Thermal Resistance Measurement Technology for Semiconductor Packages

Table of contents

 

Introduction

Are you sure you’re measuring the thermal resistance of the semiconductor correctly?


Diode

 

Correct measurement may not be possible without understanding the characteristics of the diode (PN junction), which is often used as a measuring element.

 

What does the diode I-V curve mean?

●I-V curve of diode is expressed by exponential characteristics

I-V characteristics of diode (PN junction)
■ Shockley’s diode equation
    I=Is〔exp(qV/nK T)-1〕
        ≒ I=Is〔exp(qV /nK T)〕
Is : Saturation current
T : Temperature [K]

Q : Elementary charge (1.602E-19)
N : Ideality Factor
K : Boltzmann constant (1.38E-23)

Specifically, what does that mean?

  
Looking at the current side (Y axis) logarithmically … 

⇒If log(If) vs Vf is drawn as a straight line,
then it is easy to select the current/voltage conditions when acquiring thermal resistance.

But actually …

When determining measurement conditions:
★Select a current condition that is as unaffected by the series resistance element as possible.
★You can neglect the diffusion current of (C) in most cases because its current range is sufficiently lower than the realistic current setting range.

 

Explaining series resistance element in more detail

The K factor (Vf vs temperature characteristic, used for temperature estimation) cannot be affected by the series resistance element … No problem then, right?
: Ideal temperature drift : When a series resistance component is included  

 

In reality, there often is a problem.
When measuring the thermal resistance, it is often assumed there will be a uniform temperature rise in both the portion causing the series resistance element and in the intrinsic diode portion. However, in many cases this is actually not the case.

 
If you do not consider this unideal condition…

Comparing our measured thermal resistance of TEG with another company’s

  Thermal resistance θja  
Company A measurement results 169.1 ℃ / W  
Our measurement result 117.4 ℃ / W ⊿52 ℃ / W

Simulation result: 116.7 ℃ / W

Considering the parasitic elements in the diode goes a long way in precise measurement.

Analyzing the causes of measurement differences…

When obtaining K-factor When acquiring thermal resistance

 

It is the series resistance element that makes measured temperature unreliable!
    ⇒ When acquiring K factor, all the portions of the diode are assumed to be a uniform temperature, leading to a large drift of series resistance.
    ⇒ During thermal resistance acquisition, drift of series resistance is reduced.

 

There are many things to consider even with simple TEG models…

 
 

TEG measurement issue #1:
   Does Di really characterize intrinsic diodes?
  ⇒ No potential fixing structure for semiconductor is provided

TEG measurement issue #2:
   Are the R Heat and Di temperatures the same?
   ⇒ Thermal distance unexpectedly affects the temperature distribution

 

Many matters must be considered— in addition to the measured current value explained so far— in order to understand the thermal characteristics. ⇒Below are those examples.
 Parasitic circuit
  ⇒It is necessary to understand the circuit structure including parasitic elements caused by island separation structure in the semiconductor.
 Temperature difference between heat source and temperature detection element arising from the distance between them.
  ⇒We recommend estimating the actual temperature of the heat source by fitting the parameters of the simulation.

 

TEG measurement pitfalls

Point #1: Current value conditions for temperature-sensing element
⇒ It is necessary to set conditions that eliminate the effects of series resistance elements as much as possible.

Point #2: Influence of parasitic circuit
⇒ Depending on the biasing condition of TEG, parasitic circuits may form due to the device structure. This greatly affects measurement results.

Point #3: Heat source and temperature detection element
⇒ The correct heat source temperature may not even be monitored, depending on the positional relationship between the heat source and the temperature detection element.

 

⇒If the above points are not considered,
  the correct thermal resistance cannot be measured even with the measuring thermal resistance TEG!

 

Package thermal resistance measurement technology in product mounting environment

Sophisticated thermal resistance measurement technology is required in order to accurately determine package thermal resistance in the product environment.

Thermal characteristics evaluation of actual devices

  • Conventional method
  • Problems with the conventional method
  • The WTI method and case examples

 

Conventional method

Heat generation and temperature detection using parasitic Di formed by reverse biasing power supply and GND terminals


Conceptual diagram where the power and ground of inverter circuit are reversely biased

Inside the semiconductor there are innumerable structures that can deliver PN characteristics when properly biased.

 

Problems with the conventional method

The reality is that not only the intrinsic PN diodes, but also the parasitic circuits, operate at the same time.


Example of parasitic circuit operation in inverter circuit

 

What kind of operation will occur if a diode and a parasitic circuit are mixed?

