Understanding Probe-Contact a-Spot Oxidation During Elevated-Temperature Wafer Test

Elevated-temperature wafer test presents many engineering challenges for probe-card design and construction. Variations in contact resistance (CRES) magnitude and stability are often attributed to interfacial phenomena occurring between the probe tip and contaminants from the aluminum (Al) bond pads.1, 2 This assumption does not fully address the dramatically unstable CRES behavior of some probes at elevated temperatures. A better understanding of the contact mechanisms will assist probe-card designers and test engineers in selecting the probe materials best suited for elevated-temperature test applications.

Tungsten (W), tungsten-rhenium (WRe), beryllium-copper (BeCu), and Paliney-7® (35Pd-30Ag-14Cu-10Pt-10Au) are the traditional probe materials used in probe-card production. Each material has advantages and drawbacks for elevated-temperature testing. Advanced Probing Systems recently introduced a proprietary metallic-alloy material, NewTek™ Probe, that may provide benefits over the existing materials.

W and WRe probes are the most commonly used needle materials for probe-card construction. These materials have high elastic moduli and excellent hardness and are effective in scrubbing through the oxides and residues on the aluminum pads. During elevated-temperature (>70°C) wafer test, aluminum and aluminum oxides (Al2O3) adhere to the probe tips and cause dramatic increases in CRES. Abrasive cleaning is used to recover low CRES; however, frequent cleaning reduces the service life of the probe card.

Pd-alloy, or Paliney-7, provides excellent oxidation resistance at all test temperatures. BeCu maintains low CRES for test temperatures below 125°C; however, above 125°C, it develops a thick insulative oxide film.3 Pd-alloy and BeCu have elastic moduli and microhardness values significantly lower than those of W and WRe. As such, they may not provide sufficient pressure to scrub through the contaminants on actual product and will wear rapidly on an Al wafer.2 Typically, these materials are used for probing softer gold pads and solder bumps.

Relative to BeCu and Pd-alloy, the metallic-alloy probes have comparable oxidation resistance with higher elastic modulus, strength, and hardness values. They could be used for applications requiring low CRES, smaller probe diameters, and higher contact forces.

For this comparison, the low and stable CRES of NewTek, BeCu, and Pd-alloy probes at 85°C is contrasted with that of W and WRe probes. The concept of the conductive a-Spots is used to evaluate probe CRES behavior at 30°C and 85°C and approximate localized temperature variations. The results provide significant insights into the electrical contact phenomena that occur during elevated-temperature production wafer testing.

Wafer Test and Electric Contact Theory

During wafer test, probe tips contact the Al pads of the DUT. At the start of overtravel, the probe toe makes initial contact; at the end of overtravel, intermetallic contact is made at the surface asperities of the probe heel (Figure 1a). Initially, the deformation is elastic, and with increasing overtravel, the softer material plastically deforms until the contact force is supported. The size of these plastically deformed regions is directly proportional to contact pressure and inversely proportional to material hardness.4 As a result, the real probe tip contact is considerably smaller than the observed scrub mark, approximately 65 to 85% (Figure 1b).5

The test current flow between the probe tip and the aluminum pad is constricted to intermetallic contact areas, known as a-Spots, and across any thin conductive or semiconductive films.6 The CRES between the probe tip and the contact pad is comprised of the constriction resistance plus the interfacial film resistance.

Current flow distortion causes increased constriction resistance, and the film-resistance contribution depends on the thickness of the contaminant layer. Subsequent increases in CRES cause further Joule heating, higher localized temperatures, and oxidation growth. Although the conductive regions cover a fraction of the contact area, the localized processes that occur solely determine the electrical contact reliability.7

Experimental Design and Test Results

Several 40-pin epoxy-ring probe cards were constructed from materials rated for elevated-temperature conditions. Design 1 was built with the metallic-alloy, Pd-alloy, and BeCu probes; Design 2 was built with the metallic-alloy, W, and WRe probes. All cards were evaluated in a controlled, pseudoproduction environment using a hot chuck wafer prober and a 3-mil overtravel. A minimum of 100k touchdowns was made on aluminized 8″ wafers with nonoverlapping scrub marks.

