Main NIST Website Materials Reliability Division

 
Mechanical Characterization of Thin Films and Interconnects

Objectives

We perform research benefiting those interested in materials for the microelectronics, optoelectronics, and nanomaterials industries. Our efforts concentrate on the effects of dimensional constraints on mechanical performance and reliability. The primary objective is to develop experimental measurements to understand and quantify the mechanical behavior of materials at the nanoscale, especially far-from-equilibrium behavior such as response to high forces and high stresses, and stability of nanoscale structures over time against monotonic and cyclic mechanical, electrical and thermal stresses.

Background

The electronics industry has been particularly aggressive in recent years in commercializing nanoscale structures such as advanced transistors and interconnect systems manufactured on silicon wafers. This industry regularly publishes a roadmap, indicating that successful integration of new materials is a critical element needed to keep up the pace of product improvement. As the active device and interconnect structure become more complex and smaller, the materials metrology problems become more severe. Complex materials issues such as measurement of mechanical, thermal and electrical properties of materials in nanoscale structures, and verification of the models used to predict the behavior of these structures lie at the core of our work.

Integral to designing for reliability of interconnects is accurate knowledge of mechanical properties such as yield strength (sys), ultimate strength (sUTS), and ductility, as well as thermal fatigue response. Reliability issues include various modes of mechanical failure (fracture, delamination, fatigue cracking, void formation). Accelerated lifetime testing, to verify that a design has the expected reliability, traditionally uses thermal cycling as a key technique. The issue is then, does the acceleration scheme actually test relevant failure mechanisms. A key limitation of nearly all previously developed measurement methods is that they can be used only on blanket wafers or special test structures, as opposed to patterned wafers. This is problematic, as it is well-established that the microstructure, and hence the behavior, of patterned structures is fundamentally different from that of blanket material, particularly for damascene processing. A thermal-cycling-based scheme for mechanical testing overcomes these limitations for nanostructures of any size that are built on substrates or that can be rigidly attached to substrates.

Approach

An electrical approach to material measurements is most appropriate for the target microelectronics industry. Our method is based on the application of controlled joule heating to films and interconnects, by means of a 4-point probe test. Conditions are such that electromigration does not take place. Rather, heat is generated and dissipated within each power cycle. Cyclic thermal strain (De) is caused by the mismatch in coefficient of thermal expansion (Da) between the film/interconnect and the surrounding materials, due to the temperature change (DT) induced by each power cycle:

De = DaDT

Knowledge of Young’s modulus (E) enables a determination of stress versus time, as shown in figure 1. All tests are run using a.c., under conditions of low frequency (100 Hz) and high current density (j) (~ (6 to 30) MA/cm2).

The main value of this approach will be its capability to provide quantitative strength values for material characterization; a secondary value is as a method of local, rapid thermal cycling. Our goal for this year is to further demonstrate that the AC stressing scheme causes specimen failure through a predominantly mechanical failure mechanism, and to verify the quantitative link between electrical resistance, temperature, and stress. Clearly the quantitative details of the stress and strain fields and the failure mode depend on the specimen configuration. Particularly important is the constraint of the specimen imposed by the surrounding structure. For instance, a line buried in a stiff dielectric would be constrained more than a line buried in a compliant dielectric, and both would be constrained differently than an exposed line. The utility of the AC stressing technique depends on understanding these constraint cases and their associated failure modes.

Figure 1: Current, temperature, and stress in the "mechanical testing by AC" experiment.

Results

A key milestone was recently reached. Both microtensile and AC fatigue tests were performed on a single thin film material. Analysis of the results, in particular accounting for the residual stress in the electrical specimens, showed a good correspondence between the fatigue strength and the ultimate tensile strength. This milestone built on electrical tests of a few different thin-film materials that showed that the “S-N curve”, which portrays the increase in lifetime with decreasing cyclic stress amplitude, had not only the same form but also similar parameter values for the AC fatigue data as the S-N curve for macroscale mechanical fatigue data on structural materials; data from an electrical test are shown in figure 2, along with a correction for effects of non-zero mean stress. The fatigue exponent in the Basquin equation, which is the most important parameter in fitting such curves, was in the same range for electrically-tested films and structural materials as reported in the literature. The work will appear in a forthcoming issue of Metallurgical and Materials Transactions A.

Figure 2: Electrical data produced a fatigue strength of about 260 MPa for an aluminum film; the corresponding microtensile result for sUTS was about 240 MPa.

Selected Publications

R. R. Keller, N. Barbosa III, R. H. Geiss, D. T. Read, "An Electrical Method for Measuring Fatigue and Tensile Properties of Thin Films on Substrates," Key Engineering Materials, in press (2007).

N. Barbosa III, R. R. Keller, D. T. Read, R. H. Geiss, R. P. Vinci, "Comparison of Electrical and Microtensile Evaluation of Mechanical Properties of an Aluminum Film," Metallurgical and Materials Transactions A, in press (2007).

R. R. Keller, R. H. Geiss, N. Barbosa III, A. J. Slifka, D. T. Read, "Strain-Induced Grain Growth during Rapid Thermal Cycling of Aluminum Interconnects," Metallurgical and Materials Transactions A, in press (2007).

D. T. Read, R. R. Keller, N. Barbosa III, R. H. Geiss, "Nanoindentation Round Robin on Thin Film Copper on Silicon," Metallurgical and Materials Transactions A, in press (2007).

R. H. Geiss, R. R. Keller, D. T. Read, Y. -W. Cheng, "TEM-Based Analysis of Defects Induced by AC Thermomechanical versus Microtensile Deformation in Aluminum Thin Films," Proc. Mater. Res. Soc. Symp. v. 863, 283-288 (2005).

R. R. Keller, R. H. Geiss, Y. -W. Cheng, D. T. Read, "Microstructure Evolution during Electric Current Induced Thermomechanical Fatigue of Interconnects," Proc. Mater. Res. Soc. Symp. v. 863, 295-300 (2005).

R. R. Keller, R. H. Geiss, Y. -W. Cheng, D. T. Read, "Electric Current Induced Thermomechanical Fatigue Testing of Interconnects," AIP Conf. Proc. 788, 2005 Int'l. Conf. Characterization and Metrology for ULSI Technology, (2005), 491-495.

R. Mönig, R. R. Keller, and C. A. Volkert, "Thermal fatigue testing of thin metal films," Rev. Sci. Instrum., 75, 4997-5004 (2004).

Materials Reliability Division

NIST Materials Science and Engineering Laboratory

Last modified on May 18, 2007

The National Institute of Standards and Technology
is an agency of the U.S. Department of Commerce's
Technology Administration

Privacy Statement/Security Notice
Disclaimer | FOIA