Semiconductor Thermal Testing Principles
The following discussion is a brief introduction to the technical fundamentals of semiconductor device thermal testing using the electrical method of junction temperature measurement. For an in-depth treatment of these topics, download the document entitled The Fundamentals of Thermal Resistance Measurement.
Junction Temperature Sensing Principles
Semiconductor junctions possess useful characteristics for the measurement of junction temperatures. These characteristics are called "temperature sensitive parameters" or TSPs. The most commonly used TSP is the relationship between the forward-biased voltage and the junction temperature in response to a constant, forward current. Using this property, any diode junction can be utilized as a temperature sensor. If the sense-diode junction is part of a semiconductor device, the temperature rise caused by heat dissipation in the device can be measured. This capability forms the basis for a variety of component tests related to the heat dissipation characteristics of the semiconductor devices and their associated packaging.
Sense Junction Calibration
Device calibration is performed with the device immersed in a uniform temperature bath with only the sense-current for the chosen TSP applied. Since the sense-current is too small to cause significant junction-heating the device case temperature will nearly equal the internal junction temperature with a stable bath temperature. By collecting data over a range of bath temperatures, data comprised of sense junction voltage versus junction temperature can often be fit to a straight line. Using this linear calibration relationship, sense junction voltages can be readily converted into the corresponding junction temperatures during powered operation of the device.
Thermal Resistance Test
This test method requires the measurement of device junction temperature during the application of continuous heating power. Thermal resistance is analogous to electrical resistance and defined by the following equation once steady-state thermal equilibrium has been reached:
Rjx = (Tj - Tx) / P
Tj = junction temperature (°C)
Tx = reference temperature (°C)
P = heat dissipation (watts)
Rjx = thermal resistance junction-to-ref. temp. (°C/watt)
The "reference temperature" is specific to the type of thermal resistance being measured. Reference temperature selections typically include ambient air, case temperature, lead temperature, or some other specific site on the or its mount/holder. The selection of reference temperature defines the type of resistance measurement, i.e., junction-to-ambient, junction-to-case, etc. For most devices, the junction temperature is measured during brief intervals when the otherwise continuous heating power is interrupted and the sense current is applied to the sense junction(s). When the device heating power is interrupted, sense junction voltages are immediately sampled at a high rate and the calculated junction temperatures are then extrapolated back to the instant of heating power interruption. The entire cycle: interrupting the heating power, applying the sense current, measuring the sense voltage, and restoring the heating power, occurs over a fraction of millisecond. This technique ensures accurate measurement of junction temperatures over a wide range of device types.
Die Attachment Evaluation
Die-Attachment Evaluation, also called a Power Pulse Test, is quite distinct from the thermal resistance test, since it is a non-equilibrium test and does not approach steady state thermal equilibrium as required for thermal resistance testing. In this test, a short, precisely defined heating power pulse is applied to the device and the junction temperature is subsequently measured. A sufficiently short but "hot" pulse can heat the chip without significantly effecting the package temperature. For an appropriately selected pulse duration, the measured thermal impedance will be a sensitive indicator of the quality of the die attachment. Die attachment testing provides a split-second, non-destructive means for accurate die-attachment evaluation. In component production, devices with die bond voids can be quickly detected and eliminated.
Heating Characterization and Transient Thermal Response
Thermal impedance is calculated identically to thermal resistance but prior to steady state equilibrium; once thermal equilibrium is reached, thermal impedance become equal to thermal resistance. Heating Characterization determines the device thermal impedance as a continuous function of the heating duration. The important difference between thermal impedance and thermal resistance is the time aspect: thermal impedance is measured before steady state is reached and thermal resistance is measured only after thermal equilibrium is reached under steady, continuous operation.
Heating characterizations are frequently expressed in heating curves. [Sofia, JW, "Analysis of Thermal Transient Data with Synthesized Dynamic Models for Semiconductor Devices", IEEE Transactions CPMT, Vol 18, 1995, pp 39-47] -Download Heating characterization also encompasses the time-constant spectral decomposition of the transient data which provides the basis for transient thermal models, and structure function analysis. From this data analysis, the internal thermal resistance components associated with die thermal spreading in the chip, die attachment, package thermal spreading, and dissipation to the ambient can be determined. Such data provides an invaluable tool for packaging development.
The data resulting from the Heating Characterization can used to perform accurate mathematical thermal-simulations of devices operating under non-steady power conditions. Analysis Tech Thermal Analyzers provide this simulation capability for a wide variety of library waveforms as well as for user-defined, arbitrary heating waveforms.