Features-X-RAY Inspection Applications for Chip-Scale Packages

July 1998

eMail the Editor

X-RAY Inspection Applications for Chip-Scale Packages

By Craig W. Rahn, Nicolet Imaging Systems, San Diego, Calif.

"The traditional through-transmission, two-dimensional approach can provide a great deal of information about the characteristics of the conductive material used for the connection interface in a chip-scale package-including location, orientation and volume."

Introduction

Compared to older package formats, CSPs offer several advantages, including a package not much larger than the die itself. CSPs also come in several different variations, including flex circuit, rigid substrate, custom leadframe and wafer-level packages.

There are also new challenges involved in using this packaging technology. This year, CSP volumes will be very low compared to the total market for packaged devices, but more companies are ramping up production and coming online every week.

The total number of devices expected to be shipped by the end of 1998 will be between 400,000,000-600,000,000. By the year 2000, this number is expected to increase by five times.

To understand what X-ray technology can offer to both the manufacturer and user of CSP devices, we must first review the basics of X-ray inspection systems.

X-Ray Imaging

There is much confusion surrounding the advantages or disadvantages of the two major approaches to X-ray imaging. The traditional through-transmission, two-dimensional approach can provide a great deal of information about the characteristics of the conductive material used for the connection interface in a chip-scale package„including location, orientation and volume.

The X-ray signature presented to the user can characterize and differentiate a good connection from a bad one.

At first glance, the desirability of being able to slice the connection and dissect it using tomosynthesis or laminography techniques seems tempting; however, the additional information gained is typically not commensurate with the expense and effort required to capture the image.

In addition, off-axis imaging can be achieved using 2D techniques by positioning the sample device, or assembly, at an angle to the X-ray source. There is general agreement that bridges between the solder balls are straightforward to detect and that the challenge is in detecting opens.

Although 3D technology provides a view between the ball and its connection point, off-axis 2D imaging indicates if there is a problem with that connection. Needless to say, it requires an experienced eye; however, great amounts of information concerning the package connections can be obtained using this technique. If high volume inspection is needed, consideration should be given to an automated system.

Figure 1. Building blocks of an X-ray system

Automated Vs. Manual Systems

There are two key criteria used to differentiate between manual and automated systems: how the parts are handled and where the defect decision is made. A manual system is defined as one where the operator loads and unloads the parts to be inspected from the cabinet X-ray system.

These can be individual parts, tubes or trays. In addition, the defect decision is made by the operator after viewing the image. Many times the operator may want to electronically enhance the image or perhaps take a measurement. All manual systems include some sort of image enhancement hardware.

Since the defect decision is made by the operator visually, any tools employed to aid the human eye will reduce the chance of a false call. The technique of image averaging provides a "smoothing " effect to the image, helping to define the edges and aid in defect detection. In addition, by calibrating the number of pixels for a known distance, the operator can quantify measurements of void areas and diameters of solder bumps and balls.

With automated systems, the operator and the machine have different roles. The system may perform the component or assembly handling duties automatically, using conveyor feed, or the operator may place the inspection samples into a robotic system for feeding through the inspection process and even routing the defective units to a rework or reject area.

More importantly, the system will make the defect call without relying on the operator's visual interpretation. With the image routed to computer's processor, the image can be examined pixel-by-pixel, and, with the appropriate application of algorithms, the system will be able to locate and logl. Taking it one step farther, if the measurement data can be crunched and formatted, you can now determine the direction of your manufacturing process. To be more useful, the information is available at or near realtime.

Figure 2. X-ray image showing unreflowed solder

The differences between manual and automated inspection are distinct: A manual system provides more analytical examination, but it does so at the expense of throughput and with the addition of operator variability, while the automated system provides through-put and repeatability at the expense of the initial set up plus the measurement details that have to be defined. There is also a cost difference. An automated system will typically cost more to build with the additional mechanical assembly and processing power needed.

Building Blocks

We can break down the X-ray inspection system into four major blocks of functionality to see what is needed for each specific application.

The first block is the manipulation or sample-handling element. For a manual system, this consists of a stage area for each component or tray of components. The stage area should accommodate a mechanical remote-controlled manipulator to view the part off-axis.