Design 1 cards were not cleaned throughout the test, but Design 2 cards were abrasively cleaned once after 100k touchdowns to evaluate the level of probe CRES recovery. After burnishing, the cards were not cleaned until final metrology at 500k touchdowns.

The initial and final metrologies were performed with a probe-card analyzer. CRES data was collected every 2.5k touchdowns from Design 1 and Design 2 cards at 85°C until the end of test (500k touchdowns). CRES data also was gathered from Design 1 cards at 30°C until the end of test (250k touchdowns) using a 48-channel parametric analyzer with a 50-mA forcing current. Between measurement intervals, no current was forced through the probes.

The CRES values of Design 1 and Design 2 cards were filtered so 60 W was not exceeded (Figure 2a). The initial CRES of the metallic-alloy, Pd-alloy, and BeCu probes was approximately 500 mW , 275 mW , and 125 mW , respectively, and remained below 750 mW . Debris shed by the probes caused a few readings to exceed 1.5 W .

The W and WRe probes did not demonstrate such CRES stability (Figure 2a). The W and WRe probes started with CRES values below 1 W , but with repeated touchdowns, CRES increased and became unstable.

To further investigate room-temperature W and WRe probe CRES behavior, additional Design 1 cards were evaluated at 30°C. Figure 2b shows the CRES behavior during the first 200k touchdowns at 30°C and 85°C.

Upon initial touchdown at 30°C, the probes demonstrated high CRES values that were attributed to the presence of tungstenates on the probe-tip surface. After 2.5k touchdowns, the CRES values of the 85°C W and WRe probes were higher than those obtained at 30°C; however, all probes had values of less than 750 mW . After that point, only the 30°C probes remained stable and below 1 W .

At 100k touchdowns, these cards were abrasively cleaned with 30 nonoverlapping touchdowns on a 3-µm grit burnishing pad. Immediately after cleaning, the 30°C probes recovered CRES values comparable to their initial values; however, the 85°C probes did not demonstrate a full recovery (Figure 2b).

At the next measurement interval (102.5k), the CRES of the 85°C probes had increased significantly and again was unstable. Although electrical current was forced only at the measurement intervals, the experimental CRES values were comparable to those of an actual test floor. During production-level testing, probe CRES also will be affected by other contact-pad contaminants such as dirt, passivation residue, or adherent oxides.

a-Spot Temperatures and Probe-Tip Oxidation

Semi- or nonconductive films composed of tungstenates and adherent aluminum oxides cover portions of the probe-tip contact area (Figure 1b). As the size of the nonconductive regions increases, all the current is forced through the a-Spots, and the current density actually can exceed 106 A/m.2 Consequently, significant Joule heating occurs at the a-Spots and in the immediate vicinity.6

The a-Spot temperature is defined by the voltage drop across the contact interface and the material properties of the conductors. A first-order approximation for the a-Spot temperature (Ta-Spot) can be determined from:2

where: U = voltage drop

TBulk = ambient temperature

a = temperature coefficient of resistivity

r = bulk resistivity

l = thermal conductivity

First approximations of the a-Spot temperature are shown in Figure 3a.

The metallic-alloy, Pd-alloy, and BeCu probes demonstrated low CRES values so the a-Spot temperatures were below 125°C. These materials probably benefit from a self-cleaning phenomenon where scrubbing removes physisorbed contaminants.8

Upon microscopic inspection, adherent Al from the pads was observed on the metallic-alloy, Pd-alloy, and BeCu contact surfaces. Increased probe CRES typically is attributed to Al2O3 film formation; however, these results indicate the presence of additional mechanisms that may significantly contribute to unstable probe CRES at elevated temperatures.

