Concept and prototype of the structured LED lighting system

In 2D-type defect inspection, the light source must provide light with a proper incident angle and large spread range to suitably illuminate the critical segments of bonding wires.

While in 3D-type defect inspection, the lighting device should provide light with an exact spread of incident angle so that sagged wires can easily be distinguished from good wires.

To provide suitable light with a controllable incident angle and spread range for each of the four sides of an IC chip without mechanism movement, we designed a prototype structured lighting system containing four LED arrays for this objective. Each LED array was mounted on a soft PCB. The four soft PCBs were fixed symmetrically at the north, south, east and west strips of a hemispheric holder, as shown in Figure 17.

(a)

(b) (c)

Figure 17. (a) A side-view of the proposed structured lighting system. For clarity, only the east and west strips of LED arrays and associated soft PCBs are depicted. (b) A 3D illustration of the structured lighting system (c) A photograph of the prototype of the proposed structured LED lighting system

24

-This configuration matches the four LED arrays with the bonding wires at the four sides of an IC chip. LEDs on different rows can be programmed to turn on or off independently. That is, both the incident light angle and the spread range of the incident light can be totally controlled electronically, highlighted the bonding wires without any mechanical movement. According to Eq. 3-1, the light with a smaller controllable spread range of incident angle can distinguish a slight sagged defect. Both the smaller LED and the larger working distance result in smaller spread range of incident light angle. In the developed prototype, each LED array contains 30 rows of LEDs and the working distance is set at 160mm so that each LED row can provide a fixed 2.5 degree spread range of incident light. Following Eq. 3-1, each row of LEDs in the prototype can illuminate bonding wire with a slope in the range of 1.25 degree. The lowest row is set to provide light at a 6 degree incident angle, so that this row of LEDs is a little higher than the highest point of bonding wire to avoid collision. The number of LEDs on each row can be controlled to provide sufficient light to illuminate the bonding wires within the FOV (field of view) of CCD without unexpected reflection. The size of FOV is decided by the pixel size and the resolution of CCD. To inspect the bonding wire with 20 um diameter, the pixel size of an AOI system must smaller than 5um. The estimated maximum FOV of CCD will be about 10mm x 10mm and the number of LEDs in each row is set to be three.

When the LEDs mounted on the i-th to j-th rows are turned on, the range of slope of the illuminated bonding wire can be calculated using Eq. 4-1. For example, the LEDs on the 5th to 8th rows can illuminate the bonding wires with slopes from

8 325 to

8

375 degree.

( 90 - upper boundary of incident light angle) * 0.5

< inclined slope of illuminated bonding wire

< ( 90 - lower boundary of incident light angle) * 0.5 That is,

{ 90 - [ 6 + ( j ) – 1 ] * 2.5 + 1.25 ] } * 0.5

< inclined slope of illuminated bonding wire

< { 90 - [ 6 + ( i ) – 1 ] * 2.5 - 1.25 ] } * 0.5 Eq. 4-1

In practice, the incident light angle of LED will not equal to the theoretical value. Not only the wires located in the center of FOV of the CCD will be reflected but also those within the FOV will cause some deviation of the incident light angle, as shown in Figure 18.

If an LED is set in θ slope and its incident light angle ranges in Δθ, the actual incident light angle of the LED can be represented as Eq. 4-2.

Figure 18. An illustration of the inaccuracy of incident light angle of the structured LED lighting system

26

-For example, if the width of FOV is the maximum estimated value of 10mm, the actual incident light angle of the 11-th row of LEDs will range from 28.19 to 31.85 degree. The difference between actual and theoretical value, from 28.75 to 31.25 degree, will only result in small extension of reflective segment of the bonding wire. Such deviation of the incident light angle will be neglected in the following part of the dissertation to simplify the description.

In Section 2.2, three types of lighting device were tested for 2D and 3D bonding wire inspection. In this section, the illumination results of the structured LED lighting system will be presented. In Section 4.2, the structured LED lighting system was set to provide light with a proper incident angle and spread range to illuminate the critical segments of bonding wires on one side of the IC chip for 2D wire inspection. In Section 4.3, the structured LED lighting system was tested for 3D wire inspection. Three IC chips with good, slightly sagged and seriously sagged bonding wires were used as test samples. These IC chips were specially bonded and only one wire was bonded in one boundary for grabbing the profile of wires. For 2D wire inspection, the LED was set to provide light with a proper incident angle and spread to illuminate the critical segment of bonding wire. For 3D wire inspection, the structured LED lighting system was set to distinguish the good, slightly sagged and seriously sagged bonding wires.

4-2 Illumination effect for 2D wire inspection

When the slope of bonding wire surface was obtained from the CAD data, the LEDs on proper rows were turned on, according to Eq. 3-3, to provide good illumination. For example, the slope of the bonding wire surface of the 2D sample IC is around 32 degree, as shown in Figure 19. The proper incident light angle should be 26 degree and the LEDs on the 5th to 13th rows of the developed structured lighting system were turned on. These

LEDs provides light in the incident angles from 17.25 to 34.75 degree onto the wires.

According to Eq. 4-1, all the bonding wires spread in the slope range from 27 8

5 to 36 8 3

degree will be clearly highlighted. The positions of the bonding wires and the 2D-type defects can easily be found from the grabbed images. That is, the structured LED lighting system can illuminate the bonding wires clearly.

(a) (b)

Figure 19. The image of bonding wires grabbed using the structured LED lighting system (a) all good wires (b) some shifted wires

4-3 Illumination effect for 3D wire inspection

According to Eq. 3-1, each row of the structured LED lighting system with an incident angle θ can illuminate bonding wires with a corresponding slope of σ. Figure 20 shows the profile of a good bonding wire and the critical segment which was with the slope of around 20 degree. In Table 3, the first column shows the LEDs in that row was turned on, the second column shows the slope range of the illuminated wire according to Eq. 4-1 and the third column shows the grabbed images by using the structured LED lighting system. First, only one row of LEDs on the structured lighting system was turned on and then more rows were turned on to provide the best illumination. For clarity, the images without illuminated bonding wire can be omitted, when the LEDs on the first to 12th rows and the 22nd to 30th rows were turned on. When the LEDs on the 16th to 19th rows were turned on simultaneously, the mainly critical segment of the bonding wire was illuminated perfectly.