If the part is mounted on an assembly, such as a printed circuit board, then the stage must accommodate the area of this assembly. For a 2D automated system, the part or assembly is positioned via the pre-programmed CAD data, using an X/Y table, in a position to allow passage of the X-rays through the part.

For 3D, the part is placed in a fixed position, and the X-ray source is passed through at multiple angles to achieve a single image. In addition, a conveyor interface to the X/Y table, going in and going out, is needed. In some automated systems, the part stays on the conveyor, and the X-ray source and detector move in tandem around the assembly or part, stopping to collect images.

The second block is the image train that consists of the X-ray source and the detector system. These elements will determine the quality of the X-ray image and the ability to resolve defects. In addition, parameters such as magnification and field-of-view are determined by these elements and their relative position to each other.

Keep in mind that an X-ray with sufficient penetrating capability is needed to penetrate heat sinks and similar parts. As the trend to finer-pitch packages continues, the challenge is for the detection system to provide adequate resolution so that the image can be useful for the operator viewing the image or for the analysis software used in the automated category.

Array detection technology is improving all the time, providing for increased resolution. The obvious benefits for automated systems include a fewer number of views required if the resolution can be increased or maintained, along with a field-of-view increase.

This will have a direct benefit on the throughput that can be achieved. For manual systems , the image viewed on the video monitor begins as the output from the detector system. If the quality in the beginning of the enhancement process is the best possible, then the resulting image will be that much better. This has obvious implications for failure analysis applications, as we will discuss later.

Figure 3. A typical manual X-Ray inspection system

The next building block is the image processor or image analysis element. As noted earlier, this takes the raw output from the detector system and enhances it through the use of software manipulation, typically evolved from medical applications.

For manual systems, it is critical that the quality of the image reaching the operator be the best available. Likewise, for an automated system, if the image is less than ideal, then the analysis software will report a high level of false calls and the repeatability needed will be absent. Since we are dealing with a black and white image with nothing but varying levels of gray to work with, tools to help the human eye are needed. With the ability of the image processor to divide this gray image into 256 distinct levels of gray, we can, through the power of software, assign color to a particular bandwidth. Techniques such as "shift and subtract" result in a contour relief map of the image and edge enhancement provides crisper detail.

For automated systems, the image is pixel mapped, using 256 levels of gray. Each pixel has an expected level of gray that will change if the amount of X-ray- sensitive material changes in volume at that location. With the appropriate application of algorithmic software, the amount and location of this material can be determined. If we are dealing with a predefined material reflow process, it is possible to measure slope and voids to insure that this part of the process is properly completed. An adequate amount of processing power becomes necessary due to the enormous amount of data that can be generated with each image. This data must then be compared to the expected measurements and noted in a defect file. The notation must include enough detail to enable the operator to locate the problem for process adjustments or rework. It is easy to conclude that the software is key to an automated system, with the manipulation system and image train considered the "front end."

Figure 4 - An automated X-ray system placed inline

The last element in this group of four is the display, print, store, and transfer block. This block consists of the operator interface. For manual systems, there is a video monitor and printer for hard copies. You also need the ability to store the image for later review or additional enhancement. With the ability to transfer the digital image to your own memory device, you can then use all the modern techniques to share the image with others at any location.

With automated systems, there are typically a large number of images collected. Also, with the large amount of data collected, the emphasis is placed on software that can provide user-friendly summaries of defect trends. It is possible to network the defect and SPC information to separate work stations for rework and process monitoring.

Next, we will examine different X-ray inspection applications for the new packaging technologies.

"One of the driving forces and requirements of new packaging technology is the need for high yield with the ability to produce packages at a low cost."

Defining the Manufacturing Process

One of the driving forces and requirements of new packaging technology is the need for high yield with the ability to produce packages at low cost. After the package design is completed, a manufacturing process that will produce the package reliably and repeatably must be defined. This activity could also be considered design verification. After the first few parts are produced, they should be thoroughly analyzed. Most semiconductor manufacturers have established a failure analysis lab within their facility. The first test to establish the performance of the die within the new package is electrical. This testing does not address the long-term reliability of the internal connections of the package. Unfortunately, techniques such as cross-sectioning are very time consuming. A failure analysis X-ray inspection system has the advantage of being non-destructive and will provide information about the condition of the part after burn-in and after any environmental testing. The failure analysis X-ray system requires a multi-axis manipulation unit capable of holding one part at a time. The image train should provide at least 200X magnification to show detailed images of the internal connections, whether they are pins, wires or balls. The detection system should also offer sufficient resolution to resolve down to the .001" range or less for a complete analysis. This type of system may also require an X-ray source with energy output great enough to penetrate ceramics and possibly heat sink material. This type of system is typically found in a lab environment and is at the high end of the price range of manual systems, costing upward of $200,000.

