Problem 2: Heat Spreader

Practical Training: Microsystem Design

Group 2

Authors
CHAITRA BASURU SHANKARALINGAPPA (567523)
SNEHITH MENASAMAKKI KUMAR (570585)
MUHAMMAD USMAN RAFIQUE (533878)
MUHAMMAD SHAMOOKH (539375)
HARSHITA VISWANATH (565219)
TAHA AHMED (527985)

Lab Instructor
KONSTANZE SCHOBER
 CONTENTS

LIST OF TABLES i

LIST OF FIGURES i

HEAT SPREADER 1
Introduction1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Approaches For Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
a) Analytical Calculation of Rth , ∆T and Tmax 45° Model1 . . . 2
b) Analytical Calculation of Rth , ∆T and Tmax using D.P. Kennedy
Model2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
c) Comparison of results from analytical models with simula-
tion.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
d) Temperature Profiling4 . . . . . . . . . . . . . . . . . . . . . 7
e) Analysis of the model with change in dimensions of the heat
source5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

CONCLUSION6 13

TASK DISTRIBUTION 14

REFERENCES

1
Muhammad Usman Rafique
2
Muhammad Shamookh
3
Snehith Menasamakki Kumar
4
Taha Ahmed
5
Chaitra Basuru Shankaralingappa
6
Harshitha Viswanath
 LIST OF TABLES

1 Final D.P. Kennedy model values . . . . . . . . . . . . . . . . . . . 5

2 Comparison of values from Analytical and Simulated Models . . . . 7
3 Comparing results from part (e) with part (c) and (d) . . . . . . . . 12
4 Partition of Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

LIST OF FIGURES

1 Heat Spreader . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2 Half Heat Spreader . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
3 Spreading resistance factor H2 according to the Kennedy model . . 4
4 Line heat source properties . . . . . . . . . . . . . . . . . . . . . . . 5
5 Thermal insulation and bottom surface Temp. of substrate . . . . . 6
6 Mesh for the model . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
7 (a) 2D Temperature contour model (b) Max. and Min. Temperature 7
8 (a) Cut line of substrate for radius of heater’s bottom surface (b)
Temperature (K) vs. Height (mm) Graph at r = 0 . . . . . . . . . . 8
9 Maximum and Minimum Substrate Temperature (K) . . . . . . . . 8
10 (a) Cut line of substrate for radius of heater’s bottom surface (b)
Temperature (K) vs. Radius of heater’s bottom surface (mm) . . . 9
11 Parameters of heat source . . . . . . . . . . . . . . . . . . . . . . . 10
12 Mesh Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
13 (a) Maximum heat source and Minimum Substrate Temperature
(K) (b) Maximum and Minimum temperature . . . . . . . . . . . . 11
14 Temperature (K) vs. Radius of heater’s bottom surface (mm) . . . 11

i
 HEAT SPREADER

Introduction1
Head Spreader is a device which transfer heat from a hotter surface to a colder
surface. The difference between heat sink and header spreader is, heat sink has
fins on it body to increase the area to sink the heat while in heat spread we
have big single colder block attached with smaller heated block. The application
can be seen in computer hardware when we apply heat sink or heat spreader
on microprocessor. A heat spreader can be modeled using the heat conduction
equation, which describes how heat flows through a material[1]:

qs = −kA(dT /dx)

where q is the heat flow rate (in watts), k is the thermal conductivity of the
material (in watts per meter per Kelvin), A is the cross-sectional area of the
material (in square meters), dT/dx is the temperature gradient across the material
(in Kelvin per meter), and the negative sign indicates that heat flows from higher
temperatures to lower temperatures.

Assuming that the heat is generated uniformly by a heat source with a heat flux
qs (in watts per square meter), the temperature at the center of the heat spreader
can be calculated using the following formula[1]:

Tc = (qs ∗ L)/(λ ∗ W )

where Tc is the center temperature of the heat spreader (in Kelvin). This formula
shows that the center temperature of the heat spreader is directly proportional to
the heat flux and the thickness of the heat spreader, and inversely proportional
to the thermal conductivity and the width of the heat spreader. This means
that increasing the thickness of the heat spreader or using a material with higher
thermal conductivity can help to reduce the center temperature and improve the
efficiency of heat dissipation.
1
Muhammad Usman Rafique

1
Answers to Questions

a) Analytical Calculation of Rth , ∆T and Tmax 45° Model1

Consider an axissymmertric problem with a heater of radius a = 1cm and power
P = 30 W on a ceramic substrate of radius b = 5cm and thickness w = 2.5mm.
the ceramic substrate has a thermal conductivity of λ = 33 W/(m K), a density
of ρ = 3.9 × 103 kg/m3 , a heat capacity of cp = 850 J/(kgK) and the ambient
temperature is T = 20 ◦ C.

