Understanding Thermal Analysis of RF Devices Application Note
Understanding Thermal Analysis of RF Devices Application Note
Introduction
Thermal design and analysis are critical to improving component reliability. This document provides information related to thermal
analysis and functionality for Qorvo® products and their applications. It outlines the basics of thermal design, such as resistive
analysis shown in Figure 1, for Qorvo components in customer applications and describes details to assist engineers with their
system-level design.
Figure 1. Thermal resistance used to determine the junction or channel temperature of a device.
From an environmental perspective, temperature is the biggest reason most semiconductors fail in their field application, as shown in
Figure 2. This can be due to a harsh operating environment without adequate protections built in at the system level, which in turn can
result from an incomplete understanding of the thermal aspects of semiconductor devices, packaging, layout and heat sinking.
In addition, thermal measurements on RF devices at design verification can be a bit tricky.
In this document we will break down the many aspects of temperature and design for RF semiconductor product applications.
The goal is to help you design easier, faster and more reliably, so your systems do not break down due to thermal issues.
Table of Contents
1. Introduction
2. Table of Contents
3. Reference Documents
4. The Basics of Semiconductor Thermal Measurement
5. Thermal Resistance
6. Determining Junction or Channel Temperature
7. A Review of Application Material Properties
8. Thermal Vias in an Application
9. Thermal Interfaces
10. Thermal Measurement Methods for an Application
11. High-Power GaN Application Considerations
a. Thermal Measurement
b. Improving Thermal Conductivity in Applications
i. Die Only Applications
ii. QFN and Surface Mount Packaged Applications
iii. Copper Plated (CP) and Flange Packaged Applications
12. System Level Thermal Analysis
13. In-Closing
14. Definitions
15. Additional Information
16. Contact Information
Referenced Documents
The reference documents below take precedence over the contents of this application note and should always be consulted for the
latest information.
Application Note: GaN Device Channel Temperature, Thermal Resistance and Reliability Estimates:
https://www.qorvo.com/products/d/da006480
Video: Understanding GaN Thermal Analysis:
https://www.qorvo.com/design-hub/videos/understanding-gan-thermal-analysis
Application Note: GaN Thermal Analysis for High-Performance Systems:
https://www.qorvo.com/resources/d/qorvo-gan-thermal-analysis-for-high-performance-systems-white-paper
Application Note: High-Performance GaN Thermal Evaluation – Limitations of Infrared Microscopy:
https://www.qorvo.com/resources/d/qorvo-high-performance-gan-thermal-evaluation-infrared-microscopy-limitations
Application Note: Application of Arctic Silver 5 Thermal Compound and Indium Shims for Qorvo CP-style Packaged
Components: https://www.qorvo.com/products/d/da006842
White Paper: GaAs and GaN Die Assembly and Handling Procedure:
https://www.qorvo.com/resources/d/qorvo-gaas-gan-die-assembly-handling-procedures-white-paper
White Paper: CW Operation of QFN-Packaged Pulsed GaN Power Amplifiers:
https://www.qorvo.com/resources/d/cw-operation-of-qfn-packaged-pulsed-gan-power-amplifiers-white-paper
Application Note: Bulk Thermal Models:
Available upon request from applications engineering.
Note, all three of these above methods in Figure 3 measure temperature without touching the surface of the semiconductor. IR imaging
simply measures the thermal radiation coming off the surface of the semiconductor. Thermoreflectance provides its own radiation at the
surface and then detects it in addition to the surface’s natural radiation (which then gets subtracted out). Micro-Raman is a more
expensive and time-consuming measurement technique, but it is the most accurate because it measures the temperature within the
epitaxial layer with extreme resolution.
Thermal Resistance
Thermal resistance is the measurement of a material’s ability to resist heat flow. In general, when talking about semiconductors, we
would like our semiconductor material, bonding material, PC board material, etc., to have a very low thermal resistance level, so heat
flows freely to the ambient air. So, in semiconductor applications, low thermal resistance (Rth) is our goal. Figure 4 illustrates some of
the formulas associated with heat transfer and thermal measurement.
• Thermal conductivity – the ability of a given material to conduct/transfer heat.
• Thermal resistivity – the ability of a given material to resist the conduction/transfer of heat.
• Thermal resistance – is like resistivity but takes the shape and size of the material into account.
