Interfacial fracture toughness measurement for thin film interfaces
Introduction
Delamination of intrinsically or residually stressed thin films is commonly encountered in microelectronics and MEMS systems. Thin films typically accrue stresses through microstructural variations caused by physical vapor deposition, thermally induced stresses imposed due to thermal mismatch, and/or externally introduced forces. These stresses can reach upwards of 1 GPa and can easily exceed the strength of the metal thin film interface. To help prevent delamination, often titanium or chromium is used as an adhesive layer. Understanding and preventing delamination between Ti (or any other metal) and a substrate is the motivation of this paper.
This study discusses the development of a new experimental method to measure the interfacial fracture toughness for thin film interfaces. Knowledge of the mode mixity dependent interfacial fracture toughness (Γ) is necessary to predict if delamination will occur. However, measuring Γ is a challenge for thin film interfaces. Typical testing methods such as bimaterial cantilever, microscratch, peel, bulge, or edge lift-off are limited to organic films, cause complex stress fields, can only measure a single mode mix, or cannot achieve the large energy release rates (G) typical of metal thin film interfaces. This approach, based on the work of Bagchi et al. [1], is called the modified decohesion test (MDT), and proposes to eliminate these shortcomings of current testing methods. It is a semi-empirical test that is capable of determining the interfacial fracture toughness using one test specimen, uses common IC fabrication techniques thereby ensuring that the interface is representative of a microelectronic package surface, has a mechanics-based solution, and allows full variation of the mode mix.
Validation of this test is performed through a finite element model (FEM) analysis examining several different scenarios of the test. Interfacial fracture toughness results will be shown for a Ti/Al2O3 and Ti/Si interface prepared using standard IC fabrication techniques.
Section snippets
Test methodology
A proper test to measure interfacial fracture toughness should be able to account for the energies accompanying fracture beyond that of surface energy (or work of adhesion, Wad) and be able to determine how the mode mixity (ψ) affects Γ. Also, the test must meet some other minimum criteria: (a) sample preparation should be simple, (b) deformations should be elastic so the problem can be modeled, (c) a mechanics-based analytical solution exists, and (d) test is inexpensive and efficient.
Bagchi
Modified decohesion test
Given the deficiencies of DT discussed above, some modifications are proposed. When absent of extrinsic loading, energy release rate is equal to the ratio of change of strain energy (dU) to the change in crack surface area (dA).
Inspection of Eq. (2) shows modifying the change in energy or the change in crack surface area will vary the energy release rate. In the decohesion test, G is varied by varying dU and keeping dA constant (i.e., changing the superlayer stress or superlayer
Numerical validation
Extending the non-adhesive layer underneath the interface layer should modify G proportionally to the bonded area fraction (1−ζ) and not affect the mode mix. A three-dimensional finite element analysis is used to confirm that the test follows theoretical expectations.
Analytical method
An important aspect of the MDT is that it has an analytical solution that can accurately relate the test parameters (film thickness, stress, and geometry) to the interfacial fracture toughness metrics. Since the MDT is primarily used for metal/substrate interfaces, the superlayer intrinsic stress is typically on the order of GPa. For low to moderate mode mixes (i.e., dominant bending mode), this will cause the test strips to deflect many times their thickness. This requires the analytical
Test design
The analytical model used to provide the mechanics-based solution for the MDT may also be used to design and make an initial guess at the parameters for the experiment. By analytically examining the combinations of different superlayer and interface layer thickness and superlayer stress, surface plots can be generated that give (a) the range of energy release rates produced on a single sample, (b) the mode mixity generated, and (c) energy in the superlayer. Using these surface plots, the
Titanium/silicon interface
The MDT was applied to characterize the interface fracture toughness of a DC sputter deposited titanium thin film on a test grade silicon substrate. A very thin gold layer (40 nm) was used as the non-adhesive layer, titanium was the interface layer material, and chromium was used as the superlayer. All three metals were deposited using a Unifilm™ DC sputter deposition chamber. This deposition chamber achieves 99% film thickness uniformity and very little biaxial planar stress anisotropy.
Conclusions
A modification to the decohesion test, called the modified decohesion test, has been developed to measure thin film on substrate interfacial fracture toughness (Γ). The original test uses stressed metal layers to promote interfacial fracture. The modification of varying the change in crack surface area, dA, rather than the change in energy, dU, allows the ability to create a range of energy release rates on a single substrate. The modification of using a non-adhesive material that adheres
Acknowledgements
The authors would like to thank Dave Fork of PARC, Inc. and Don Smith of NanoNexus, Inc. for their valuable input and guidance in designing the modified decohesion test. The authors also appreciate valuable discussions on the design, fabrication, modeling and use of microcontact springs with all of the consortium members at Xerox PARC, NanoNexus, and the CASPaR Lab at Georgia Tech. In particular, the contributions of F.C. Chong, Joe Haemer, Jeff Wise, Reagan Wynn, and Frank Swiatowiec are
References (12)
- et al.