Figure 1: Heat Spreader

Due to axis-symmetry, taking half part for better computation speed on COMSOL

Figure 2: Half Heat Spreader

Since the substrate is circular so the area can be calculated as A = π ∗ r2 , and the
model is 45° heat spreader, the equation can be changed to A = π ∗ (a + w)2 with
tan(45◦ ) = 1. Then the thermal resistance is given as,

dw
dR =
λ×A
Z Rth Z 2.5×10− 3
dR 1
dw = dw
0 dw 0 λ ∗ π(a + w)2
2.5∗10−3
(a + w)−2+1

Rth =
(−2 + 1) ∗ λ ∗ π 0
1
Muhammad Usman Rafique

2
  2.5∗10−3
−1
Rth =
λ ∗ π ∗ (a + w) 0

Applying limits and putting values

 
1 −1 1
Rth = +
λ 33 ∗ (2.5 ∗ 10−3 + 10 ∗ 10−3 ) 33 ∗ 10 ∗ 10−3

K
Rth = 0.201
W
Since,
∆T
Rth = (1)
P
∆T = 0.201 ∗ 30 = 6.03K

Tmax − Tamb = 6.03K

Tmax = 299.03K (Tamb = 293K)

b) Analytical Calculation of Rth , ∆T and Tmax using D.P.

Kennedy Model2
The D.P. Kennedy model is a mathematical model to describe the thermal behav-
ior of heat spreaders. The D.P. Kennedy model is based on the assumption that
the heat spreader is treated as a two-dimensional plate and the model assumes
the bottom surface to be at constant temperature. All other surfaces are ther-
mally insulated and the heater itself is assumed to be infinitely thin. Using these
assumptions, the D.P. Kennedy model provides a mathematical expression for the
thermal resistance (Rth ).
1
Rth = H2 (2)
πaλ
where,
a = radius of the heater
λ = thermal conductivity (W/mK)
H2 = heat spreading factor

The heat spreading factor (H2 ) describes how effectively a heat spreader distributes
heat away from a heat source. It is the measure of temperature drop across the
heat spreader (w/b) to the temperature drop across the interface between the heat
spreader and the heat source (a/b). In our case, H2 has been calculated from the
graph[2]. The graph is divided into three lines where the curve lines represent
2
Muhammad Shamookh

3
the w/b ratio, straight vertical lines represent a/b and straight horizontal lines
represent H2 . Therefore, as per values from the task a/b and w/b is calculated
as 0.2 and 0.05 respectively and the position of H2 is at the intersection point.
From figure 3 H2 is calculated as 0.23. After extracting H2 from the graph, using

Figure 3: Spreading resistance factor H2 according to the Kennedy model

equation 2 we calculated Rth

1 1
Rth = H2 = (0.23)
πaλ π(0.001)(33)

Rth = 0.23K/W

As, ∆T is dependent on thermal resistance and power of heat source i.e. P = 30W.
This is given by equation 3.
∆T
Rth = (3)
P
∆T = 6.66K

Later, the maximum temperature (Tmax ) is calculated by taking the difference

between the two extreme points with Tamb = 293K

∆T = Tmax − Tamb

Tmax = 299.83K

The final results using D.P. Kennedy model are grouped in table 1.

4
Table 1: Final D.P. Kennedy model values

Rth [K/W] ∆T [K] Tmax [K]

0.22 6.66 299.83

It can be concluded, numerical analysis gives the analytical intuition into the
problem and later on simulations are performed for more accurate results where
the plausibility of the simulated result is understood comparing with numerical
analysis in the later section.

c) Comparison of results from analytical models with sim-

ulation.3
In the previous section we used normal analytical and Kennedy Model to calculate
the temperature parameters and profiles of a heat spreader. In this part we exam-
ine the steady state analysis of given configuration of heater and ceramic substrate
by using heat transfer in solids using COMSOL. The results of simulation and an-
alytical calculation are compared. Here we consider a thin line heat source with
a given radius of 10 mm and a power of 30 W (applied on the ceramic substrate)
and heat flux boundary is placed from (0,2.5) to (10,2.5) . The considered ceramic
substrate are given properties mentioned above after introduction.

Figure 4: Line heat source properties

The Figure 4 shows the line heat source as the heat source of radius 1cm situated
on top of the substrate with coordinates (0,2.5) to (10,2.5). The substrate chosen
here is a basic ceramic material with mentioned parameters in the task.