Without a forcing current, W and WRe chemical reactivity results in chemisorbtion of Al atoms from the pad to the tip surface.1 In the present experiment, the CRES of the 30°C W and WRe probes remained below 1.5 W , and the localized a-Spot temperatures did not exceed 70°C.

Conversely, the a-Spot temperatures of the 85°C probes crossed the W-oxidation threshold after 25k touchdowns (Figure 3b). By considering the W oxidation characteristics, elevated test-temperature effects on the a-Spot properties can be assessed (Table 1).

Between 30°C and 300°C, W follows a temperature-dependent parabolic oxidation law.9 Before the probes ever touched the Al wafer, the initial CRES on a tungsten-carbide checkplate at room temperature was from 6 W to 12 W . A light, abrasive cleaning reduced the values below 1.5 W , indicating the existence of an insulative film on virgin tips.

At 85°C, the tungstenate layer forms on the tip surface more rapidly than at 30°C. Oxygen and other gases continually diffuse into the interfacial gap and attack the a-Spots.10 As the amount of adherent oxides and contaminants increases, localized temperatures due to Joule heating will rise (Table 1 and Figure 3b). Once a voltage drop of 140 to 150 mV occurs, semiconductive tungsten oxides start to form around the a-Spots.

Gradual a-Spot oxidation can be used to explain periods of increasing resistance.11 While the a-Spot temperature is below the Al melting point, CRES recovery is attributed to metal softening in the conducting areas. At the tungstenate allotropic transformation temperature (500°C), the oxide layer cracks, allowing regions of intermetallic contact.12 Once the Al melting point is exceeded, the intermetallic contact regions sink together. Increased a-Spot size causes a reduction in CRES, the localized temperature falls instantly, and the microscopic regions of molten material solidify.

Less than a micron away from the a-Spot, the temperature will be slightly above the ambient test temperature.7 The cyclic CRES drops shown in Figure 2b may be caused by this a-Spot melting phenomenon. If the CRES recovery fails to keep pace with localized oxidation growth, the interfacial voltage drop will cause the a-Spot temperatures to increase.11 High forcing currents and contaminants present during production-level test will exacerbate these effects.

The CRES differences between the W and WRe probes were attributed to the addition of rhenium (Re). Alloying W with Re changes the interfacial energy, and continuous oxide films do not form at the grain boundaries.13 Without this continuous film, the WRe probes will make a more consistent electrical contact.

Summary and Conclusions

The CRES of W and WRe probes tested at 85°C increased dramatically after a few thousand touchdowns; however, for probes tested at 30°C, it remained low (<1.5 W ). Such CRES instability is not due solely to adherent Al and Al2O3 since localized probe-tip oxidation of W and WRe significantly contributes to CRES variations during wafer test. Various tungstenates form on W and WRe probe tips at room and elevated temperatures according to a parabolic oxidation law.

At any test temperature, some Al and Al2O3 from the wafer probably adhere on the probe tips and form insulative or semiconducting films. In the test environment, the high interfacial current densities (>106 A/m2) that occur with low forcing currents will cause electromigration, increased constriction resistance, and localized power dissipation. Rapid oxide growth on the probe tips further reduces the a-Spot size, causing excessive localized Joule heating that initiates oxide growth around the a-Spot periphery and increasing CRES.

Although Al and Al2O3 adhered to the tip-contact surfaces, the metallic-alloy, Pd-alloy, and BeCu probes did not suffer from oxidation mechanisms such as the W and WRe probes do. At 85°C, these three materials demonstrated low and stable contact resistance values independent of touchdown.

During preliminary production-level beta testing on actual devices, NewTek Probes had fewer continuity failures, needed less cleaning, and experienced less downtime.14 Although some infrastructure and procedural changes will be required on the test floor, the new metallic-alloy probes will provide a viable probe-material alternative for peripheral and vertical probe applications where low contact forces and stable CRES are critical.