• Thermal impedance – is like resistance but does not factor in the footprint size of the material and often includes interfacial
resistances to provide an effective value for a layer within a stack.
Figure 4. Physics of heat conduction – thermal resistance formulas and application calculation.
The analogy of an electrical resistor network in Figure 4 illustrates the thermal resistance of a packaged semiconductor device.
This represents the electrical resistor equivalent circuit – starting from the heat source (transistor junction or channel) through the two
possible pathways the heat will transfer. In this electrical resistor model, the electrical resistance is defined by the potential difference
(voltage) across the resistor divided by the current through that resistor. In thermal resistance, the thermal potential difference
(temperature) divided by the thermal current (heat flow) through the thermal resistor defines the thermal resistance Rth.
In the first pathway, heat transfers from the transistor junction or channel through the mold compound by conduction, and then to
the air surrounding the device by convection. In the second pathway (dominant path), which is parallel with the first, heat flows
from the junction or channel of the device through the lead, through the PC board, into the chassis by conduction and finally to the air
surrounding it by convection. The second path is the primary focus of calculating the junction or channel temperature since the majority
of the heat generated in the device transports through this pathway. This pathway is also the one we will concentrate on in detail in this
application note.
Figure 5. IR Imaging versus micro-Raman thermal measurement (Note: micro-Raman measures below the material surface).
Note most companies use IR imagery to make this Tj or Tch determination, as it is the easiest to use, cost-effective and most
accessible, but it provides a lower measurement value due to averaging across the surface region above the channel.
Let’s review some basics of measuring an application using a molded semiconductor on a PC board. As shown in Figure 6, there are
several layers and measurement areas to consider when looking at thermal measurements of your application. The first thing to
consider is the heat path. In a package as shown in Figure 6, the heat path for this type of design must go through the bottom of the
package through the ground pad. Therefore, the application design should incorporate a PC board that pulls the heat from the bottom
of the package through the device ground slug. Note, in this type of package, attempting to draw heat from the top of the package will
not work due to the poor thermal conductivity of the mold compound; trying to cool from the top will cause device degradation/failure
due to high junction/channel temperature. The primary heat paths in semiconductors require heat to be pulled away from the device
through the bottom of the die in the most effective way possible (unless the manufacturer notes otherwise). Therefore, in this scenario,
pulling the heat down through the ground pad represents the lowest resistance to heat flow from the device junction or channel to
the outside world.
Figure 6. This overmolded part shows heat flows from the top of the die down through the package ground pad.
Figure 7 illustrates the layers of an application using a packaged part. Note, when designing a system or final product, it is important to
use as much information and material from the semiconductor vendor as possible, such as S2P parameters, PC board Gerber files,
etc., so the proper thermal heat extraction methods are employed. For example, placement of via holes and creation of these vias are
very important to extract heat from your device and increase part reliability – these types of instructions and placements of via holes are
generally located in the PC board Gerber files offered by the semiconductor vendor.
As you can see in Figures 6 and 7, the semiconductor device is located above the ground pad of the device using a die attach material.
Between the ground slug and the PC board will be the solder attach, which should be a high-grade material effective for optimizing
thermal dissipation. The via holes’ location, size, and style are typically spelled out by the vendor for optimized performance and should
be used with vendor direction. The placement, style, and type of via holes are important. We will dig deeper into this later in this
application note.
Figure 8 below shows the electrical Rth equivalent circuit for a semiconductor device application shown in Figure 7. As seen in
Figure 8, the largest portion of the thermal path goes through the devices, die attach and package ground pad into the PC board
vias through solder. It is critical the vias are precisely made to ensure a low resistance pathway for the heat. The amount of variation
between PC board materials and design can be significant (see Table 1); therefore, airflow, board size, board thickness, via hole
location and type can all affect the thermal resistance of the PC board. Moreover, the solder attachment material and methodology
are also critical here. When using a heat sink as shown in the image, attaching the heat sink to the lowest impedance path to the IC
junction or channel is best.
The k-value of a material determines how fast heat can spread through it. A high k-value means better heat conduction. For metal, the
k-value usually increases along with the electrical conductivity. Many dielectric materials have a high k-value, as shown in the tables in
Figure 9. Moreover, a material’s k-value has some dependence on temperature and may be anisotropic or direction dependent.