As given in the task all surfaces are thermally isolated from the surroundings
except the bottom surface shown in figure 5 where, bottom surface temperature
of ceramic substrate is kept at a constant of 293.15K.

We have used “Extremely fine mesh” (collection of more data points) for evaluating
the results of temperature profiles for obtaining the best output but its a trade-off
3
Snehith Menasamakki Kumar

5
 Figure 5: Thermal insulation and bottom surface Temp. of substrate

between computational speed and accuracy. The max. and min. element size are:

• Maximum element size = 0.5 mm

• Minimum element size = 0.001 mm

Figure 6: Mesh for the model

Thermal resistance can be calculated using equation 1 by examining Tmax =

300.348K and Tamb = 293.150K from figure 7(a).

[300.348 − 293.150] K
Rth = = 0.2399
30 W

As mentioned earlier both analytical heat spreading using 45° method and Kennedy
method using heat spreading factor are used to obtain the maximum temperature,
the temperature difference and the equivalent thermal resistance over the ceramic
substrate . With the given parameters we simulated in COMSOL to compare both
methods and their results, we conclude that rather than 45° analytical method
Kennedy model uses heat spreading factor and gives more precise results with
respect to COMSOL results.

6
Table 2: Comparison of values from Analytical and Simulated Models

Methods Tmax △T Rth

45° model 299.03 K 6.03 K 0.201 K/W
D.P. Kennedy 299.83 K 6.66 K 0.23 K/W
model
Simulation Model 300.348 K 7.198 K 0.2399 K/W

Results

Figure 7: (a) 2D Temperature contour model (b) Max. and Min. Temperature

d) Temperature Profiling4
In this task, we are required to analyze two line graphs: one showing the relation
between temperature and height at radius r = 0, while the other depicting the
relation between temperature and radius at the bottom surface of the heater.
These graphs are under consideration as in the context of the model of the previous
task.

Figure 8(b) shows the linear relationship between the temperature and height at
r = 0 where cut line is considered at the edge where r = 0 shown in figure 8(a),
the thickness of the substrate i.e., w=2.5mm.

The Figure 8(b) shows the linear change from the bottom of the substrate to
the top of the substrate where heater is placed, so it is clearly observed that
temperature goes up as the height increases (linear relation). In other words,
as the distance from the heater decreases, temperature of the substrate increases.
Considering Figures 8(a) and (b) , the temperature varies from 293.15 K to 300.34
4
Taha Ahmed

7
 Figure 8: (a) Cut line of substrate for radius of heater’s bottom surface (b)
Temperature (K) vs. Height (mm) Graph at r = 0

K from bottom to top of the substrate. This analysis can be compared with the
contour image below figure 9. The contour shows the linear behavior by defining
different color as the maximum and minimum temperature from top to bottom
respectively.

Figure 9: Maximum and Minimum Substrate Temperature (K)

The second graph, illustrated in Figure 10(b) shows the relation between the
temperature and radius of the bottom surface of the heater i.e., top surface of the
substrate that touch the heater and has a radius of a = 1 cm. It is marked with
red line in Figure 10(a), which is the half-cut section of the substrate.

The curved line depicted in the Figure 10(b), illustrates the exponential function,
which corresponds to the behavior of the temperature, it is also observed that,
at the center of the heater the temperature reached its maximum value i.e., 300.

8
 Figure 10: (a) Cut line of substrate for radius of heater’s bottom surface (b)
Temperature (K) vs. Radius of heater’s bottom surface (mm)

34K and decreases exponentially as it heads toward the outer side of the heater,
in this case until reaching the lowest temperature value of 293.15 K.

e) Analysis of the model with change in dimensions of the

heat source5
Substrate and Source geometry: As per task (c), the heat source is con-
sidered as a line segment but here we modify it to rectangular heat source with
thickness of 0.5mm which is made of Silicon with thermal conductivity 130W/mK
is changed to 150W/mK. It is placed above the rectangular substrate with the
same dimensions and material i.e., ceramic (nitrites or oxides) and here we have
used Silicon Nitride material for Substrate as in task (c).
Note: Uniform Power density does not imply same temperature throughout heat
source.

The heat source is provided with power of 30 W from which we calculate heat rate
which shows power distribution around the volume of the source as,

P ower Po
Qo = = (W/m3 )
V olume V

Change in maximum and minimum temperature location occurs with change in

dimensions. Temperature (max. and min.) is irrespective of bending/shapes on
5
Chaitra Basuru Shankaralingappa

9
the surface.

