / ∆Tmax
Tcase Tcold 54.8°C
73.9°C
10 0.1
∆T=0°C

CoP
RθTIM3 0.4°C
120W

CoP
55.2°C 0.8 0.2 4
8 10°C
Rθspread 9.6°C
Effective case-to-ambient thermal resistance, RθC-A um 0.3

Coefficient of performance
Tspreader 64.8°C 0.6 tim
op 3

Coefficient of performance
73.9 − 40 6

Normalised temperature
No TEC = 0.283°C/W 0.4
120 RθTIM2 2.7°C range of 15°C
67.5 − 40 Tcase 67.5°C optimal performance 0.5
With TEC = 0.229°C/W 4
120 0.4 2
120W 0.6
20°C
2 0.7
Figure 5.59. Typical thermal resistor equivalent networks with and without a TEC. 0.2 1
30°C 0.8
0
40°C
0 10 20 30 40 50 60 70 50°C 0.9
Although a TEC adds additional thermal interfaces, it interposes a negative ∆T in the thermal path. Also 60°C
0 0
a TEC allows a simpler heat sink, one designed for a uniform heat input where spreading in the heat TEC element °C ∆T = Thot - Tcold
sink is not necessary. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
(a) (b) TEC element normalised current Ite / Imax
With TECs, assume the four TECs operate with a CoP of 3. For practical applications of switch cooling,
the CoP will likely be between 2 and 3.5, as shown in figure 5.60a, corresponding to TEC ∆T 's between Figure 5.60. Generic Bi2Te3 thermoelectric characteristics:
10 and 20°C. (a) theoretical optimum (maximum) CoP vs ∆T and
(b) CoP as a function of current and generic thermoelectric performance curves.
227 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 228

Using examples 5.17 and 5.18 as a benchmark cases, when considering utilising a TEC: 5.21.3 Thermionic and thermotunnelling cooling
• The TEC should operate near its maximum (optimum) CoP for the ∆T chosen. As shown in
Figure 5.60b, off-optimum operation results in significantly higher TEC power consumption Thermionic cooling is based on the principle that a high-work-function cathode preferentially emits hot
(lower CoP). This results in a larger ∆T across the hot side thermal interface and a larger electrons. Materials available have a work function of 0.7eV or higher, which limits the use to the higher
additional rise in heat sink temperature, thus reducing the net gain from the TEC negative ∆T. In temperature ranges (>500K). Vacuum thermionic devices based on resonant tunnelling have cooling
addition, for the same CoP and thus the same input power, an off optimum TEC will not achieve capabilities of 20 to 30°C with kW/cm2 cooling power densities achievable. However, since the operating
the same ∆T, thus further reducing the ∆Tnet. currents for the device are as high as 105A/cm2, effects such as Joule heating (I2R) at the metal-
• Analysis of Figure 5.59 shows there will be cases when a TEC should not be used - even if it is semiconductor contact resistance and reverse heat conduction limit cooling to <1°C.
optimally designed. Even for an optimally designed TEC that behaves according to Figure Devices based on quantum tunnelling through a small gap, with the spacing between the cathode and
5.60a, if the heat sink performance is poor, the additional rise in heat sink temperature, ∆Ths-amb the anode of the order of 10 nm provide much larger cooling power than thermoelectric superlattice
will offset more of the negative ∆T from the TEC.. coolers, specifically 10kW/cm2 for 50K cooling at room temperature.
• The calculations in Example 5.17 were performed at the design power heat dissipation at the
maximum ambient conditions 40°C. Typically, the switch may not operate for prolonged periods 5.22 Cooling by phase change
at these extreme conditions. When the switch heat dissipation is less than 120W or when the
ambient is less than the maximum, the realized TEC CoP can be significantly greater, rising Phase change materials and heat accumulators
exponentially as the required TEC ∆T is decreased (see Figure 5.60a). CoPs higher than 5 Phase change materials for electronics thermal management are limited to applications where time-
would be likely under typical operating conditions when only small ∆T would be required by the dependent phenomena play a role.
TEC in order to maintain the maximum junction temperature below its specification. Likewise, In the case of heat accumulators, the use of composite materials based on a granulated open-porous
operating the TEC at a constant input power under these non-peak conditions offers the matrix filled with a hygroscopic substance is an approach to accumulate heat. The advantage is a
opportunity for significant reductions in fan speed with favourable acoustic benefits. significant increase in the heat that can be stored as compared to sensible heat (surrounding heat,
potential energy) and latent heat. For example, for a 100°C temperature rise, copper absorbs 40kJ/kg.
TEC requirements Evaporation of water is associated with an absorption of 2260kJ/kg. The enthalpy of a reversible
chemical reaction can reach a value of 7000kJ/kg. A principal advantage of reversible chemical
In the TE enhanced heat sink example 5.18, four TECs and a large spreader were utilized, where each reactions for heat accumulation is their ability to store the accumulated energy for a long time, if the
TEC pumped 30W giving the required total of 120W. The generalized thermoelectric performance curve reaction is controlled by the presence of either a catalyst or a reagent.
given in Figure 5.60b gives an indication of the type of TEC required. For a CoP of 3 operation, the
optimum ∆T is 14.5°C (figure 5.60a). For a typical TEC with a ∆Tmax of about 67°C, the ratio of ∆T/ Phase change material thermal properties
∆Tmax is about 0.21. From Figure 5.60b, the curve for optimum (lowest power) operation, P/Pmax is When power electronics is operated under transient conditions, increasing the thermal capacitance is a
approximately 0.17. Therefore, for a 30W TEC heat load, the Pmax for each TEC should be useful technique for limiting temperature increases and/or minimizing the performance requirements of a
approximately 30W/0.17 = 176W. heat sink. An effective method of increasing thermal capacitance is to utilise a material that undergoes a
change of phase at a desirable temperature – at which temperature there is thermal energy transfer
The four TECs and a large cold side spreader add cost and weight to the system. A single TEC with
associated with the change of phase. Utilizing phase change for temporary thermal energy storage has
partial spreading of the heat prior to the TEC and additional spreading between the TEC and heat sink
the benefit of allowing heat rejection to occur over a longer time interval, which in turn allows a smaller
could offer some performance/cost tradeoffs. However, if the entire heat from the switch were pumped
heat sink. Examples of phase change that have been used in power electronics cooling include solid-
by a single TEC, the Pmax of that single TEC would need to be 120W/0.17=700W, which exceeds
liquid, liquid-vapour, and solid-solid (for example, crystalline structure to amorphous). A solid-liquid
available limits of about 550W (for ∆T=0).
phase change is most common for systems that require reuse of the phase change material (PCM). The
thermal energy required to melt the PCM is described as the latent heat of fusion and latent heat of
5.21.2 Superlattice and heterostructure cooling
vaporisation is applicable to a liquid to gas phase change. Directing the waste heat from the power
The strategy to improve thermoelectric cooling has turned to the nano scale level, where coherent and electronics into the PCM can result in a near isothermal heat sink while the PCM is changing state -
incoherent transport plays an important role in electron and phonon diffusion. ZT values (a figure of melting or vaporising.
merit parameter, see Chapter 22) between 2 and 3 at room temperature are obtained with Bi2Te3/Se2Te3 Selecting a PCM requires knowing the range of expected temperatures (the melt or vapour temperature
superlattices. Cooling power density is as high as 700W/cm2 at 353 K compared to 1.9W/cm2 in the bulk of the PCM must be high enough such that state change does not occur until needed). Another
material, figure 5.61. specification is duration index, DI, which is a comparison of time of the PCM at constant temperature
Thin-film, based on SiGe/Si, gives a cooling power density of almost 600 W/cm2 for a temperature during the phase change.
difference of 4K below ambient for a 40 x 40 micrometer area. A superlattice approach produces a ZT h ×ρ
larger than 1.4.
DI = f (5.95)
∆T
The list of candidates is restricted when issues such as material compatibility, electrical conductivity,
toxicity, availability of thermal property data, and cost are considered.
2

