semi合集-English.pdf - 第6214页

SEMI G43-87 © SEMI 198 6, 1987 2 the flu id velocity and by decreas ing the power density seen by the flu id. The device under tes t should be m ounted such that heat transfer to th e fluid is not impeded. For leaded dev…

SEMI G43-87 © SEMI 1986, 19871

SEMI G43-87

TEST METHOD FOR JUNCTION-TO-CASE THERMAL RESISTANCE

MEASUREMENTS OF MOLDED PLASTIC PACKAGES

1 Purpose

The purpose of this test is to determine the thermal

resistance of molded plastic packages using thermal test

chips. This test method deals only with junction-to-case

measurements of thermal resistance and limits itself to

fluid bath testing environments. For this test,

conduction through the leads is minimized, thus

providing information on the ability of the plastic

package material to dissipate heat. Due to the

thermophysical properties of the heat transfer fluids

used and the effects of the variable nature of the fluid-

stirring and package-mounting procedures, this test

method should only be used for comparing the thermal

characteristics of plastic packages in the same fluid bath

system.

2 Applicable Documents

2.1 SEMI Specification

SEMI G32 — Guideline for Unencapsulated Thermal

Test Chip

3 Definitions

The following definitions and symbols shall apply for

the purpose of this test:

case temperature, T

— in degrees Celsius. The case

temperature is the temperature at a specified accessible

reference point on the package in which the

microelectronic chip is mounted.

junction temperature, T

— in degrees Celsius. The

term is used to denote the temperature of the

semiconductor junction in the microcircuit in which the

major part of the heat is generated. For purposes of this

test, the measured junction temperature is only

indicative of the temperature in the immediate vicinity

of the element used to sense the temperature.

power dissipation, P

— in watts, is the heating power

applied to the device causing a junction-to-reference

point temperature difference.

thermal resistance, junction to specified reference

point, R

θJR

— in degrees Celsius/watt. The thermal

resistance of the microcircuit is the temperature

difference from the junction to some reference point on

the package divided by the power dissipation P

temperature-sensitive parameter, TSP — the

temperature-dependent electrical characteristic of the

junction under test which can be calibrated with respect

to temperature and subsequently used to detect the

junction temperature of interest.

4 Apparatus

The apparatus required for these tests shall include the

following as applicable to the specified test procedures:

a. Thermocouple material shall be copper-constantan

(type T) or equivalent, for the temperature range

-100 to + 300°C. The wire size shall be no larger

than AWG size 30. The junction of the

thermocouple shall be welded to form a bead rather

than soldered or twisted. The accuracy of the

thermocouple and associated measuring system

shall be ± 0.5°C.

b. Suitable electrical equipment as required to provide

controlled levels of conditioning power and to make

the specified measurements. The instrument used to

electrically measure the temperature-sensitive

parameter shall be capable of resolving a voltage

change of 0.5 mV.

c. Controlled temperature chamber or fluid bath

capable of maintaining the specified reference point

temperature to within ± 0.5°C of the preset

(measured) value. A typical temperature-controlled

fluid bath assembly is presented for illustrative

purposes only.

4.1 Fluid Bath Assembly — A typical temperature-

controlled fluid bath for thermally characterizing the

microelectronic device under test is shown in Figure 1.

In this figure, the package is mounted in a fluid bath

separate from the fluid circulator, although it can be

immersed directly in an integrated fluid circulator/bath

unit. The fluid in the bath should be continuously

stirred or agitated to ensure the required temperature

stability and uniformity. Since this working fluid is

being used as an infinite heat-sink, the case-to-fluid

(ambient) temperature difference at the case

temperature reference point of interest should be

minimized, i.e., ≤ 20°C. For case-to-fluid temperature

differences > 20°C, accuracy and repeatability

difficulties may occur due to a large variable

temperature gradient in the fluid film boundary layer at

the package-fluid interface. The case-to-fluid

temperature difference can be minimized by increasing

SEMI G43-87 © SEMI 1986, 1987 2

the fluid velocity and by decreasing the power density

seen by the fluid.