Although copper is a great conductor and a common ingredient in electronics packaging, not all copper is the same. As shown in
Table 1, sometimes the process used to fabricate the copper will change the k-value. Generally speaking, copper is affected by both
the alloy and process techniques. For example, copper lead-frames for over-molded QFNs use an alloy with a k-value 35% lower than
pure copper (copper-110), while the plated copper in PC boards will have a k-value that is around 15% lower than a rolled copper foil
that starts out on a board.
Additionally, the PC boards’ thermal resistance is proportional to the overall thickness of the board – and thermal vias also play a part in
this resistance value. Thus, a very thin PC board like 8-mils is often used for high-power GaN in a QFN package to keep the thermal
resistance as low as possible.
The thermal performance of a PC board material is entirely dependent on the way the copper is laid out. To move heat downward to
where a heatsink is attached you will need to use thermal vias at a minimum. These are drilled holes plated with copper to provide
thermally conductive pathways from one copper layer to the next. The thicker the copper, the better, and it does provide a great
conductor for heat to dissipate, but the more copper used, the higher the cost. So, both system designers and PC board manufacturers
must consider tradeoffs regarding the amount of copper used.
For example, for multi-layer PC boards built on FR4, common sizes for drilled and copper-plated thermal vias are 10-mil and 12-mil
diameter. For Rogers’ materials, thin two-layer boards and 8-mil thermal vias are common. See above in Table 1.
Another consideration in PC board layout and materials is the layer count. Many applications today have introduced several layers of a
PC board to route all the complex connections between circuit components and features. But those additional layers also provide a
benefit when addressing heat. In lower-power applications adding layers to the PC board has a big impact on heat conduction away
from components. The difference between a two-layer and a four-layer board can be as much as 20°C depending on how well the
copper planes are laid out for heat spreading. The four layers help to move the heat as shown in Figure 10. The various dielectric and
copper layers impact the PC boards’ thermal conductance below the device. The copper layers will dominate the heat flow because the
copper k-value is high. In the FR4 layers, the k-value is low, so it is best to use the vias with copper plating to provide the thermal path
from one copper layer to the next. But ultimately from a thermal low-power perspective, the more PC board layers, the better.
Although adding layers of PC board is a benefit in a low-power situation, doing so in high-power applications has the opposite effect.
Adding PC board layers in applications such as GaN high-power devices, especially above 10 to 15 Watts of power, increases
resistance and interferes with the heat path. Instead, in applications where a heat path must be optimum, like in high-power dissipation
scenarios –a thin PC board is recommended rather than a multi-layered PC board. We will review this further in the GaN applications
section in this application note.
Using these vias, the thermal resistance can be significantly lowered. The placement, size and number of vias used in creating a
thermal via array within the die pad area are critical. Maintaining a copper ground-plane on the top or bottom copper layer with as few
breaks as possible to create a heat spreader on the PC board is recommended. In multi-layer PCBs, successive vias between copper
layers should be aligned with the vias running between the top two metal layers (don’t stagger them – make sure the heat can travel
directly downward wherever possible).
Ultimately to determine these variables, it is recommended to take thermal measurements and perform thermal modeling to accurately
create the most optimum number, size or type of via needed in your desired application. Ideally, it would be best to have solid vias and
as many as possible underneath the die pad, so the maximum amount of heat is transferred, but this is not cost-effective nor practical.
So, the best practice is to use the guidance of the semiconductor vendor as they will be able to provide optimized PC board files
showing the location, type and number of thermal vias that should be used.
In regard to via holes used to remove heat from the device, thicker copper plating means a higher effective k-value of the total via
cross-section and better heat transfer. In Figure 11, kEFF or k effective, k refers to the effective thermal conductivity. The effective
overall kEFF value for the via cross sections increases as more copper plating is used. For example, kEFF = 282 W/(mK) for, a .010”
diameter via with two-ounce plated copper, but kEFF = 91 when half-ounce copper is used. PCB designers must make tradeoffs on the
cost vs. the effectiveness of thermal vias, but more copper is always better.
Some designs such as high-power GaN require via holes to be solid copper rather than relying on solder filled. Based on our internal
analysis we recommend customers use a copper-plated solid via but do not advise fully plated vias using solder, as it may lead to
voiding at the package interface. Also, for GaN operating above 10 or 15 W of power we recommend a thermal analysis, which may
show the need for the device to be mounted using a copper coin under the package or die. We delve deeper into this topic in the GaN
section of this document.