800
W cm

Heat of fusion (solid to liquid)

Desirable characteristics of a solid-liquid PCM include high heat of fusion per volume (DI), congruent
600 Superlattice melting and freezing characteristics, high thermal conductivity, minimal supercooling, and low thermal
Cooling power density

Thot-side = 353 K expansion. Table 5.28 lists four categories of solid-liquid PCMs with melting temperatures in the range
of 40° to 120°C, with water characteristics provide as a reference.
400 The main drawback of metal eutectic PCMs is high density, resulting in heavy package solutions. Salt
hydrates have handling and safety issues. Two salt hydrates with high volumetric heat of fusion are
Superlattice Lithium Nitrate Trihydrate (679×106 J/m3) and Barium Hydroxide Octahydrate (656×106 J/m3). The first is
200 Thot-side = 298 K a severe oxidizer with a moderate health rating and the second carries a severe health warning. The
organic non-paraffin category has the largest variability. Acetamide is an example of a non-paraffin
organic PCM that melts at 81°C and has only moderate handling and material compatibility issues.
0 The organic paraffins represent a reasonable compromise between handling and performance and are
0 1 2 3 4 available in a wide range of melt temperatures. In general, higher melt temperature paraffins are more
expensive, especially for high purity levels. The low thermal conductivity of these materials usually
Current I (A)
requires the use of an imbedded matrix material to help conduct heat into the PCM. The volume change
Figure 5.61. Estimated power density for superlattice devices as a function of current. between solid and liquid states limits their packaging density.
229 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 230

