Enhanced Programming Tips June 2012

Bond Functions - First Bond (ball)

Force: Using the correct force is second only to ultrasonics as the most important of the bond functions. Since different wire diameters and a multitude of other factors determine the amount of force on a specific bond, precise force amounts are impossible to enumerate. It is imperative that the force applied does not overwhelm the ultrasonics. Usually a balance can be obtained by keeping the force and ultrasonics numerically equivalent within the specific bonding modes of voltage and current. Example: If the force in current mode is set at 40 grams, the ultrasonics should be close to 40. As always, there are exceptions. If the force is set too high, the ultrasonics will be overwhelmed and the result will be a thermocompression bond (using very high stage temperature, time and force without ultrasonics), not a thermosonic bond (using low stage temperature, time and force combined with ultrasonics). The bond will most likely be weak since the temperature is far below that required for a good thermocompression bond.

Time: The bonding time is the most forgiving of the parameters and usually, but not always, can be left at the machine default. Experimentation is encouraged.

Ultrasonics: The function that deforms the ball through high frequency vibrations of either 60 Khz or 128 Khz. This deformation is not due totally to frictional heating as has been the accepted theory in the past. The intense high frequency vibration causes the atoms of the wire and the bond pad to electronically bond forming a micro weld. The physics describing ultrasonic wire bonding is only partly known and the discussion is beyond the scope of this paper.

Overtravel: The distance the bond head travels after touchdown indicated in one tenth mil increments (0.0001 inches) on digital machines. In the older machines (2460-I/II) the overtravel was indicated in 0.8 mil increments (0.0008 inches). On standard gold or aluminum pads with a silicon substrate, a suggested 3 counts (0.0024 inches) of overtravel was the default on the 2460-I/II machines. The same starting measurement is suggested on the digital machines (2460-III through 2460-V) and would be 24 counts. As you can see, there is much more flexibility and accuracy in the amount of overtravel in the digital machines (24 counts compared to 3 counts). Lower amounts of overtravel may be required and used on sensitive surfaces such as gallium arsenide. Excessive overtravel may cause deformed balls and damage to the crystalline structure underlying the bond pad (cratering) may not be obvious during final inspection.

Search Speed: The speed at which the bond head travels through the tolerance interval for touchdown. A low number (3 - 5) may be used on a standard pad comprised of gold or aluminum. Excessively high search speeds may cause excessive deformation of the ball and possible cratering by the dynamic force induced at touchdown.

Tolerance: The height above the bonding surface the bond head travels at search speed. If the height of the part to be bonded varies considerably, a higher number is suggested. Excessive tolerance can cause the machine to run slowly. Experimentation and testing may be required to find the best tolerance interval.

Bond Functions - Second Ball (stitch)

Force: The same general advice on parameters indicated on the first bond applies to the second bond.

Time: The default is usually adequate. Experimentation is encouraged.

Ultrasonics: The ultrasonic energy deforms the wire along with the force and, to some minute degree, the stage temperature. Setting the ultrasonic parameter too high (out of balance with the force) will cause the vibrations of the capillary to disturb the weld as it is being formed. Some have referred to it as the "jackhammer effect" which will form a weak bond by constantly breaking the bond as it is formed. Setting the ultrasonics parameter too low will also result in a weak bond. Again, a balanced approach with the force and ultrasonics approximately numerically equal will usually give the best result.

Stage Temperature: Stage temperatures run the gamut from ambient (yes, you can bond without stage heat but it is difficult) to 200° C. Most bonding Iíve seen is between 150° C and 175° C. Temperatures higher than 175° C usually results in very little, if any, bond strength increase and may be dangerous to the operator.

Ultrasonic Mode: The choice is voltage or current mode. Voltage mode is much less aggressive and allows the transducer to reduce the amplitude of the vibrations as it meets more resistance during the force application of the bond. Current mode is more aggressive and attempts to maintain the amplitude of the vibrations by increasing the electrical current as it meets more resistance during the force application of the bond. Iíve seen voltage mode on bond pads and current mode on substrates and straight voltage mode and current mode on everything. I prefer current mode on everything but voltage mode does allow for more latitude since higher numbers are usually used and the energy "attached" to each increment is lower.