Acknowledgments

The authors acknowledge the help of Carol Whann, Paul Elizondo, and Steve Beaver of Micro-Probe, San Diego, CA; Scott Lindblad and John Strom of Applied Precision, Isaquah, WA; and David Monroe, Ph.D., and Scot Swanson of Sandia National Laboratories, Albuquerque, NM.

References

Molkenthin, J.P., et al., Wear of Materials, 10, 1991.

Logan, J.K., EE-Evaluation Engineering, 8, 1985.

Carbonéro, J.L., Morin, G., and Cabon, B., IEEE Transactions. Microwave Theory and Tech., 43, 1995.

Timosheko, S.P., and Goodier, J.N., Theory of Elasticity, McGraw Hill Book Company, 1934.

Strom, J., IEEE Proceedings of the Southwest Test Workshop, 1998.

Holm, R., Electric Contacts, Springer-Verlag, 1967.

Greenwood, J.A., and Williamson, J.B.P., Proceedings of Royal Society, A246, 1958.

LaRosa, R., and Griffing, S., Microelectronic Manufacturing and Testing, 9, 1989.

Gulbrasen, E.A., and Wysong, W.S., Transactions. AIME, 175, 1948.

Williamson, J.B.P., IEEE Proceedings Holm Conference Electrical Contacts, 1981.

Slade, P.G., IEEE Transactions Components, Hybrids, Manufacturing Technology, CHMT-9, 1986

Barth, V.D., and Rengstorff, G.W.P., DMIC Report 155 ,1961.

Lement, B.S., and Perlmutter, I., J. Less-Common Metals, 2, 1960.

Broz, J.J., and Rincon, R., IEEE Proceedings of the International Test Conference, (at press 1999).

About the Authors

Jerry J. Broz, Ph.D., earned a doctorate in mechanical engineering from the University of Colorado. Since 1993, he has been the director of R&D at Advanced Probing Systems. Previously, he was a research scientist at the Denver Research Institute. Advanced Probing Systems, P.O. Box 17548, Boulder, CO 80308, (303) 939-9384, e-mail: [email protected].

Reynaldo M. Rincon is the probe coordinator for ASP product groups at Texas Instruments. During 20 years at TI, he also has worked as the ASIC product engineering manager and the ASP test engineering manager. Mr. Rincon received a B.S.E.E.T. from the University of Houston. Texas Instruments, 13020 Floyd Rd., MS 3616, Dallas, TX 75265, (972) 917-4303, e-mail: [email protected].

Voltage

Drop

(mV)

Estimated

a-Spot

Temperature

Effects on Contact Materials

0

85

Initial tungsten probe needle oxidation. The oxide layer (WO or WO2) is approximately 50 to 60 Å thick and grows according to a parabolic rate law. This layer appears to be easily removed by abrasive cleaning.

140 to 150

300 to 325

W-oxidation threshold. The adherent, compact film of higher tungstenates forms on the probe tip. This oxide is a protective, porous, blue-colored film controlled by oxygen diffusion.

220 to 240

500

Allotropic transformation occurs, and the oxide begins to crack and becomes unprotective. The oxidation growth along these cracks no longer is self-healing, resulting in metal-to-metal contact. Intermediate oxide film of lower tungstenates also begins to form on the surface.

280

660

Melting temperature of Al.

295 to 300

700 to 715

Onset of rapid oxidation. Loose WO3 forms on the outer surface of the intermediate layer of the adherent lower tungstenates.

400 to 410

1,000

W-softening temperature. Above this temperature, evaporation of the tungstenates rapidly increases. Semiconducting oxides, such as W3O8, form through the reaction of W and WO3.

480 to 490

1,100 to 1,200

Onset of catastrophic oxidation in the air. The W oxides evaporate as soon as they are formed.

580

1,470

Melting temperature of WO3.

960 to 970

2,500 to 2,600

Boiling temperature of Al.

1,100 to 1,200

3,100 to 3,200

Softening temperature of tungsten.

Copyright 1999 Nelson Publishing Inc.

September 1999

Comments

To join the conversation, and become an exclusive member of Electronic Design, create an account today!