Table 5.28: General solid-liquid PCM thermo-physical characteristics Compared to a solid to liquid phase change, the volume change associated with a liquid to vapour
phase change is significantly larger, often greater than two orders of magnitude. Even though heats of
property
Inorganic Organic Organic Inorganic Inorganic vaporization are typically larger than heats of fusion for a particular material, the large volume change
water paraffin non-paraffin salt hydrate metal eutectic limits the ability to take advantage of this property for simple thermal energy storage. When it is not
Phase change temperature °C 0 37 2 - 164 7 - 117 <0 required to contain and reuse the vapour, such as in a single use application, expendable coolants that
undergo a liquid-vapour phase change as they absorb heat are a consideration.
Latent heat of fusion hf kJ / kg 334 230 - 290 120 - 240 170 - 340 30 - 90
3
kJ / m More common in power electronics cooling, the use of heat of vaporization occurs in closed systems
Latent heat of vap hfv 2 22.7 190 - 240 140 - 430 250 - 660 300 - 800
x 10 where heat removal is augmented by the vaporization property and then released elsewhere in the
Density ρℓ kg / m 3
1000 ≈ 810 900 -1800 900 - 2200 ≈ 8000 system when the vapour is condensed. Examples for cooling hardware using this property include heat
pipes, pumped refrigerant, and spray cooling. The operating pressure of these systems can be adjusted
Thermal conductivity λ W / mK 0.6 ≈ 0.25 ≈ 0.2 0.6 - 1.2 ≈ 20 to provide an attractive boiling point for the coolant. Desirable qualities for a coolant that will operate in a
Duration Index DI J/ m3K 13.32 2.96 5.44 two-phase mode include a high heat of vaporization, acceptable boiling temperature and pressure, low
corrosion potential, low toxicity, environmentally friendly, and low cost. The heat of vaporization and
Thermal expansion low high moderate low low melting temperature for several common substances at standard temperature and pressure are listed in
Congruent melt yes yes some do most do not yes Table 5.29.

Supercool yes no no most do no

Corrosion low low some are highly some are
Reading list
Toxicity no no some are highly some are
Simple to use Cheap Fishenden, M. and Saunders, 0. A., An Introduction to Heat Transfer,
Advantages
No nucleating agent Non-flammable Oxford University Press, 1982.
Careful preparation
More expensive
Disadvantages Need stabilising
Some combustible
additives [Link]

[Link]

Table 5.29: Heat of vaporization at standard temperature and pressure (see Table 5.23) [Link]

[Link]
Substance Boiling Point Heat of Vaporization
°C kJ / kg [Link]
Helium -268.9 21 [Link]
Hydrogen -252.8 461
Nitrogen -195.8 199 5.23 Appendix: Comparison between aluminium oxide and aluminium nitride
Propane -42.1 427
R12 -38.8 165
Aluminium Nitride is made by nitridation of aluminium or by conversion of alumina Al203 to aluminium
nitride. It is a covalently bonded material and has a hexagonal crystal structure. Because of its
Ammonia -33 1369 resistance to sintering, an oxide-forming additive such as Y203 is needed to form a substrate. General
R134a -26 178 comparison properties of aluminium oxide and nitride are listed in Table 5.30.
Fluoroketone 49 88 Aluminium nitride is a cost effective, non-toxic alternative to beryllium oxide and has a thermal
Acetone 56 518 conductivity nearly eight times higher than alumina (aluminium oxide). Advantages of aluminium nitride
include good thermal performance, low thermal expansion, and non-toxicity. Aluminium nitride offers:
FC-72 56 88
• High thermal conductivity
Methanol 64.7 1100 • Low thermal expansion coefficient closely matching silicon
Ethanol 78.4 846 • Good dielectric strength
• High electrical sensitivity
Water 100 2256
• Low toxicity and therefore excellent replacement for Be0
Ethylene glycol 197 800 • Good shock and corrosion resistance
Glycerine 290 974 • Low dielectric loss
• High temperature stability
Mercury 347 295
• High flexure strength and light weight
• Resistant to wafer processing gasses and plasma erosion
• Conducive to finishing operations such as lasering, lapping, and polishing
Heat of vaporization (liquid to vapour) • Substrate for direct bond copper DBC and filled vias
• Good adhesion for thin and thick film applications
The energy required to change a substance from a liquid state to a vapour state is termed the heat of • Uniform lapped and polished surfaces for resistor networks
vaporization, and is directly related to overcoming the intermolecular bonding forces in the liquid state. • Polished and lapped surface finishes to 12nm, Ra, with minimum pullouts
The energy released when a vapour is condensed to a liquid is equal to the heat of vaporization but has • Lapped surface finishes to 150nm, Ra, where is the roughness average profile
an opposite sign and is commonly termed the heat of condensation. The most common applications of 1 n
this thermodynamic property in power electronics cooling are associated with boiling heat transfer. Ra = ∑ y i
n i =1
 231 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 232

-1
10 K
-6
10

coefficient of thermal expansion

9

6
0 0.2 0.4 0.6 0.8 1 1.2
copper thickness mm

Figure 5.62. Thermal expansion dependence on copper base plate thickness.