Temperature distribution of microcomputer chip surface when microcomputer’s VCC-GND is reversely biased to apply forward voltage to PN junction.

Although uniform heat generation is expected, the chip surface temperature is uneven. Why?

 Cause #1:
The diode and the parasitic circuit operate in parallel. As a result, the current flow inside the chip becomes non-uniform, and there forms an area where current concentration occurs.

 Cause #2:
Since the VF voltage of the diode drifts in the negative direction as the temperature rises, hotspot proceeds to form like so: Vf decrease ⇒ current concentration ⇒ Vf decrease ⇒ current concentration.

 Cause #3:
Large enough current is required to heat the entire chip. However, if a high current flows into the chip, Joule heat generated in the wire bonding pads and bonding wires affects the accuracy of thermal measurements.

 

The WTI method and case examples

Local heat generation using clamp characteristics of VCC terminal

When a high voltage is applied to the VCC terminal, the PN junction is reverse biased, and it breaks down over a certain voltage (avalanche breakdown).

   ⇒Below are previous cases where WTI utilized this phenomenon.

 

How our technology works

★ WTI sought solutions for conventional method problems

What existing methods cannot do
 ⇒Use a heat source to reliably estimate the operating area, even if it is not the whole chip.
 ⇒Suppress the applied current to avoid unwanted temperature rise by Joule heat.

Solution:
 Clamp the applying voltage with the breakdown voltage and generate heat with low current and high voltage
 High clamping voltage can be applied (preferably breakdown voltage of 10V or higher) to AVCC

 

★ Benefits of using voltage clamp in measurements

When the clamped voltage generates heat, the breakdown voltage shifts to the positive side along with the temperature rise.
Due to this mechanism, the temperature of the voltage applied region can be kept uniform.

⇒ Relatively large current flows through the low breakdown voltage part and raises the temperature there.
 As the temperature rises, the breakdown voltage drifts toward the positive side, and as a result, the entire region is spontaneously controlled to have a uniform breakdown voltage and a uniform temperature.

 

★ Is it possible to find the thermal resistance of the product in normal operation where a different part generates heat?

If a terminal like AVCC is selected for thermal measurement, the operation area can be limited to the analog block. But aren’t there problems calculating the package thermal resistance from that? That’s the whole point, after all.

Are you concerned about the difference between the block used for thermal resistance measurement and the one generating heat in the product in normal operation?

 

Combining measurement results and thermal simulation technology is the solution!

 

Specific examples: confirmation example using SH2 microcomputer

Introduction of specific examples ... Confirmation example using SH2 microcomputer

 

The area framed in red corresponds to the analog block area.
When current is supplied to AVCC and the voltage is clamped, the analog area heats up evenly.
In addition, the effect of Joule heat generation on the pad and wire is almost insignificant compared to cases of forward bias between GND and VCC.

If any one of the I / O ESD protection Di in the analog area is used, the temperature of the corresponding area can be accurately known.

Here’s what was confirmed when we tried a thermal simulation…

Actual measurement: θja = 55.1 ° C / W
Simulation: θja = 53.2 ° C / W

-3% error margin

In the example with the conventional method, there was a confirmed difference of about + 188%. (Actual measurement was too low.).
Thermal characteristics results obtained via the conventional method cannot be used for analysis.

Therefore, it is not advisable to use such unreliable data as key parameters for thermal design.

 

How to find the thermal resistance in consideration of the heat generation area during product operation?

If a thermal simulation model that can calculate temperatures corresponding to measured data is created in the local heat generation model, the thermal resistance at arbitrary heat generation can be solved via simulation.

 

Furthermore, with this technology
it is possible to verify the thermal resistance with high accuracy and precision in the product environment.

 

Our thermal resistance tester

Along with offering consistent services for everything from measurement to simulation, WTI even has the technology to design thermal resistance testers independently.

Introduction of thermal resistance inspection system

Detailed system specifications

  • Parallel measurement possible on up to 10 channels
  • Supports 3 modes: constant current, constant voltage, and constant power
    ⇒Constant power control possible within the constraint of 2-terminal control
  • Heat generation operation via PWM is possible in 3 modes: constant current, constant voltage, and constant power
  • Supports automatic measurement for both steady and transient (transient is conditional)
  • All waveform information is automatically saved and can be acquired with the viewer function
  • The inspection device itself (measurement in the cooling process) is even compatible with the new JEDEC standard “JESD51-14″.

★ All channels are compatible with the new method proposed here.
We at WTI can support all market needs.
The example listed is a device developed exclusively by our company. We will gladly propose a system with optimum specifications according to your request.

 

 Copyright © 2020 Wave Technology Inc. all rights reserved.