 Taking the lead in innovative electrical test solutions.
|
|
The Future of Semiconductor Substrate Testing: non-contact Technology |
||||
non-contact Capacitive Sensor approach replaces moving probe, mechanical probe, and shorting rubber technologies in order to meet high throughput, high accuracy demands.by: Don Hague & Michio Kaida |
||||
| Non-contact measurement is emerging as the leading choice
for testing semiconductor substrates. Part of an overall effort to combine
increased quality and automation, especially in electrical test, non-contact
measurement is enabling PCB fabricators to meet the semiconductor industry's
demand for a several hundredfold increase in the production of laminate
chip carriers. A recent technology transfer from the loaded board test
industry, non-contact measurement is getting the job done without the staining,
damage, or expense associated with earlier test methods.
Typically semiconductor substrates are built on either a ceramic, rigid laminate, or flexible substrate. Designs vary, but in all cases there is a relatively large land to which a ball is attached for a BGA package, or a pin is inserted in a PGA package. Increasingly due to size constraints, the BGA and u-BGA package is becoming the package of choice. These packages come in face up and face down styles depending on whether the semiconductor die is mounted on the same side as the lands or on the opposite side respectively. The lands are then connected to bond pads, typically on a pitch of 2 - 8 mils. which are plated with soft gold. Some packages, such as the Pentium Pro, assume three dimensional characteristics as the bond pads are laid out on multiple tiers in order to achieve maximum packaging density. Testing these new laminate chip carriers presents many challenges not previously encountered in conventional bare board test. In particular, due to the very high volumes of parts to be tested there is a greater requirement for accuracy and durability in the tester and fixture. Traditionally there have been three methods to test semiconductor substrates: moving probe, shorting rubber, and direct probing. Each of these technologies has widely experienced, known deficiencies which will limit their ability to keep pace with the demands of the semiconductor industry: |
||||
| Moving probe - Flying probe testers
are available with very fine accuracies, optical positioning systems, and
high probing rates. These systems are widely used in testing semiconductor
substrates but even the fastest Flying Probe Test Systems have a relatively
high unit test cost as throughput is limited by the need to mechanically
probe each of the lands and bond pads. Also of concern is that the prober
should not make any noticeable indentations on the soft gold of the bond
pads as customers are very concerned about the impact of marks on subsequent
bond reliability |
 Figure 1. With Moving Probe and Direct Probing technology
a growing concern is that the prober should not make any noticeable indentations
on the soft gold of the bond pads as customers are very concerned about
the impact of marks on subsequent bond reliability. |
|||
 Fig. 2 . Shorting Rubber testing technology continues
to struggle with the problems of staining, the short life span of shorting
rubber parts and their high replacement cost . |
Shorting Rubber - For quite some time companies have used shorting rubber to short out the bond pads and spring probes to access the lands. In order to get good test results with shorting rubber substrates must be extremely clean. In addition the rubber (which can be quite costly) wears out after several thousand hits. Also the potential for silicon oil leaking onto the bond pads can be unacceptable in some bonding applications and can cause unwanted stains | |||
| Direct Probing - In most cases (even for tiered
devices) a spring probe fixture can be manufactured to contact both the
lands (typically 20 - 30 mils dia.) and the bond pads down to 6 mil pitch.
Such fixtures are quite expensive, $15K +, and require a CCD alignment
system in order to align the spring probes with the fine bond pads. A sophisticated
pressure control system is required to prevent gouging of the delicate
bond pads. This type of fixturing can be automated with pick and place
handling, but typically throughput suffers in comparison to the shorting
rubber system as more time is required for alignment and test results are
not as trustworthy as with shorting rubber.
non-contact Semiconductor Testing Technology has moved beyond the evaluation stage to initial acceptance as a main stream tool for semiconductor substrate testing. Adapting a technology widely used in the loaded board test industry, approximately 50 fabrication and test facilities are currently utilizing non-contact testing technology from firms like Read Electronics. This technology is proving effective not only for the current demands from the semiconductor industry but should also be effective in meeting projected demands for the foreseeable future. Several engineering managers from firms utilizing non-contact technology for testing semiconductor substrates have made the observation: "The suppliers tell us that this technology can detect faults on circuits as small as 1 mil line with 1 mil space. Although we can't fabricate to those specs now, we're headed in that direction and this non-contact technology is staying ahead of our needs." |
||||
| non-contact Testing Of Semiconductor Substrates: In non-contact testing, semiconductor substrates lands are accessed with an inexpensive spring probe which can hit the relatively large target accurately. The bond pads are accessed through a capacitive sensor which is mounted in close proximity to the bond pads. The sensor is tooled to conform to the exact geometry of each die size and style. A signal is injected through the spring probe and a response is monitored via the non-contact sensor. In the event of an open circuit the signal level is greatly reduced (as shown in Fig. 3.) |
 A non-contact test fixture. Photo courtesy Read Electronics Corp. | ||||||||||||||||||||||||||||||||||||
 Fig. 3 Diagram of the non-contact testing of a PCB. Note the variation in signal levels when an open circuit is detected. | |||||||||||||||||||||||||||||||||||||
| As shown in Fig. 4 the sensor can still pick up the signal effectively even when the bond pads are shifted from nominal thus enhancing the trustworthiness of the test result. As with the shorting rubber approach, once continuity is established, additional tests at high DC voltages can be made for shorts or leakage and R,L,C values and other specialized measurement can be made with standard GPIB instrumentation. In field use non-contact testing has proven to be fast and repeatable using moderately priced, long lasting, fixturing. Furthermore, as the sensor does not come in contact with the device under test, there is no possibility for unacceptable marking on the bond pads. | |||||||||||||||||||||||||||||||||||||
 |
Fig. 4: Test Chart: A, Variation in sensor output from 100 cycle of the same part. B, Variation in sensor output from +/- 2 mil shift. C, Continuity judgement threshold typical-10.5db of nominal) D, Sample continuity failure sensor output. | ||||||||||||||||||||||||||||||||||||
| With fully automated systems throughput of more the 900 pieces per hour can be achieved resulting in extremely low unit test cost. In high volume applications pick and place technology allows the parts to be automatically loaded from trays onto a pre stage for optional tests such as co-planarity, RLC measurements, prepositioning, etc., and then placed on an index table for test and sorting onto a tray for good parts or optional sort categories for opens, shorts, and other types of failures. (Shown here in Fig. 5). | |||||||||||||||||||||||||||||||||||||
 |
Fig. 5 non-contact testing technology is capable of detecting faults on circuits as small as 1 mil line width with 1 mil spaces, a test capability which for the moment, exceeds the fabrication capability of virtually all PCB fabricators. | ||||||||||||||||||||||||||||||||||||
Conclusion:As pressure increases to make better substrates faster and at lower cost, non-contact testing technology will grow increasingly popular. Not only is this technology currently outperforming traditional methods for substrate testing, it exceeds the anticipated needs of the market for the next five years.
| |||||||||||||||||||||||||||||||||||||
|
|
|
|