The coefficient of thermal expansion for direct copper bonded (DCB) substrates with a layer of 0.6 mm
alumina sandwiched between Cu layers of various thicknesses, is shown in figure 5.62.
Table 5.30: Substrates properties

Beryllium Silicon 5.24 Appendix: Properties of substrate and module materials

Material Aluminium nitride Aluminium oxide
oxide carbide Zn Ag
symbol Aℓ3N4 Aℓ2O3 pure metals
Colour Tan/Gray white white Ni Al
Melting point/ maximum >2200/800 alloys
°C 2054/1700 2507/1800 -/1650
use temperature oxidizing
plastics ice oxides
Purity (wt %) 98% 99.6% 99.5
nonmetallic solids
3
Density (g/cm ) >3.25 >3.65 2.85 3.2
Foams fibres
Thermal Conductivity (W/m. K) 100 - 300 27 265 270
-6
insulation systems
(x10 /K)
Thermal Expansion @ >20°C
<4.3 <7.7 8.0 3.7
Oils H2 O Hg
Specific Heat (J/kg K) 740 880 750 liquids
Dielectric Strength (kV/mm) >15 >12 9.5
CO2 H2
(at 1MHz)
Dielectric Constant @ 20°C
8.7 9.2 6.6 40 gases
(x10-4 @ 1
Dissipation Factor MHz)
3-7 3 4
>1012 at 20oC 0.01 0.1 1 10 100 1000
14
Volume Resistivity (ohm-cm) >10 1015 102-106
>108 at 500oC Thermal conductivity W /mK
(kgf /mm) >250 >280 120-150 450
Flexural Strength MPa 300 345 150-200 550
Modulus of elasticity GPa 331 372 345 410 Table 5.31: Substrate characteristics
Hardness (Knoop) GPa 11.8 14.1 9.8 27
Poisson’s ratio 0.22 0.21 0.26 0.14 Material Al2O3, BeO, AIN, Quartz, Silicon, Sapphire, Ferrite

toxicity nontoxic nontoxic toxic nontoxic Surface Finish Al2O3 As Fired, 0.05µm maximum;

Substrate Specification Polished to 0.12µm

Maximum Dimension (mm) 140 x 100mm, 100 x 200 Dimensions 0.5mm x 0.5mm to 100mm x 100mm

Thickness (mm) 0.63 - 0.2 mm 0.63 - 0.30mm Dimensional Tolerance ± 0.01mm scribed; (± 0.002mm saw cut)
as fired - 0.3 as fired - 0.3 as fired - 0.3 Thickness 0.012mm inches to1.2mm
Surface Roughness (micron) as lapped - 0.075 as lapped - 0.075 as lapped - 0.1
as polished - 0.025 as polished - 0.025 as polished-0.08 Thickness Tolerance 0.12mm standard to as tight as 0.04mm
Sputtered Resistor Material NiCr, Ta2N Sputtered Metallization Ti, TiW, Pd, Ni, Au, Al
Electro-Plated Metals Au, Cu, Ni, Solder
At room temperature, the thermal conductivity of aluminium nitride ceramics is independent of Aℓ3N4
grain size or number of grain-boundaries, but is controlled by the internal structure of the grains, such as Electroless Plated Metals Sn, Ni, Au
the degree of oxidation (oxygen contamination). Thermal impedance is compared in the following figure.
 233 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 234

Table 5.32: Coolant properties Table 5.34: Power electronic component properties

specific temperature
thermal Melting
material heat density coefficient
Specific Heat Density Thermal Conductivity conductivity point
capacity of expansion
Material cp ρℓ λ 20°C λ c ρℓ CTE
J/kgK kg/m
3
W/mK W /m K J/ kg K kg/m3 ppm/K K
air (STP) 0.026 1004 1.2 -
Coolanol 25 1,838 903 0.12
silicon Si 120 700 2330 3.5 1685
Diala-X 1,840 870 0.14
solder PbSn 50 150 8400 24.1 183
20/80 EGW Solution 3,817 1023 0.57
copper Cu 385 385 8930 17 1358
50/50 EGW Solution 3,283 1064 0.39 alumina Aℓ203 22 80 3720 6.5
Fluorinert®, FC-77 1,028 1771 0.063 aluminium nitride Aℓ3N4 170 725 3300 4.5
Hydraulic Oil 1,842 868 0.12 aluminium silicon carbide AℓSiC 170 3000 7
Polyalphaolefin 2,180 794 0.14 polyimide 0.2 1100 1400

SAE 10W Oil 1,901 875 0.12 dielectric layer 0.3 1400 1120
encapsulation 0.5 2000
SAE 30W Oil 1,901 875 0.12
aluminium Aℓ 205 900 2710 22.5 775
Stainless Steel, 316 500 8025 16.20
gold Au 315 126 19320 14.2 1336
Water, H2O 4,184 998 0.59
platinum (90%) Iridium (10%) 31.1
Air 1,008 1.1 0.27 platinum Pt 70.9 1448 21450 9.0 1728
Mercury 137 13,800 137 tungsten W 188 130 19300 4.6 3410
molybdenum Mo 140 250 10200 4.9 2610
lead Pb 235 130 11340 23 327
tin Sn 66 227 7300 23 232
Cu/Mo/Cu 5.8
silicon carbide SiC 700 250 3.21 3.7
Table 5.33: Properties of module materials sapphire 2700 419 3900 8.4 2040
diamond 2300 509 3500 2.4 3100
brass CuZn 111 343 8490 18.0 920
linear thermal
specific temperature steel (low carbon) 48 460 7850 11.5 1370
relative dielectric loss expansion
thermal coefficient of type
material permittivity factor coefficient Mica K Mg3 AlSi3 O10 (OH)2 0.6 1700
conductivity εp
beryllium Be 230 1088 2880 5.9 1280
@1MHz 10GHz, 25°C @25°C
εr tanδε λ ∆l / l /∆T ∆ε / ε∆T silicon thermal grease 0.8 2093 2.8 -