Figure 5 - Automated X-ray system reporting software

Monitoring the Manufacturing Process

After the design has been shown to be reliable, the manufacturer should monitor the process, at least on a sample basis. Since most manufacturing takes place using automated equipment, there is always the risk of a production tool moving out of tolerance quickly, resulting in many bad parts. If the X-ray system is to be useful, it should process has not strayed from the nominal.

Sometimes, in certain critical situations, engineering data on a large sample is required to be collected. This would be to characterize the process, for example. In that case, an automated system would provide the repeatability and precision measurements needed.

Other applications for an automated system include the need for high reliability in the component and the end product. By verifying each package with 100% inspection, you are providing the users of the device assurance they are not adding value by using your part and then having a failure either later in the manufacturing process or in the field.

Incoming Inspection and Specification Verification

An X-ray system can be used by the customer of CSP components who would like to verify, on a sample basis, that these parts are meeting specifications. Positioned in the incoming inspection area, a certain percentage of parts can be checked. The level of performance of the system will depend upon the accuracy and detail needed. Typically, you will want to quantify the measurements to detect trends and then to alert your supplier. Once an acceptance procedure is established, the required system performance is defined. Every effort should be made to verify the quality of the product before any value is added when it becomes part of an assembly. Eventually, if the assembly fails in-circuit test or functional test, the question that must be answered will be, is it a component failure or an assembly process problem?

"It is now common practice to incorporate X-ray technology into the assembly line either off-line, on a sample basis or even in-line for 100% inspection of BGAs and other surface mount solder joints."

Monitoring the Assembly of CSPs

Placement of any of the new packaging types is a challenge. BGAs are relatively forgiving when it comes to positioning on a printed circuit board, but verification of process performance is another matter. It is now common practice to incorporate X-ray technology into the assembly line either off-line, on a sample basis or even in-line for 100% inspection of BGAs and other surface mount solder joints. X-ray technology can not only be used to identify bad solder joints, but can also point to the area in the process that created the defect.

For example, when the X-ray system detects insufficient solder, it points to the paste-dispense operation as the source of the problem. When the X-ray inspection process shows that the part is skewed, the placement system should be checked. Voids in the solder reflow indicate that a check of the oven profile is in order. More importantly, as the automated X-ray system collects measurement information on each joint, it provides immediate feedback on the performance of the assembly process. For a chip-scale package, measurements include:

excess solder
insufficient solder
bridging
opens„using ball size
voids
skewed placement
missing ball
no reflow

This SPC information should be available as close to realtime as possible and provide a summary of defect type and location. The manual off-line systems will need enough manipulator flexibility to handle different sized assemblies. The performance of the system will be in the mid range with the emphasis on ease of use by production personnel. These systems are typically found in a high-mix contract manufacturing environment.

The automated systems, especially for the in-line category, are best suited for low-mix, high-volume lines. With the increased amount of overhead necessary to bring a new application on line, most assemblers do not have the luxury of investing time in this effort.

Repair and Rework

In high volumes, CSP packaging may be inexpensive. Currently, however, after adding the cost of the die and the rest of the assembly, finding a defect will necessitate a repair. After removing the part from the assembly, using one of the systems designed for repair and rework, the operator needs verification after completing replacement. A low-cost manual X-ray system will verify that the work was done properly.

In the automated, in-line X-ray environment, defect information is provided to the rework area via the system network. Here the operator can call up the defect type and location and even look at the actual X-ray image to verify that it is not a false call. As a result, no effort is expended in repairing a good connection. The process control engineer will then have access to a summary of the defects by type and location.

Summary

X-ray technology can be a valuable inspection tool for new packaging technology. It is now widely accepted by the BGA industry and provides X-ray system manufacturers with a growing market.