Wafer Testing With Tungsten and Tungsten-Rhenium Probe Needles
Most integrated circuits are tested with cantilevered blade-type and epoxy-ring probe cards. Contact resistance, balanced contact force and card-service life are directly dependent on probe material and shape. As a result, a critical design step for either card is the specification of the probe needle.
In blade-type cards, the probe needles are soldered to metallic or ceramic blades that are, in turn, mounted to a printed wiring board (PWB) to match the contact pad pattern of the device under test. For epoxy-ring cards, the probes are arranged in the contact-pad pattern, embedded in an epoxy ring, and then the oven-hardened assembly is glued and soldered to the PWB.
Within the past two years, the use of tungsten-rhenium (WRe) needles in the construction of probe cards has greatly increased. Traditionally, tungsten (W) probes were used for many reasons—excellent stiffness, hardness and wear characteristics.
The growing acceptance of WRe as an alternative probe-needle material appears primarily fueled by favorable experiences of users. Wafer test engineers have found that both W and WRe materials are effective in scrubbing contact pads. But because of their greater hardness, WRe probes demonstrate superior resistance to tip fouling and better overall service-life characteristics.
Although efforts have been made to quantify probe-needle parameters that affect card behavior, some misconceptions still exist regarding the fundamental properties of W and the WRe probes.
Probe-Needle Metallurgy
Probe needles are fabricated from precision-straightened W and WRe fine-diameter wire. Wire drawing is accomplished by pulling and annealing swaged rods through incrementally smaller dies. As the wire cross-sectional area is reduced, the dislocations per unit area are increased, dislocation motion is hindered, and the material is considerably cold-worked. As a result, the strength and hardness values are systematically increased.
The addition of Re to W wire raises the recrystallization temperature and modifies the grain structure after recrystallization. Upon alloying, the Re atoms substitute into the tungsten body-centered cubic lattice, causing the lattice to contract, yielding a denser material (Table 1). The Re atoms restrict dislocation movement, increasing the strength, hardness and wear characteristics.
Re also enhances the ductility. With a 3% Re content, the ductility of a WRe probe approaches 34%, where a W probe demonstrates a ductility of approximately 8%. The WRe probes evaluated in this article have a composition of 97% W and 3% Re.
During drawing, the grains are highly elongated and a fine-grained anisotropic microstructure is obtained. The grains that form the tough, interlocking, fibrous longitudinal microstructure of a W probe are visible in Figure 1a. With the addition of 3% Re, the grain boundary area significantly increases and the interlocking structure becomes even more pronounced (Figure 1b).
From a geometrical standpoint, grain boundaries impede dislocation motion and contribute to increased strength and hardness. Visually, the tip surface of the coarser-grained W probe (Figure 2a) is not as smooth and shows more wear than the fine-grained WRe probe (Figure 2b).
Electrical and Mechanical Properties
The benefits of cold-working and alloying are obtained at the expense of increased resistivity (Table 1). Dislocations within a W probe serve as electron scattering centers; and as the wire diameter decreases, the electrical resistivity increases.
For WRe probes, the Re atoms and dislocations both act as scattering centers. As a result, WRe probes have higher resistivity and initial contact resistance values than those of W probes. Regardless of the probe diameter, the electrical resistivities of W and WRe are less than platinum and beryllium copper.
The elastic modulus is a function of the interatomic forces and is unaffected by wire drawing (Table 1). Conversely, the combination of alloying and cold-working has dramatic effects on the other mechanical properties. Before any cold-work, alloying W with 3% Re increases the strength values by 10% to 15%.
As the W and WRe probe diameters decrease, the cold-work effects and microstructural anisotropy exacerbate. As a result, the flexural strength values significantly increase almost four times greater than the tensile values (Table 1). The flexural properties are more relevant than tensile properties to probe-card design and analysis, needle bending and wafer testing.
Hardness and Wear Behavior
Probe wear and tip fouling are major issues during aluminized wafer testing. Upon exposure to air, an oxide 50- to 100-Å thick forms on the surface of the aluminum contact pads. To contact the underlying metal, which has a hardness of 50 to 170 kg/mm2, the probe must penetrate the oxide and any organic contaminants.
Although the hardness values of the W and WRe probes range from 665 to 738 kg/mm2 and 745 to 877 kg/mm2 respectively, the particles of the oxide are considerably harder, or 1,200 to 1,500 kg/mm2. The frictional forces associated with penetration cause localized plastic deformations of the tip surface into which bits of the oxide and organic residue can adhere.
Upon touchdown, the contact resistance drops and becomes asymptotic to the intrinsic contact resistance between the probe tip and the contact pad. As the pad is scrubbed, practically all the plastic deformation takes place in the softer aluminum pad. Penetration of the oxidation initially cleans the probe tip. But after a sufficiently large number of touchdowns, enough surface contamination accumulates to foul the tip and to increase the contact resistance beyond acceptable limits.
Microhardness measures resistance to localized plastic deformation and is the best indicator of probe mechanical-wear characteristics. Not surprisingly, alloying and cold-working significantly increase the material hardness. Depending on the diameter, WRe probes are 15% to 20% harder than W probes (Table 1). As such, the harder WRe probes have superior wear characteristics, durability and resistance to tip scarring than the softer W probes.