×10-4 W/m K 10-6/K 10-6/K thermal conducting plastic 20

Aℓ203 99.5% 9.8 1 37 6.5 136 insulator

sapphire 9.4 1 42 6 110 insulator
Quartz glass 3.78 1 1.7 0.55 13 insulator 5.25 Appendix: Emissivity and heat transfer coefficient
Beryllium oxide ceramic insulator
6.3 60 210 6.1 107 3
Be0 8900kg/m Emissivity is not only a material property but also a surface property, at least for opaque materials.
GaAs 12.9 20 46 5.7 semiconductor Consequently, coatings (oxides, grease, and water film) influence the value measured under pristine
Silicon conditions. For example, the emissivity of a copper surface covered with 2 µm oxide increases from 0.03
3 11.9 150 145 4.2 semiconductor to 0.2. In practice, surfaces that are initially shiny are covered with oxide and dust after one year of
ρ = 10 Ωcm
PTFE 2.2 3 0.2 106 350 plastic operation. Additionally, surface texture can influence the emissivity because of a strong dependence on
angle.
polyolefin 2.32 7 0.5 108 480 plastic
copper 1.0 393 17 24.1 metal For heat transfer calculations, total hemispherical emissivity is found by integration over all wavelengths
aluminium −1300+i1.3×1014 220 23.8 metal and all angles. What is being measured with an IR camera is the normal spectral emissivity restricted to
the wavelength band of the detector.

As a rule of thumb: for unpolished metals the ratio of hemispherical to total emissivity is 1.1 to 1.3, and
for non-conductors 0.95 to 0.97. Another angle-dependent difference between metals and insulators is
that under a shallow incident angle, the emissivity of metals tends to one and insulators tend to zero.
235 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 236

A complication is, the emissivity of many materials of interest is strongly dependent on the wavelength. highly polished metals, foils, and films
That is, there is a big difference between the visible band and most common IR bands (0.8 to 3 µm, 3 to
8 µm, 8 to 14 µm). Si and Ge are notorious examples. It is relevant to low-temperature applications polished metals
(<100°C) because the bulk of the radiation is in the long-wavelength region: 8 to 14 µm. This is the main metals untreated
reason why the colour of paints is not relevant for heat sinks and casings. All colours are black. Apart
from the fact that it is simply impossible to account for all the physics of radiation in a practical situation, metals oxidised
the problem is only relevant when radiation is an important contribution to the total heat transfer. For
natural convection cases, this contribution might well be 40%. When the choice is between an emissivity oxides and ceramics
of 0.05 and 0.1, the radiation contribution is 2 to 4% - minimal. When the choice is 0.7 to 0.9, matters
are different. Most heat transfer simulation software that support radiation heat transfer, assume the carbon, graphites
following simplifications:
minerals, glasses
• surfaces diffuse, grey (sometimes specula [mirror-like]), opaque
• surfaces isothermal vegetation, skin, water
• surfaces uniformly irradiated
• medium is transparent for all relevant wavelengths special paints, anodised finishes

In summary, differences exist between experimental data and simulation results when radiation is a
0 0.2 0.4 0.6 0.8 1
major factor, even when all other data are known to within 1%.
Surface emissivity ε
Table 5.35 shows typical material normal total emissivity values @ 20°C.

Table 5.35: Normal total emissivity values @ 20°C

visible

Metals ε Coatings ε
Aluminium polished 0.04 Aluminium bronze 0.3
sheet 0.09 Aluminium paint 0.35
Brass polished 0.05 Blackbody paint 0.97
X-rays infrared
oxidized 0.22 Enamel 0.82
ultraviolet microwave
Chromium polished 0.06 Lampblack 0.95
rough 0.74 Paint 0.89 thermal radiation
gamma rays
Copper polished 0.03
Gold polished 0.025 Various
Graphite polished 0.42 Glass, Quartz 0.93
Inconel polished 0.2 Ice 0.98
Iron polished 0.06 Plastics 0.8 λ µm
oxidized 0.85 Paper 0.8
ground 0.24 Porcelain 0.92 Electromagnetic spectrum

cast 0.16 Silk, Cotton, Wool 0.75

Mercury 0.09 Stone 0.8-0.9
Molybdenum polished 0.05 Water (> 0.1mm) 0.95 Natural convection heat transfer coefficient

@ 2600 K 0.29 Wood 0.9 The convective heat transfer coefficient h, for various geometry arrangements are given by equations
Silver polished 0.025 (5.96) and (5.97), as applicable, in conjunction with Table 5.36.
¼
 ∆T 
Steel polished 0.06 h = kh   (5.96)
oxidized 0.6
 D 
¼
 ∆T 
Silicon difficult 0.3 - 0.8 h = kh   (5.97)
 L 
Tin bright 0.07
where ∆T is temperature difference, K
Tungsten polished 0.05 L is length, m
D is diameter, m
237 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 238