Another consideration in thermal vias is breaks in the thermal path. Effective heat spreading across the PC board improves the overall
heat transfer. Adding traces too close to or perpendicular to the heat flow path near the source will hamper heat spreading and create
hot spots. If traces are unavoidable due to design, attempt to make them run parallel to the predominant heat transfer path rather than
across it.
Thermal Interfaces
Generally, when putting two surfaces together, like a ground slug of the underside of a QFN package and a PC board, there is an
expectation that the two solid metal bodies have smooth surfaces that bond well. However, this is far from reality as these surfaces
have air gaps at the microscopic level. When two solids meet, the actual surface area touching is relatively small, which is the opposite
of what one may think.
Heat will flow across and between the air gaps, but it will not flow as easily without the help of a thermal conductor such as thermal
grease or other interface material. There are two main types of thermal interfaces:(1) bonded thermal interface (e.g., solder joints,
sintering metal pastes, and epoxies), (2), compressible thermal interface materials.
Solders and sintering metal pastes form metallurgical bonds and are thermally the highest-performing interfaces. Epoxies are usually
doped with thermally conductive particles like silver to improve their thermal performance. Solder joints have the advantage of being
reworkable, while the sintering metal pastes and epoxies are not re-workable once hardened.
For packages such as QFNs that are normally soldered down, it is a good practice to minimize solder joint voids between the exposed
pads and interconnections. There is a tradeoff between solder paste coverage and having the package “float” or “skate” on the PC
board connection. Typically, the preferred solder paste coverage is 50% to 80%. On the ground slug, 80% coverage is preferred.
Compressive thermal interface materials (TIMs) are generally lower-performing and place structural requirements on components,
PCBs, and heat sinks. Examples include thermal greases and graphite shims which must be held under compression to work
effectively, meaning the two parts in contact must be stiff enough to handle the stress. Using thermal grease (e.g., Dow Corning 340)
does not require mounting holes on the cold plate, because there is enough adhesion due to the grease layer surface tension. They are
popular because they are easily re-workable. While these TIMs may have performance comparable to an epoxy-bonded interface, they
can’t approach the thermal performance of a good solder joint. Additionally, graphene film is also good for a TIM as long as there is
sufficient pressure to obtain an optimal heat transfer.
Thermocouple measurements are one of the most commonly used sensing methods in engineering labs. (See Figure 12)
Thermocouple measurements are simple and consist of two wires of dissimilar metals joined together at one end. The other end,
where the wires are not joined, is connected to a multimeter or thermocouple meter. The junction of the two dissimilar metals
(e.g., copper and constantan), produces a thermoelectric effect which gives a constant potential difference of only a few millivolts
between them. Since the thermocouple is a differential device rather than measuring an absolute temperature, the reference
junction temperature must be known.
As shown above, in Figure 12, thermocouples can be made from a variety of different materials, enabling measurement of extreme
temperatures from -200°C to over 2000°C – the examples shown are the most commonly used in electronics. When using the
thermocouple to measure temperature in a PCB with a device mounted on it, one should make certain to get as close as possible
to the ground slug located under the packaged part, so the measured value is accurate to determine the true Rth.
If micro-Raman thermography and FEA modeling are not an option and only an infrared (IR) camera is available, the accuracy
limitations of IR imaging must be clearly understood. IR cameras have a spatial resolution that is an order of magnitude larger than the
gate length of a FET channel and yield area-averaged surface temperatures which are much lower than the true maximum channel
temperature. It is recommended to work closely with the GaN device applications teams to ensure the GaN device is operating cool
enough in your application. The applications team can also provide a product bulk thermal model to drop into your system-level thermal
model to estimate the device operating environment more accurately in your system-level model and determine the resulting junction or
channel temperature.
Gold-tin solder and many thermally conductive epoxies have a low thermal resistance value and the ability to accommodate coefficient
of thermal expansion (CTE) mismatch stress (often seen when mounting GaN die to high thermal conductivity materials). It is important
to ensure the gold-tin solder joint is void-free, especially under the active area of the die. If using a conductive epoxy, the attachment
must be uniform and void-free, and have a thin bond line (to maximize thermal conductance).