The device under test should be mounted such that heat

transfer to the fluid is not impeded. For leaded devices,

the leads should be oriented in such a manner so as not

to interfere with the heat transfer to the fluid and

provide freedom to any thermal currents caused by the

power dissipation within the package. The microcircuit

package shall be mounted such that conduction cooling

through the leads or test socket or both shall be small

compared to the other cooling mechanisms. To

minimize conduction through the leads, a special socket

jig that connects No. 36 AWG wire to the device socket

should be used.

The case temperature of the device under test should be

measured with a thermocouple that is attached to the

package/lead and should not be assumed to be at the

fluid temperature. The working fluid should have a

thermal conductivity at 25°C of at least 0.0006

W/cm°C. Working fluids such as inert fluorocarbon

liquids and silicone oils are suitable as cooling media.

5 Procedure

5.1 Direct Measurement of Reference Point

Temperature, T

= T

— For the purpose of measuring

a microelectronic device thermal resistance, the

reference point temperature shall be measured at the

package location of highest temperature which is

accessible from outside the package. This reference

point location is determined with the device operating

in free air and with no external heat-sinking. In general,

this reference point is found to be on the surface of the

body of the package, or on a lead near the body, in the

major path of heat flow from the chip heating surface to

the ambient fluid. The package surface may be altered

to facilitate this measurement provided that such

alteration does not affect the original heat transfer paths

and, hence, the thermal resistance, within the package

by more than a few percent.

5.1.1 Case Temperature, T

— The microelectronic

device under test shall be mounted under specified

conditions so that the case temperature can be held at

the specified value. A thermocouple shall be attached

on the surface of the device package directly under the

chip (i.e., on the base plane of the package). A

conducting epoxy may be used for this purpose. The

thermocouple bead should be in direct mechanical

contact with the package of the microelectronic device

under test. Care should be taken to minimize exposure

of the thermocouple bead to the high temperature

gradient in the fluid film boundary layer at the package-

fluid interface.

If it is found that attaching the thermocouple directly to

the case is impractical, an alternate approach using a

thermocouple welded to one side of a thin metal disk

should be used. This can be accomplished by parallel

gap welding the crossed thermocouple wires to one side

of a 0.25 cm (0.094 in) diameter, 0.02 cm (0.008 in)

thick beryllium-copper disk and then, with a thin layer

of adhesive, bonding the other side of the disk to the

case at the point of interest. The exposed thermocouple

bead/wire on the disk shall be covered with epoxy or

silicone rubber. The attached thermocouple should not

unduly interfere with heat transfer to the fluid.

5.2 Thermal Resistance, Junction-to-Specified

Reference Point, R

θJR

5.2.1 General Considerations — The thermal

resistance of a semiconductor device is a measure of the

ability of its carrier or package and mounting technique

to provide for heat removal from the semiconductor

junction. The thermal resistance of a microelectronic

device can be calculated when the case temperature and

power dissipation in the device and a measurement of

the junction temperature are known.

When making the indicated measurements, the package

shall be considered to have achieved thermal

equilibrium when halving the time between the

application of power and the taking of the reading

causes no error in the indicated results within the

required accuracy of measurement.

5.2.2 Indirect Measurement of Junction Temperature

for the Determination of R

θJR

— The purpose of the test

is to measure the thermal resistance of integrated

circuits by using particular semiconductor elements on

the chip to indicate the device junction temperature. In

order to obtain a realistic estimate of the operating

junction temperature, the whole chip in the package

should be powered in order to provide the proper

internal temperature distribution. During measurement

of the junction temperature, the chip heating power

(constant voltage source) shall remain constant while

the junction calibration current remains stable. It is

assumed that the calibration current will not be affected

by the circuit operation during the application of

heating power.