Table 5.36: Heat transfer coefficient constant, kh DCB AMB

Dependant geometrical constant

Geometry equation Eutectic copper- copper
parameter kh
copper oxide Solder
Vertical cylinder or plate Vertical height L 1.42 (5.97) Ceramic Silver/copper/titanium
Horizontal cylinder Diameter D 1.32 (5.96) initial Aluminium oxide
fracture
4 × area Upper surface hot 1.32
Figure 5.63. Direct copper bonding, DCB and active metal brazing, AMB.
Horizontal plate L= (5.97)
perimeter Lower surface hot 0.59
Vertical populated PCB Vertical height L 2.44 (5.97)
Advantages of the DCB-technology compared to other structures are mainly the high current
Sphere Diameter D 1.92 (5.96) conductivity due to the copper thickness, good cooling features due to the ceramic material, the high
adhesive strength of copper to the ceramic (reliability), and the optimal thermal conductivity of the
ceramic material. Possible failure due to cracking, termed conchoidal fracture, starts at the copper edge,
5.26 Appendix: Ampacities and mechanical properties of rectangular copper busbars as shown in figure 5.63, and progressively extends under the copper interface area.

Effect of emissivity and number of busses on ampacity (current carrying capacity) – data in Table 5.37 AMB (Active Metal Brazing)
shows how higher emissivities improve ampacity. Multiple busses also affect ampacity in a nonlinear The AMB process (brazing of metal foil to a substrate) has been developed based on DCB technology.
relationship. Ampacity may be raised by increasing heat dissipation with convection cooling or surface The advantages of AMB-substrates with AℓN-ceramic materials compared to substrates with Aℓ203-
treatments. Surface treatments which improve emissivity are oxidation or thinly coated, flat, inorganic ceramic materials are lower thermal resistance, lower coefficient of expansion, and improved partial
based spray paints. discharge capability.

Table 5.37: Ampcapacity Figure 5.63 illustrate the differences between DCB and AMB. AMB offers higher partial discharge levels
than DCB.
Ampacity, A
number of 30°C rise 50°C rise 65°C rise IMS TFC
6mm x 100mm
busses emissivity ε emissivity ε emissivity ε Chip: Si, 280µm
copper Printed conductor
0.15 0.4 0.7 0.9 0.15 0.4 0.7 0.9 0.15 0.4 0.7 0.9 Solder: SnAg, 80µm
Cu 100µm Cu, 30 - 200µm
1 1100 1250 1400 1600 1500 1700 1900 2000 1700 1950 2200 2300 Isolation Isolation Al2O3, 380µm
2 1900 2050 2200 2300 2550 2750 2950 3100 2950 3200 3400 3600 Polyimid 25µm
Adhesive: silicone, 35µm
3 2500 2700 2850 3000 3400 3600 3850 4000 3950 4200 4500 4600 Baseplate
4 3100 3300 3450 3600 4200 4400 4700 4800 4900 5100 5400 5600 Cu or Al, 3mm

6mm spacing.
Ampacities of bus bar systems of other configurations must be calculated taking into account size, spacing, number
of bus bars, and overall skin-effect ratio.
Figure 5.64. Basic module structure of: (a) an IMS power module and (b) a TFC power module.

5.27 Appendix: Isolated substrates for power modules

IMS (Insulated Metal Substrate)
Currently used isolated substrates for power modules are: IMS is mainly used in the low-cost, low-power range and is characterized by direct connection of the
isolation material to the module base plate. For insulation, polymers (such as epoxies, polyamides) are
Isolation material applied to an aluminium base plate, as seen in figure 5.64a. The upper copper layer is produced in foil
Ceramic: aluminium oxide Aℓ203 Organic: epoxy form and glued onto the isolation substrate (similar to PCB production) and is patterned by etching.
aluminium nitride AℓN polyimide (Kapton) Advantages of IMS are low costs, filigree track structure (possible integration of driver and protection
(beryllia oxide Be0) circuitry), substrate high mechanical robustness, and relatively wide substrate areas, compared to DCB.
(silicon nitride Si3N4) The thin isolation layer, however, leads to comparably high coupling capacitances associated with the
Substrates mounting surface. Also the thin upper copper layer only provides a comparably low heat spreading,
Metal sheets: DCB (Direct Copper Bonding) Metal sheets: IMS (Insulated Metal Substrate) which is improved by additional metallised heat spreading layers under the chips or by adding Aℓ-
AMB (Active Metal Brazing) Multilayer-IMS particles to the isolation layer.
Thick film layers: TFC (Thick Film Copper) TFC (Thick-Film-Copper)-thick film substrates
Just as with DCB, the basic material for thick film substrates is an isolation ceramic, which is glued
DCB (Direct Copper Bonding)
directly onto the base plate or a heatsink by means of silicone or applied by soldering, as shown in
Power modules with IGBTs (or MOSFETs) and freewheel diodes commonly use substrates made of DCB-
figure 5.64b. The tracks on the top of the ceramic substrate are made of copper and are applied by
ceramics with Aℓ203 or AℓN isolation that combine good thermal conductivity and high isolation voltage.
screen printing. The power semiconductor chips or other components are soldered or glued onto the
For DCB, copper surfaces 200µm to 600µm thick, typically 300µm, are applied to the top and bottom
track pads.
surfaces of the isolation substrate material (0.25mm to 0.85mm thick, typically 0.5mm) by eutectic
melting at between 1065°C and 1083°C. The sandwiched copper oxide layer helps adjust for the
TFC technology can also be combined with standard thick film technology. Since low resistances may
different thermal expansion rates. After the necessary track structure for the module circuitry has been
be produced by the paste materials which are usually applied in thick film technology, and since isolated
etched into the top side copper surface, the chips are soldered on, and contact connection on the chip
tracks can be arranged on top of one another and connected together, quite a number of system
top side is effected by bonding. The bottom side copper of the DCB-ceramic substrate is fixed to the
components may be densely integrated. However, the filigree tracks, typically 15µm thick, limit the
module base plate (about 3mm thick copper) usually by soldering, as seen figure 5.8. Other module
current capability of such structures to about 10A.
types do not necessarily require a base plate and the soldering procedure may be avoided.
 239 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 240