Geometrically nonlinear stress–deflection relations for thin film/substrate systems
Int. J. Eng. Sci.
(1993) - et al.
The influence of interfacial reactions on the fracture toughness of Ti–Al2O3 interfaces
Acta Mater.
(1995) - et al.
A new procedure for measuring the decohesion energy for thin ductile films on substrates
J. Mater. Res.
(1994) - et al.
Interface crack between two elastic layers
Int. J. Fract.
(1990) - Modi M, Sitaraman SK. Effect of adhesive layer properties on fracture in thin film high density interconnects. In:...
- et al.
A method for calculating stress intensities in bimaterial fracture
Int. J. Fract.
(1989)
Cited by (26)
Mechanical Properties of Solar Cell Structures
2023, Comprehensive Structural IntegrityInterfacial mixed-mode fracture toughness characterization of flexible layered materials
2020, Engineering Fracture MechanicsCitation Excerpt :Moreover, the effect of friction load between the glass plate and sample was neglected (<1N) due to the small transverse load and low friction coefficient. Calculation of G is complicated due to the dissipative behavior of the substrate plasticity during the crack extension and the edge effects in thin films [48], and closed form expressions like those used for mode I and mode II standards may not be feasible. Therefore, a numerical 2D plane-strain model was employed to calculate G as shown in Fig. 4.
Effect of temperature and humidity conditioning on copper leadframe/mold compound interfacial delamination
2020, Microelectronics ReliabilityCitation Excerpt :Important consideration should be given to mechanical reliability of such interfaces in microelectronic packages at high temperatures and humidity conditions to lower costs and improve design and process guidelines [10,11]. Numerous experimental and analytical studies have been conducted to investigate the delamination issues in bimaterial interfaces such as copper/epoxy molding compound, copper/dielectric, die/die attach, die/underfill, silicon/metal, and glass/polymer systems [12–29]. Studies have been conducted to evaluate how residual stresses and mode mixity affect interfacial strength [12,13,30,31].
Prediction of interfacial adhesion strength of nanoscale Al/TiN films passed through patterned BEOL interconnects
2015, Materials Science in Semiconductor ProcessingCitation Excerpt :Aggressive reduction of the dielectric constant of insulations and barrier layer thickness to the nanoscale order remarkably increases the risk of device failure resulting from interfacial fracture. Consequently, adhesion of dissimilar media in stacked thin films have been widely investigated to enhance interfacial adhesive strength, such as through nanoindentation [2], blister measurements [3], and modified decohesion test [4]. Four-point bending test (4-PBT) is one of the regular methods used to measure de-bonded fracture toughness of thin coatings [5–9].
Overview of interfacial fracture energy predictions for stacked thin films using a four-point bending framework
2013, Surface and Coatings TechnologyCitation Excerpt :Thus, assessing the interfacial adhesion of a thin coating adhered to the adjoining material by different measured techniques is needed and should be widely developed [1,2]. Several methods, such as blister test [3], nano-indentation [4–6], modified decohesion test [7], nanoscratch test [8,9], double cantilever beam test [10,11], laser spallation test [12,13], and four-point bending test (4-PBT) [14–25], have been utilized for quantitatively extracting the adhesion parameters of dissimilar films. An appropriate model should be selected for the indentation and nanoscratch methods after analyzing the related failure mechanisms of stacked films; consequently, 4-PBT has been popularly applied in electron packaging [14–16] and back-end-of line (BEOL) interconnection structures [17] due to their accurate and reproducible measurements over a wide range of bond characteristics, strength, and fracture toughness compared with other testing techniques.
Investigation of interconnect design on interfacial cracking energy of Al/TiN barriers under a flexural load
2013, Thin Solid FilmsCitation Excerpt :These issues have given rise to an urgent need to enhance interfacial adhesive strength, a stringent standard for guaranteeing interconnect reliability. Several approaches are used to examine the interfacial adhesive strength of stacked films; these methods include nano-indentation [2], blister tests [3], the modified decohesion test [4], and the nanoscratch test [5,6]. Among these techniques, the four-point bending test (4-PBT) is emerging as the preferred technique for studying the adhesion of interconnects in semiconductor devices [7–12].