Bond Testing Ultra Fine Pitch Wirebonds
Although the pitch between bonds reduces year on year, the challenges of meeting this demand has fallen on other processes more than the science of bond testing. With industrial bonding at 50µm becoming common place, this is no longer the case.
Wire bonding at 25µm and below is proven in laboratories, including the shear and pull testing of these wires. Although it is possible to use existing testers at these geometries, their ease of use and accuracy is being pushed to, if not beyond, their limits. In order to accurately test and measure these finer pitches, equipment manufacturers have to address a series of issues such as:
* Lower ranges of force measurement, 0 to 25gf with no proportionate loss of accuracy
* Improved optics and illumination to enable easy alignment of the tool to the sample and grading of the failure mode
* New tool designs optimised for the changing shape of ultra fine pitch bonds and the relative increase in thickness of the passivation layer
* Improved step back accuracy when shear testing
* Anti-vibration measures to maintain accuracy and optical performance with typical environmental conditions
* Improved control and accuracy of the sample and tool manipulation
Measuring low bond strengths
The strength of ultra fine pitch (UFP) bonds is typically about 15grams. Existing low range bond tester transducers are designed to test several hundreds of grams. To maintain force measurement accuracy, transducers for UFP applications should be optimised to measure within the expected bond strength range. For pull and shear tests, a 25gram range was selected as most suitable to test at 50µm and below. Accuracy is maintained for measuring low forces by using friction free sensing elements.
Transducers that have been optimised for 25gf can be easily damaged if they are not protected. Changing a shear tool or pull hook can easily subject the measuring devise to forces many time that of the maximum range. Although the design will require delicate parts to enable sufficient output when measuring low forces, it should also be robust. New transducers that are developed should include mechanical stops that protect the transducer element from over strain during normal operation.
Traditionally bond testers use stereo-zoom microscopes for tool alignment and failure mode analysis. The tool is required to be vertically above the test site and the optical system must therefore view at an angle, as shown in Figure 1. Viewing at an angle requires a depth of focus over the testing area, which is inclined to the axis of the optical system. In addition, failure mode analysis of UFP ball bonds requires high magnification. Stereo-zoom microscopes with adequate magnification cannot provide sufficient depth of focus. In fact, the focal plane becomes so narrow that the focused image is virtually in one plane and is therefore effectively, two-dimensional. The stereoscopic image is then of no value.
Viewing angle of microscope
Stereo-zoom microscopes require large diameter objective lenses and so because of the viewing angle, their working distance is large. To obtain greater depth of focus with the required magnification, a shorter working distance is required. This is only possible if the objective is smaller and the optical system would then be monocular. Borescope technology is well suited to this requirement. Objective lens systems can be 8mm in diameter with corresponding working distances of 12mm. Although the depth of focus is still limited, it is adequate to cover the inclined bond with a resolution of 2µm. Examples of such images and corresponding conventional microscope images are shown in Figure 2.
Microscope (left) and Borescope (right) images for tool alignment (top)
and failure analysis (bottom)
The image from the Borescope is displayed on the bond tester monitor, which is then more ergonomic than using a microscope. Illumination is very important to achieve the best from the optical system and lighting should include both diffused and coaxial sources.
The principal aim of a bond tester is to apply a load such that the bond fails and its strength is measured. However, often the bond stays intact and some other failure occurs. In the case of a ball bond, the ball may shear leaving the bond untested. While this is often considered to be the hall mark of a good bonding process, because it represents a good bond, it will not reveal the strength of the bond nor any trend within the process. A better test would apply more load onto the bond so that it failed. As it is the tool that applies the load, its design is crucial to optimising the test system. By using a "cavity shear" tool developed by Dage a few years ago, it is possible to achieve superior test data compared with the more conventional "chisel tool". Reductions in pitch result in corresponding reductions in ball size. The passivation layer however remains the same thickness. Because the tool must land on this layer, it affects the minimum shear height that can be achieved above the bond (Figure 4). A conventional chisel tool limited to shear heights of the passivation thickness and some clearance will often cause the gold ball to be sheared rather than the bond fail. On the other hand, the cavity tool applies the load to the ball more evenly and bond failures are far more frequent.
Chisel and cavity tools
Minimum shear height for testing UFP. Minimum shear height = passivation
The shear heights shown are referenced from the height of the bond pad. It can be seen that with the chisel tool at 0.5µm shear height some balls failed rather than bonds. At 1.0µm, the number of failed balls is very significant. The cavity tool performed better at 2.0µm than the chisel tool did at 0.5µm. The measured bond strengths from the cavity tool represent the bond strength up to 2.0µm shear height. The results from the chisel tool become a mixture of bond strength and ball strength above 0.5µm. With typical passivation layers of 0.5 to 1.0µm the chisel tool cannot be used to accurately measure bond strength. Measured forces using the cavity tool are higher, as the tool does not distort the ball as much as a chisel tool. The typical deformation of gold balls is shown in Figure 5.
Deformation with chisel (top, 13.4gram load) and cavity (bottom, 14.5gram
Low shear heights are best for testing - a cavity tool requires 1.5µm or less to ensure optimal data. If the passivation layer is 1.0µm, accurate step back is essential to ensure that the shear height does not become too large or that the tool drags on the substrate during the test. Step back must then be controlled between 1 and 1.5µm, requiring sub micron accuracy.
As geometries reduce, ambient vibration becomes more significant. This will depend on each individual location but the bond tester should be designed to perform in typical industrial environments. Unchecked vibration will have an adverse effect on calibration, test results and image quality at high magnifications. Typical sources and frequency of vibration in the industrial environment are:
* 20-30Hz, office equipment - PCs, fans
* 10-15Hz, factory equipment - compressors, machines and lifts
* <10Hz, building rumble/sway
Vibration isolation using spring damper mounts start to effectively attenuate vibration at frequencies about 3 times the resonant frequency of the isolation system. To isolate all of the possible sources of vibration the resonant frequency of the system mount would then have to be at least one third of the lowest source. If this is factory equipment, the frequency required is about 3Hz. Manufacturers should then be looking to supply equipment that can offer anti-vibration mounts to suit a range of environmental conditions.
Low frequency mounts must include damping and self-levelling to minimise machine movement/sway. The anti-vibration system should also be designed to retain its ergonomic operation. Anti-vibration mounts are nominally 50mm tall. A machine cannot be simply placed on these without considering the operator interface. Thought should be given to the position of the controls by separating them from the isolated portion of the machine. The operator should not induce or bridge ambient vibration into the bond tester.
Improved control and accuracy
Smaller geometries require more precision when aligning the tool to the bond and during the test. X and Y movements must be designed to be easily manipulated across a 25µm pitch. The landing force of its UFP shear tools should also be reduced in proportion to the smaller landing area of their tip. If the tool face width is halved, the landing force must then be reduced by a factor of four. For pull testing, hook concentricity is greatly improved. Small tools and hooks may be easily damaged when they are being fitted or removed. Systems should be introduced that protect the tool system. An example of this can be found on the latest 5000 system from Dage. The sheath automatically retracts to expose the tool as it is fitted and retracts to cover it on removal. The location of the tool is precisely controlled and no fixing screws are required.
By incorporating more advanced software features into the systems, repetitive manipulation is more consistently controlled and performed more rapidly. For example, the tool can be automatically indexed to the next bond at the correct height for the next test. These mini-subroutines are an intuitive mixture of programmed moves and self teach operations.