High-temperature lead-free transient liquid phase (TLP) die and substrate attach methods
Ptot1 Ptotn While silicon semiconductor technology is limited to junction temperatures of about 200°C, emerging SiC
technology could exploit 600°C operating temperatures, were it not for die and substrate attachment
limitations and aluminium thermal bonding stressing.

The high-temperature, lead-free silver-tin transient liquid phase (TLP) die attach process for connecting
the SiC power devices to a nickel-plated direct bond copper (DBC) or direct bond aluminium (DBA)
power substrate (aluminium nitride or silicon nitride) shown in figure 5.66a, allows junction temperature
operation in excess of 400°C.

Similarly the high-temperature, lead-free nickel-tin TLP attachment process for connecting the nickel-
plated DBC or DBA power substrate to a metal matrix composite, MMC, base-plate allows operation to
temperatures in excess of 400°C.

The baseplate of the power module can utilize a lightweight copper-molly (CuMo) metal matrix
composite (MMC) that has a coefficient of thermal expansion (CTE) characteristic closely matching that
of the SiC power die. This CTE matching reduces thermal-stress mismatches, thus improving the long-
term thermal stressing reliability of a power module.

SiC die DBA

substrate

to
to

DBA base-plate
substrate

(a) (b)

Figure 5.66. Cross-section of various layers in the lead-free 400°C high temperature SiC power
module: (a) Ag-Sn TLP die attach and (b) Ni-Sn TLP substrate attach.

Al bonding wire Soldered CONVERSIONS

Bond Bond Cu terminal
IGBT STATIC PRESSURE
Bond
1 mmH2O = 0.0394 inch H2O
Diode 1 mmH2O = 9.8 Pa
1 mmH2O = 25.4 mm H2O
Chip Solder 1 Pa = 0.102 mm H2O
1 inch H2O = 249 Pa
Substrate
Base Plate Solder
Screw 1 Pa ≡ 1 N/m2 = 10-5 bar = 10-6 N/mm2 = 0.102 kp/m2 = 0.987×10-5 atm = 0.0075 Torr = 145.04×10−6 psi
Thermal Base plate
Grease
AIRFLOW
1 m3/min = 35.31 ft3/min (cfm)
1 cfm = 0.0283 m3/min
Heatsink 1 m3/min = 16.67 litre /s
1 cfm = 0.472 litre /s
1 litre /s = 0.06 m3/min

1 cfm = 1.7m³/h
1 litre/s = 3.6 m³/h
Figure 5.65. Power module with DCB substrate: (a) basic structure and (b) thermal model. 1 m³/s = 3600 m³/h
241 Power Electronics Chapter 5 Cooling of Power Switching Semiconductor Devices 242

