Investigation Conquers Probe-Card Problems
Information gathered from failure analysis helps optimize probe-card design and manufacture, and it helps ensure accurate wafer test.
Bahadir Tunaboylu, Chander Sekar, and Hank Scutoski, Cerprobe, Gilbert, AZ -- Test & Measurement World, 11/1/2000
| For a related article, see Analysis yields secrets of probe-assembly failuresin the January 2001 issue of Test & Measurement World. |
IC manufacturers use epoxy probe cards to test ICs in wafer form by probing the chips to determine whether the individual circuits meet design specifications. Probe cards must enable reliable, repeatable, and consistent electrical contacts between the tester and the chip pads. Components and materials used in epoxy probe cards can fail prematurely because of inherent defects, process-induced defects, design problems, or misuse. Failure analysis (FA) can determine the specific cause of probe-card failure.
A failure investigation must assess the factors that could have contributed to the failure and identify which factor was the primary cause. To conduct a successful investigation, you must be familiar with the wafer-test conditions and the probe-card manufacturing process. Figure 1 outlines the critical steps in the failure-analysis process. Initial failure observations are usually optical. For some probe card failures, although obvious visually, you may need high-magnification electron microscopy to determine the root cause. The analysis method can be either destructive or nondestructive, depending on the nature of the failure.
Figure 1. Probe-card failure analysis extends from optical identification of problems through documentation of corrective action.
Failure Mechanisms
Probes, themselves, are the most common causes of probe-card failure because they may be subject to overstress, and this article will focus on probe failure analysis. Machinable glass-ceramics used in the ring-build process and in the epoxy cards themselves may also fail; an article in the January issue of Test & Measurement World will investigate failures in these components.
Probes are susceptible to damage from a number of sources. Bending operations during manufacturing can introduce microcracks in the bend area that can propagate during touchdown cycles and cause a probe to fracture. Both excess loading (beyond specified overdrive) during burnishing and cleaning operations in wafer sort can result in excessive probe-tip wear and undesirably large probe marks. Tip bending of probes, also called fish-hooking, is commonly the result of accidentally crashing the probe card into the wafer under test, uncontrolled overdriving (during wafer test), or nonplanar (tilted) interfacing of the probe card to the wafer surface.
Abnormal wear of only a few probes in a probe card may point to a faulty card design. If exposed to chemicals or excessive humidity, probes can corrode. Finally, contamination of probe tips by foreign material picked up from the environment (thin films of ink, polyimide, regular dirt, sulfides, or oxides) can lead to high contact resistance during test.
Figure 2. (a) A probe-card device exhibits a tungsten probe fractured near the knee-bend and another deformed extensively with a visible bending mark at the knee-bend. (b) Bend marks and microcracks are visible on the probe knee-bend.
Failure Analysis and Methodology
The first step in a failure investigation is to observe the probe-card failure. You could use an optical microscope for simple viewing, but because you’ll likely need 400X magnification, you’ll get better resolution with an SEM (scanning electron microscope). You can easily identify the chemistry of features and detect the elemental spectrum, down to micron sizes, by attaching an EDS (energy dispersive x-ray spectrometer) to the SEM. Such a system typically requires very little sample preparation, and it generates data quickly. If you need to perform quantitative compositional analysis with good depth resolution of thin-film layers or contaminants, you can use electron spectroscopy for chemical analysis (ESCA), which is also called x-ray photoelectron spectroscopy, or you can use auger electron spectroscopy. Another useful tool for identification of residues is Fourier-transform infrared spectroscopy. (The detailed descriptions of these methods are presented in the ASM handbook.1)
Identifying probe fractures or severely damaged bends at knee-bend locations during wafer test may require a nondestructive SEM analysis. The microcracks that might be generated during the wire-bend manufacturing operation are responsible for the stress-rupture of probes in the test cycles, as seen in the SEM image of tungsten probes (7 and 8 mil in diameter) in Figure 2.
A parallelism problem (significant tilting of the card) during wafer sort can cause excessive damage to the tips by bending them (fish-hooking) on one side of the probe card. We identified such a failure by viewing a probe card with an SEM and simply counting the fish-hooked probes in every DUT and assigning numbers to the device layout (Fig. 3). The left side of the device contains virtually all of the failed probes—54 vs. only 2 (on DUT 2, not shown in the figure) on the right. Fish-hooking can also result from using damaged (pitted or worn) tungsten-carbide sandpaper, in which case the location of failed probes tends to be random. Frequently changing the sandpaper can help prevent fish-hooking and contamination.
Figure 3. (a) The design layout of a probe-card device shows, encircled in red, the number of probes with bent tips for every DUT. This figure shows 6 of 16 probe locations. (b) Two probes bent by impact loading are revealed in an SEM image.
Figure 4. (a) This SEM micrograph shows a worn-out tungsten-rhenium probe from a device that was burnished excessively. (b) The burrs around the mushroomed probe tip caught polymeric residue, as identified by EDS.
Probes can get mushroomed tips due to overloading during burnishing (abrasive cleaning). This can lodge debris in the probe tip, which causes jumps in electrical contact resistance during the wafer test. Such a damaged tungsten-rhenium probe is seen in Figure 4a. You can use an EDS to determine the nature and perhaps source of the debris; the EDS in Figure 4b indicates that the debris contains mostly carbon, sodium, and chlorine.
Tip splitting is another problem seen in hard and brittle tungsten or tungsten-rhenium probes; tip splitting is rare in more ductile beryllium-copper or Pd-alloy probes.2 The split tip is usually worn away as the probe tips are sanded in successive steps, leading to the fabrication of large, flat tips in probe-card manufacturing. Such cases of tip splitting in wafer test may indicate a defective, brittle needle material.
Figure 5. The probe metallization Ti/Au is shown delaminated from the blade shank surface in the SEM image (top left, magnified top right). Chemical analysis on the blade surface (bottom) shows a large Al peak from the ceramic Al2O3 and a thin layer of Ti/Au sputtered layer and trace amount of Ni, probably from a Sn/Ni solder overlayer.
Metallization failures can occur in ceramic blade cards when the ceramic surface is excessively damaged or altered by laser processing, or when sputtering of the metals is done on a contaminated surface. The result is a probe delamination from the ceramic blade shank (Al2O3) as shown in Figure 5. Using an EDS, you can identify the chemistry of the ceramic surface as well as the delaminated metallization surface. In this case, the Ti layer had poor bonding to the ceramic but was bonded well to the upper Ni or Pb/Sn solder layers, which are used for attaching the W-probe to the ceramic blade. In epoxy cards, you can typically use an optical microscope to view the solder-probe joint failures on the PCB surface. T&MW
FOOTNOTES
1. ASM Handbook: Volume 10, Materials Characterization, ASM International (www.asminternational.org), 1996.
2. Pitney, Kenneth E., Ney Contact Manual—Electrical Contacts for Low Energy Uses, J.M. Ney Co., 1973.
Bahadir Tunaboylu is a materials scientist at Cerprobe. He has a Ph.D. in materials science from the University of California, San Diego, and a M.S. in ceramic engineering from Alfred University, NY.
Chander M. Sekar has been the corporate statistician at Cerprobe since 1996. He has a Ph.D. in statistics from the University of Madras (India).
Henry P. Scutoski is the vice president of Quality & Process Management at Cerprobe. Scutoski has an M.B.A. from the University of Dallas and a B.S. in Industrial Engineering from Bradley University.
Follow-up: "Analysis yields secrets of probe-assembly failures"
Probe failure data is just part of the information needed to help manufacturers improve probe-card design, fabrication, and application. An article published in January 2001 discussed ceramic blade cards and rings.