Mounting a GaN die power device directly onto a PCB is not recommended unless it is mounted to a high-thermal-conductivity coin
(e.g., a copper coin) to ensure adequate heat transfer.
Figure 14. Example of copper coin structure under GaN device and via versus coined comparison.
One thing to note with TIMs is that sufficient pressure is required for obtaining good heat transfer. When using a thermal grease
or a thermal compound, minimum of 80% coverage is required. Refer to Figure 15 for recommended use of an indium shim
or thermal compound.
Figure 15. Example of indium shim and thermal compound under packaged GaN devices.
Semiconductor reliability is partly defined by estimating a device’s maximum junction or channel temperature to determine a projected
lifetime. These values are gathered by measuring and modeling thermal resistance, power dissipation and heat transfer. In an
application, the maximum junction or channel temperature is a key metric and depending on a single method is not good practice;
thermocouple/thermistor measurements, modeling, IR imagery, etc., should all be employed. In a package such as the one shown in
Figure 16, one should use a model to determine the maximum case temperature under the device for the purpose of estimating the
junction temperature based off package thermal resistance. Note – for GaN components refer to the application note GaN Device
Channel Temperature, Thermal Resistance and Reliability Estimates.
Another nice-to-have in system level thermal analysis is having a bulk thermal model of the semiconductor component. Having this
model from a semiconductor supplier will help customers mimic the output heat flux of the component helping customers estimates of
maximum backside temperature of the product and what it will experience when operated in the customer system design. In addition,
the bulk thermal model provides greater insight into how neighboring devices are impacted by the waste heat from Qorvo’s product.
The junction or channel-to-base temperature, as shown in Figure 17, uses a thermocouple to provide the temperature of the package
base. While this is possible, care should be taken to ensure that the thermocouple is reading the proper region. As shown in Figure 16,
the measurement should be taken at or as close to the underside of the package ground pad as possible. If the measurement is done
at the pad location, it is recommended that more analysis be done, such as modeling the system using a product bulk thermal model.
Once this case temperature is known, the next step is to calculate the power dissipation. Qorvo has an online calculator to make this
easy. (Qorvo online PAE/Pdiss/Tj calculator – see Figure 18). The online calculator can also be used to calculate the power dissipation
and the maximum junction temperature of a component. Link to calculator.
As noted earlier, there is a Qorvo video showing how to use this calculator with GaN components. The link to that video is here,
Understanding GaN Thermal Analysis. For general applications, let’s review a case study to assist in a better understanding of how this
calculator can be used.
In-Closing
Understanding and determining the thermal performance of both semiconductor devices and system designs are critical to ensuring
your product functions optimally, reliably and without problems. Although this application note does not replace the method of finite
element analysis and the software tools used in such measurements and analysis, it does provide guidance in thermal analysis design.
Using the outlined methods and tools will aid system engineers with design considerations and thermal measurements.
Definitions
• Junction or Channel Temperature – The highest operating temperature of the actual semiconductor in an electronic device.
• Absolute Maximum Junction or Channel Temperature – The temperature beyond which damage occurs to the
semiconductor device after a specific amount of time. The device may not function or meet the expected performance at this
temperature.
• Absolute Maximum Operating Temperatures – The maximum environment temperature where the device is recommended
to operate. Operating above this temperature can damage or reduce the lifetime of the device. This is sometimes referred to
as the case temperature.
• Thermal Resistivity – The ability of a given material to resist the conduction/transfer of heat.
• Thermal Resistance – Is like resistivity but takes the shape and size of the material into account.
• Thermal Impedance – Is like resistance but does not factor in the footprint size of the material and often includes interfacial
resistances to provide an effective value for a layer within a stack.
Additional Information
For information on ESD, soldering profiles, packaging standards, handling and assembly, please contact Qorvo for
general guidelines.
Contact Information
For the latest specifications, additional product information, worldwide sales and distribution locations:
Web: www.qorvo.com
Tel: 1-844-890-8163
Email: [email protected]
© 2023 Qorvo US, Inc. All rights reserved. This document is subject to copyright laws in various jurisdictions worldwide and may not be
reproduced or distributed, in whole or in part, without the express written consent of Qorvo US, Inc. | QORVO® is a registered
trademark of Qorvo US, Inc.