Table 5.38 Heat pipe working fluids 5.3. Figure 5.68a shows the circuit diagram for a power current sink which utilises a 40V source.
Both the IGBTs T and wire wound resistors R are mounted on a common heat-sink, of thermal
Boiling point @ resistance Rθ hs-a = 1 K/W. The transistor has a thermal resistance of 2 K/W from the junction to
medium Melting point Useful range the heat-sink, and 10 K/W from the junction to air via the transistor casing exposed to the air.
atmospheric pressure
°C °C °C The resistor has a mounting thermal resistance from the insulated wire to the heat-sink of 1 K/W
Helium -271 -261 -271 to -269 and 10 K/W from the wire to the air via its casing exposed to the air. The maximum transistor
Nitrogen -210 -196 -203 to -160 junction temperature is 423 K, the maximum resistor wire temperature is 358 K and the ambient
Ammonia -78 -33 -60 to 100 air temperature is 303 K.
Pentane -130 28 -20 to 120
Acetone -95 57 0 to 120
Methanol -98 64 10 to 130
Flutec PP” -50 76 10 to 160
Ethanol -112 78 0 to 120
Heptane -90 98 0 to 130
Water 0 100 0 to 150
Toluene -95 110 30 to 200
Flutec PP9 -70 160 50 to 200
Thermex 12 257 0 to 225
Mercury -39 361 150 to 350
Caesium 29 670 250 to 650
Potassium 62 774 500 to 1000
Sodium 98 892 600 to 1200 Figure 5.68. Problem 5.3.
Lithium 179 1340 1000 to 1800
Silver 960 2212 1800 to 2300
Based on thermal considerations, what is the maximum current rating of the current sink and
under such conditions, what is the heat-sink temperature?
Problems What power rating would you suggest for the 1 Ohm current measurement resistor?
Are there any difficulties in operating the transistor in the linear region in this application if it is in
5.1. A thyristor bridge switches at 1 kHz and the total energy losses per thyristor are 0.01 Joule per a 120 W dissipation package which is derated according to figure 5.67b?
cycle. The thyristors have isolated studs and a thermal resistance of 2 K/W. The heat sink has a [1.36 A, 69°C, > 2 W]
thermal resistance of 1.8 K/W. Calculate the maximum number of thyristors that can be
mounted on one heat sink if the thyristor junction temperature is not to exceed 125°C in an 5.4. A power IGBT switches a 600 V, 25 A inductive load at 100 kHz with a 50 per cent on-time duty
ambient of 40°C. What is the heat sink temperature? cycle. Turn-on and turn-off both occur in 100 ns and the collector on-state voltage is to be 2 V.
[3 devices, Ts= 94°C] Calculate the total power losses, Pd, of the switch.
The switch has a thermal resistance Rθj-hs = 0.05 K/W, and the water-cooled heatsink provides a
5.2. A transistorised switch consists of two IGBTs and two 1 Ohm current-sharing resistors, as shown thermal resistance Rθhs-w = 0.05 K/W. Calculate the operating junction temperature if the water
in figure 5.67, mounted on a common heat-sink. Each transistor has a thermal resistance Rθj-hs for cooling is maintained at 35°C.
of 2 K/W, while each resistor has a thermal resistance Rθ r-hs of 1 K/W. The maximum switching The 25 A steady state load current is stepped to 200 A. Calculate the surge power dissipation
frequency is 1 kHz and the maximum duty cycle is 99.99 per cent. The heat-sink thermal Ps, at 200 A, assuming transistor switching and on-state characteristics remain unchanged.
resistance Rθ hs-a is 1 K/W. The energy losses per transistor are 5 mJ/A per cycle. If the ambient The junction temperature for a power surge during steady-state operation is given by case (e) in
Table 5.11.
temperature is 30°C, maximum allowable junction temperature is 150°C, and the maximum
With the aid of figure 5.10, determine the junction temperature at the end of a 0.1s, 200 A pulse.
allowable resistor internal temperature is 100°C, calculate the switch maximum current rating
How long is it before the junction temperature reaches Tl j = 125°C, with a collector current of
based on thermal considerations. What are the operating temperatures of the various
200 A?
components, assuming ideal current sharing?
(Assume Rθc-hs = 0).
[6.88 A, Tr = 100°C, Ths = 88°C, Tj = 122.5°C]
[175 W, 52.5°C, 1400 W, 112.6°C, 0.5 s]

5.5 Rework example 5.6 finding the case temperature when the switching losses equal the on-state
loss.

5.6 A 20kHz, step-down, 340V dc chopper feeds an inductive load with an average current of 20A
and a peak-to-peak ripple of 20A. Thus the MOSFET switch on-state current rise from 10A to
T T 30A while the freewheel diode current falls from 30A to 10A when the switch is off. The MOSFET
on-state resistance is 0.1Ω and has switch on and off times of 100ns and 200ns respectively.
The switch duty cycle is 75% and it has a thermal resistance Rθ j-c of 0.4K/W and is mounted on
a heatsink of thermal resistance Rθc-a of 0.6K/W in a maximum ambient temperature is 40°C.
Calculate:
i. switching losses, using equations 6.9 and 6.10
ii. switch on-state losses
iii. MOSFET junction operating temperature
[3.4W + 20.4W = 23.8W; Irms = 15.8A, 25W; Tj = 88.8°C]
Figure 5.67. Problem 5.2.
243 Power Electronics

1 Explain the difference between fans, blowers, and compressors.

Which fan is used for moving flows against relatively low pressures?
2
(a) radial fan (b) backward inclined fan (c) forward curved fan (d) axial fan
If efficiency is the main consideration, which fan is selected?
3
(a) radial fan (b) backward inclined fan (c) forward curved fan (d) axial fan
Which fan is ideally suited for dusty conditions?
4
(a) radial fan (b) backward inclined fan (c) forward curved fan (d) axial fan
System resistance refers to
5
(a) static pressure (b) velocity pressure (c) total pressure (d) differential pressure
System resistance varies as
6
(a) square of flow rate (b) cube of flow rate (c) proportional to the square root (d) directly with flow rate
The intersection of the system characteristic curve and the fan operating curve is called
7
(a) design point (b) operating point (c) free flow point (d) shut off point
Varying the RPM of a fan by 10% varies the pressure by
8
(a) 19% (b) 29% (c) 10% (d) no change
Varying the RPM of a fan by 10% varies the flow by
9
(a) 27% (b) 20% (c) 30% (d) no change
Varying the RPM of a fan by 10% varies the power by
10
(a) 27% (b) 37% (c) 10% (d) no change
11 Explain the factors which change the system resistance.

12 What are the affinity laws as applied to centrifugal fans?