Overtravel: The distance the bond head travels after touchdown indicated in one tenth mil increments (0.0001 inches). In the older machines 2460-I/II the overtravel was indicated in 0.8 mil increments (0.0008 inches). Standard pads on silicon parts are usually 24 counts (0.0024 inches) on the 2460-I/II machines with 0.8 mil increments this would be an overtravel of 3 which works out to (0.0024 inches) - the same as on the digital machines (2460-III through 2460-V). As you can see, there is much more flexibility in the amount of overtravel in the digital machines (24 counts compared to 3 counts). Although a stitch bond should never terminate on a bond pad which would probably cause cratering because of the extremely hard capillary contacting the surface of the pad, if you are bonding to sensitive surfaces a reduction of overtravel from the standard of 24 counts (0.0024 inches) or 3 counts on the 2460-I/II.<=this sentence is too long Excessive overtravel may cause cut stitches and flame-off conditions.

Search Speed: The speed at which the bond head travels through the tolerance interval for touchdown. A low number (2 - 4) may be used on standard surfaces including thick or thin gold plating. Excessively high search speeds may cause cut stitches by the dynamic force induced at touchdown.

Tolerance: The height above the bonding surface the bond head travels at search speed. Usually, since stitch bonding is terminated at the substrate, the height of the part should not vary. If the height of the part to be bonded varies considerably (bonding to resistors or pins), a higher number is recommended.

Auxiliary Functions

Arc Length: Arc length is the distance in 0.1 mil increments from the end of the stitched-off wire to the EFO wand. The myth that the amount of current flowing through the wire melts the wire into a ball is just that - a myth. Itís actually the extreme temperature of the plasma (sometimes known as the 4th state of matter) which makes up over 99% of the universe. The plasma temperature easily exceeds 5000° C which almost instantly melts gold wire at its melting temperature of 1064.43° C.

Tail Length: This is the length of the wire in 0.1 mil increments from the capillary tip to the end of the stitched-off tail. This wire may not be perfectly straight out of the capillary.

EFO Time: This value, usually in the range of 100 to 255, determines the amount of time the current flows to produce a high-temperature plasma. This value usually changes with the wire size.

Tail length and EFO time work together. Usually, in order to increase ball size, the tail length must be increased (to allow more wire to be available to form a ball) and EFO time must be increased to melt more wire. Just increasing the EFO time without increasing the tail length results in melting the wire back to the chamfer of the capillary. This will cause internal capillary damage and over-annealed wire which causes extreme grain growth above the ball and results in breakage at the neck of the ball during pull testing.

Here is a quick tip on tail length and arc length: if the bonder does not catch the flame-off quickly enough to prevent a capillary hit on the next pad, damaging the pad, try increasing the arc length until a flame-off occurs and then reduce the arc length a few counts until the flame-off condition just goes away. Now the machine is setup to prevent arcing to too short "tail" inside the capillary and sensing a ball has been formed in error. This will prevent the capillary from contacting the next pad before indicating a flame-off condition and is very important if you are using a positive EFO which only senses a short when the wand hits the wire. Sometimes even a negative EFO will miss the flame-off condition and cause damaged pads.

Ultrasonic Compensation: Used only on the stitch, never on the ball, this parameter compensates for the directional mismatch of energy application in the X and Y directions. This only occurs with the 60 Khz transducer generator pair. Usually I start with about 15 which reduces the ultrasonic energy applied in the Y bonding direction by 15%. Increase as needed to eliminate cut stitches in the Y-axis. The problem is the physics of the capillary applying energy more effectively in the Y-axis and causing cut stitches seen in the straight Y direction and wires with a Y directional component.

Loop Functions

A sufficient amount of very good information is available in the Palomar 2460-V manual and an in-depth examination of loop modes exceeds the space available here.