Wear Resistance of W and WRe Probes
Mechanical wear is considered surface deterioration and material loss caused by localized stresses arising from contact; in this case, a probe tip, the oxide particles and a contact pad. Probe-tip wear results from a combination of mechanisms:
Abrasion caused by the removal of material from one body due to contact with a harder body.
Adhesion caused by the relative sliding contact of the two bodies.
Fatigue resulting from cyclic stresses caused by the relative motion of the contacting bodies.
The complexity of this phenomenon precludes the use of a universal wear test. To gain insight into probe wear, a test program simulated touchdown testing on aluminum contact pads. The experiment focused only on the material effects; that is, no current was applied, contact resistance was not measured, and the tips were not cleaned.
Four sets (N=5 per set) of W and WRe-blade-type probes were mounted to a cyclic testing machine in which the overdrive and touchdown rate were precisely controlled. The probes were repeatedly touched down on an aluminized wafer. An overdrive of < 5 mils was used since excessive overdrive causes tip shattering and delamination.
The scrub marks and tips were examined every 1,500 touchdowns at 50× magnification. The scrub marks resembled oval-shaped depressions and oxide particles were observed around the periphery. After 100,000 touchdowns, the probes were removed, cleaned with 48% hydrofluoric acid to extract any embedded aluminum, and compared with untested probes.
Some scarring parallel to the probe axis was observed, but most of the scratches were concentrated at the heel of the tip contact area. During scrubbing, the tip slides toward the center of the die, the tip’s toe lifts and the heel contacts the aluminum substrate. As a result, the heels were worn considerably more than the toes. Overall, the WRe tips had less visible scarring than the W tips.
A region of excessive plastic deformation, caused by the high frictional forces during scrubbing, was observed at the heel of several W probes. No such deformations were observed in the WRe probes. When this region was taken into account, the average tip diameter of the W probes increased approximately 6% (Table 2).
In contrast, the average WRe tip diameter increased < 2%. This translated into a 12% change in the W probe-tip area and < 5% change in the WRe probe-tip area. Finally, the average tip length of the W probe was reduced significantly more than that of the WRe probes.
Summary and Conclusions
The morphological, mechanical and electrical properties of W and WRe probe needles are affected by the diameter (cold-working) and composition (alloying). As the diameters of the W and WRe probes are systematically reduced, the strength and hardness properties are significantly improved by cold-working.
Alloying W with 3% Re refines the grain boundaries, thermally stabilizes the microstructure and improves resistance to penetration. The addition of Re also imparts higher ductility, flexural strength and hardness properties.
The high hardness values of WRe probes indicate a mechanical wear resistance superior to that of W probes. Indeed, experimental results showed that WRe probes have better dimensional stability and resistance to tip deformation than W probes during scrubbing. Consequently, WRe probes withstand tip fouling and remain cleaner longer than W probes. For an application such as wafer probing in which a complex mechanical wear phenomenon is involved, WRe probes provide a more consistent contact resistance and a longer service life.
References
ASM Metals Handbook
, Ninth Edition, American Society of Metals, 1986.ASM Metals Handbook,
Tenth Edition, American Society of Metals, 1992.Rare Metals Handbook
, Second Edition, Reinhold Publishing, 1961.
Dieter, G.E., Mechanical Metallurgy, McGraw-Hill, 1986.
Savitskii, E.M. and Tylkina, M.A., Ed., Study and Use of Rhenium Alloys, Amerind, 1978.
Tietz, T.E. and Wilson, J.W., The Behavior and Properties of Refractory Metals, Stanford Press, 1965.
Yih, S.W.H. and Wang, C.T., Tungsten: Sources, Metallurgy, Properties and Applications, Plenum Publishing, 1979.
About the Author
Jerry J. Broz, Ph.D., has worked in the probe-needle industry since 1993 and is the director of R&D at Advanced Probing Systems. Before joining APS, he was a research scientist at the Denver Research Institute. Broz received a Ph.D. in mechanical engineering from the University of Colorado. Advanced Probing Systems, P.O. Box 17548, Boulder, CO 80308, (303) 939-9384.
Table 1
PROPERTY
TUNGSTEN (W)
TUNGSTEN-RHENIUM (WRe)
Density (gm/ cm3)a
19.24
19.29
Resistivity (m ohm-cm) at 20oCb
5.59 to 5.86
9.15 to 9.65
Elastic Modulus (GPa)c
394.5 ± 6.1
395.7 ± 6.4
Tensile Yield Strength (GPa)d
2.65 to 2.90
2.90 to 3.36
Flexural Yield Strength (GPa)c
5.52 to 6.05
5.95 to 6.48
Ultimate Flexural Strength (GPa)c
9.02 to 9.30
10.00 to 10.89
Vicker’s Hardness (kg/ mm2) using 100 gm load gm loadb
665 to 738
745 to 877
gm loadb
Table 2
PARAMETER
TUNGSTEN (W)
TUNGSTEN-RHENIUM (WRe)
(WRe)
Initial Probe Geometry
Tip Diameter (mil)
1.75 ± 0.01
1.75 ± 0.01
Taper Length (mil)
125
125
Tip Length (mil)
7.00 ± 0.01
7.00 ± 0.01
Probe Geometry After 100,000 Touchdowns
fffffTouchdownsTTouchdowns:
Tip Diameter (mil)
1.85 ± 0.02
1.77 ± 0.01
Reduction in Tip Length (mil)
1.46 ± 0.25
0.40 ± 0.15
Copyright 1997 Nelson Publishing Inc.
January 1997
|
|