Figure 11: Parameters of heat source

Mesh type: As we vary element size which indicates mesh type, the data points
adding up the values to get better results can be seen. Coarse mesh type has less
data points selected compared to fine or finer mesh types. The mesh type and
maximum and minimum element sizes are kept same as before in order to get
an equal data point comparison which is more accurate. The element size with
minimum and maximum size are specified as:

• Element Size: Extremely fine

• Maximum element size = 0.5 mm
• Minimum element size = 0.001 mm

Figure 12: Mesh Model

Max/Min Surface Temperature: Top of the Silicon heat source has maxi-
mum temperature of 300.209K. The results are shown in figure 13 (a) and (b)

Line graph: The Cut Line for 2D graph is considered at same coordinates as
in figure 10(a) which highlights the surface which we are referring for observing
temperature change in the Line graph.

As per figure 14, the temperature is not linear as we move along the length from
the top to the bottom of the heat source.
Now, Temperature difference can be calculated as,

△T = Tmax − Tmin

10
 Figure 13: (a) Maximum heat source and Minimum Substrate Temperature
(K) (b) Maximum and Minimum temperature

Figure 14: Temperature (K) vs. Radius of heater’s bottom surface (mm)

where from figure 13

Tmax = 300.209K ; Tmin = 293.15K

=⇒ ∆T = 300.209 − 293.15 = 7.059K

Using equation 1 for thermal resistance with ∆T = 7.059K & P = 30W gives

K
Rth = 0.235
W

The comparison between simulated tasks is displayed in table 3

11
Table 3: Comparing results from part (e) with part (c) and (d)

Parameters part (c) and (d) part (e)

Temperature on the top 300.348 K 300.209K
surface of the heat
source (maximum
temperature)
Thermal resistance 0.239 K/W 0.235 K/W
line graph temperature varies temperature shows
non-linearly as we move non-linear variation as
along the length from we move along the
the top of the heat length from the top to
source to the bottom of the bottom of the heat
the heater source. The intensity of
heat is maximum at the
top as material itself has
a certain heat spreading
coefficient which will
distribute the heat
among itself and
therefore the
neighbouring mesh data
points at the top will
get the maximum heat.
As material always
distributes the heat
equally in all portions,
we see that as moving
along the length, the
effect of power gets
limited taking over an
ambient temperature

12
 CONCLUSION6

The task brings out a comparison between different models to describe the thermal
behavior of heat spreaders, and outlines the changes in result parameters with
alteration of heater dimensions.

In task (a), we calculated the temperature parameters analytically using the 45°
heat spreading model. In task (b), we calculated the same parameters as before
using D.P. Kennedy model. This improves the accuracy of the result since it is
based on the assumption that the heat spreader is a 2D-plate and the bottom
surface is at a constant temperature. Overall, D.P Kennedy model is better than
the 45° heat spreading model for calculation of temperature parameters analyti-
cally. One of the prominent reason could be it is difficult to build the required
pyramidal model than more easy model as in D.P. Kennedy. Nonetheless, the
spherical enclosure equation can be good option because it can provide a better
estimate of the thermal resistance but for now, we will prefer D.P. Kennedy model
for analytical calculations

In task (c), we draw out a clear contrast between the analytical models and simu-
lation model using COMSOL. The results of simulation are approximately equal
to those values computed analytically using D.P. Kennedy model. Temperature
profiling is the focus of task (d) and task (e). When the radius is considered as r
= 0 with specified value of height , the nature of heat dissipation is linear because
the surface area of heat source is more than that of the substrate. Later, when a
non-zero value of radius is considered the heat propagation is non-linear since more
ceramic substrate is available. We have considered extremely fine mesh which im-
proves the accuracy of results. In the final task, the thickness of the heater is
considered to be 0.5 mm at a uniform power density and the results are compared
with that of task (c).

For COMSOL models, we would prefer to use the rectangular heat source model
because the power is distributed in the whole volume which gives more realistic
picture for temperature distribution and thermal resistance.

6
HARSHITHA VISWANATH

13
 TASK DISTRIBUTION

Table 4: Partition of Tasks

Name Task Completed

MUHAMMAD USMAN RAFIQUE Introduction, Task (b)
MUHAMMAD SHAMOOKH Task (a)
SNEHITH MENASAMAKKI Task (c)
KUMAR
TAHA AHMED Task (d)
CHAITRA BASURU Task (e)
SHANKARALINGAPPA
HARSHITHA VISWANATH Conclusion

14
 REFERENCES

[1] Latif M Jiji and Latif Menashi Jiji. Heat convection. Springer, 2006.

[2] Konstanze Schober. Problem 2: Heat Spreader. Microsystem Design, TUHH,

2023.