The temperature-sensitive device parameter is used as

an indicator of an average (weighted) junction

temperature of the semiconductor element for

calculations of thermal resistance. The measured

junction temperature is indicative of the temperature

only in the immediate vicinity of the element used to

sense the temperature.

The temperature-sensitive electrical parameters

generally used to indirectly measure the junction

temperature are the forward voltage of diodes and the

SEMI G43-87 © SEMI 1986, 19873

emitter-base voltage of bipolar transistors. Other

appropriate temperature-sensitive parameters may be

used for indirectly measuring junction temperature for

fabrication technologies that do not lend themselves to

sensing the active junction voltages.

5.2.2.1 Steady-State Technique for Measuring T

—

The following symbols shall apply for the purpose of

these measurements:

— ring current in milliamperes.

— Value of temperature-sensitive parameters in

millivolts, measured at I

, and corresponding to the

temperature of the junction heated by P

— Calibration temperature in degrees Celsius,

measured at the reference point.

— Value of temperature-sensitive parameter in

millivolts, measured at I

and specific value of T

The measurement of T

using junction forward voltage

as the TSP is made in the following manner:

Step 1 — Measurement of the temperature coefficient

of the TSP (calibration).

The coefficient of the temperature-sensitive parameter

is generated by measuring the TSP as a function of the

reference point temperature, for a specified constant

measuring current, I

, by externally heating the device

under test in an oven or in a fluid bath. The reference

point temperature range used during calibration shall

encompass the temperature range encountered in the

power application test (see Step 2). The measuring

current is generally chosen such that the TSP decreases

linearly with increasing temperature over the range of

interest and that negligible internal heating occurs in the

silicon and metal traces. For determining the optimum

TSP calibration or measuring current, V

vs. log I

curves for two temperature levels that encompass the

calibration temperature range of interest should be

plotted. The optimum measuring current, I

, is then

selected such that it resides on the linear portion of the

two V

vs. log I

curves that were generated. A

measuring current ranging from 0.05 to 5 mA is

generally used, depending on the specifications and

operating conditions of the device under test, for

measuring the TSP. The value of the TSP temperature

coefficient V

, for the particular measuring

current used in the test, is calculated from the

calibration curve, V

vs. T

. At least three points

should be used to generate the voltage vs. temperature

curve for the determination of the TSP temperature

coefficient.

Step 2 — Power application test.

The power application test is performed in two parts.

For both portions of the test, the reference point

temperature is held constant at a preset value. The first

measurement to be made is that of the temperature-

sensitive parameter, i.e., V

, under operating

conditions with the measuring current, I

, used during

the calibration procedure. The microelectronic device

under test shall then be operated with heating power

) applied. The temperature-sensitive parameter,

, shall be measured with constant measuring

current, I

, that was applied during the calibration

procedure (see Step 1).

The heating power, P

, shall be chosen such that the

calculated junction-to-reference point temperature

difference as measured at V

is ≥ 20°C. In

accomplishing this, the device under test should not be

operated at such a high heating power level that the on-

chip temperature-sensing and heating circuitry is no

longer electrically isolated. Care should also be taken

not to exceed the design ratings of the package-

interconnect system, as this may lead to an

overestimation of the power being dissipated in the

active area of the chip due to excessive power losses in

the package leads and wire bonds. The values of V

, and P

are recorded during the power application

test.

The following data shall be recorded for these test

conditions:

a. Temperature-sensitive electrical parameters (V

, or other appropriate TSP).

b. Junction temperature, T

, is calculated from the

equation:

= T

+ V

− V

()

∆V

∆T

−1

where T

= T

c. Case temperature, T

(including specific location).

d. Power dissipation, P

e. Mounting arrangement (including method of

thermocouple attachment and fluid temperature).

5.3 Calculations of R

θJR

5.3.1 Calculations of Package Thermal Resistance —

The thermal resistance of a microelectronic device can

be calculated when the junction temperature, T

, has

been measured in accordance with procedures outlined

in Sections 5.1 and 5.2.

With the data recorded from each test, the thermal

resistance shall be determined from: