Wire Loop

After the first transistors were introduced in the 1960's, an integration revolution started. Nowadays, semiconductors are a crucial component in every electronic device and still they are getting more and more complex.

Hong Kong skyline incl Philips Transistor Electronic Microscope Chips containing millions of transistors and working with a clock frequency of several Ghz.
But still the conventional single discrete transistor is used in billions of electronic devices.
Invisible to the user, this cheap device does its job so reliable that it is difficult to express in even parts per million.
To get this level of reliability the production was developed to a very high quality level by continuously improving the production process.

This is only possible by continuously improving the measurement and test equipment.

The characteristics of every transistor on the wafer is measured and classified before the semiconductor is positioned on the lead frame. Wire Loop In line measurement_2
Once the semiconductor is fixed on the lead frame, two bonds are made to the lead frame contacts. Also, the quality of these contacts are measured before and after it is encapsulated by the molding process.
So the moment the device is leaving the factory, all the electronic characteristics are known.

But still there is no guarantee that this device is shock-proof and it is unknown how this device will perform after a year of active duty. Therefore Philips Semiconductors in Hong Kong (nowadays called NXP) makes use of vision systems to measure the quality of the wire bonding.

Vision inspection setup

To measure the quality of wire bonding, also known as the wire loop, a vision system is developed to measure the shape of the wire during production at full speed.
The device is bonded on a double lead frame so one device is in front of the other.
Because of the small field of view and the high production speed, several problems occur:
Problem 1: Creating the imaging telecentric optics with enough focus depth to minimize measurement inaccuracies as a result of vibration in the optical path direction.
To solve this problem, an existing optic setup, used for visual inspection, is used and modified. Extra optics and diaphragms of Linos are included to create the required magnification and focus depth.
Problem 2: Creating a homogeneous back light on this small scale. Here, one high quality led,two lenses and two diffusers are used. The result of this illumination is a homogeneous back light, obtained by adjusting these two lenses compared to each other.
Typical example Problem 3: Acquiring the image and do the processing of this image within a short time. Of course the illumination is triggered by the camera to reduce blur caused by vibrations. To speed up the system, only half images are taken so the images do not have the standard 3*4 ratio but 3*8 ratio.
The resolution of the captured images is horizontally 4 micron per pixel and vertically 9 micron per pixel. Because of the high-width ratio of the device this is a perfect solution. Software speed is optimized to do the measurements in time, which is possible because of a very good homogeneous back light.

A typical image of the wire loop is given above.

Measurement software

A trigger is generated after the lead frame is positioned in front of the camera. Measure criteria The computer triggers the camera and the flash illumination. After an image is acquired the next measurements are performed (compare the drawing and the real image):

1. The maximum height (measurement 1 and 2) of the wire with respect to the die is measured.

To perform this measurement the die has to be detected at first in the image, followed by a measurement of the top side of the die. Compare this measurement drawing with the real image, it is difficult to measure the top side of the die accurately. The measurement has to be done between the corner of the die and the bond.

Example flat tail 2. The minimum distance of the wire with respect to the die corner has to be measured (measurement 3 and 4).

This means the contour of the wire at a certain area the contour of the wire is measured and the distance to the corner is calculated. The contour is measured to a certain point defined above the die corner. Otherwise, the minimum distance of the wire-die corner is found above the die itself. An example from a rejected die with a minimum distance of 29 micron that exceeds the limits is given in the picture left. This type of error is known as ‘flat tail’. The height and minimum distance to the die is measured with a precision of 9 micron.

Rejected die.jpg 3. The height with respect to the frame, measured at the two leads corners, is then measured.

This is shown in the drawing but the corners of the two leads (visible with the bare eye) is not reliably detectable in the images. Therefore an other definition is introduced. The height is measured at a certain (adjustable) distance and with respect to the die corner.

An error as shown in the picture left will be detected by en electronic functional test and rejected. As you see this is actually not a case for the vision system itself, but it can easily be detected during the functional test.

The wire loop vision system is tested in production running at an inspection speed of 5 products/sec. The measured results are saved on hard disk and are analyzed afterwards. The measured total standard deviation of the production and the vision system is less then 9 micron.
Two results of the most difficult measurement are presented in the frequency plot. Left, in blue, the minimal distance of the left wire to the crystal (die)is given, right, in green the minimal distance of the right wire to the crystal is given.
First testrun left min distance wire to crystal.jpg First testrun right min distance wire to crystal.jpg