- Email: [email protected]
Magnetically actuated peel test for thin films
Thin Solid Films 520 (2012) 3987–3993
Contents lists available at SciVerse ScienceDirect
Thin Solid Films journal homepage: www.elsevier.com/locate/tsf
Magnetically actuated peel test for thin films G.T. Ostrowicki, S.K. Sitaraman ⁎ George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, Georgia, USA
a r t i c l e
i n f o
Article history: Received 9 June 2011 Received in revised form 24 January 2012 Accepted 26 January 2012 Available online 1 February 2012 Keywords: Peel test Pull-off test Interfacial fracture Thin film Magnetic actuation
a b s t r a c t Delamination along thin film interfaces is a prevalent failure mechanism in microelectronic, photonic, microelectromechanical systems, and other engineering applications. Current interfacial fracture test techniques specific to thin films are limited by either sophisticated mechanical fixturing, physical contact near the crack tip, or complicated stress fields. Moreover, these techniques are generally not suitable for investigating fatigue crack propagation under cyclical loading. Thus, a fixtureless and noncontact experimental test technique with potential for fatigue loading is proposed and implemented to study interfacial fracture toughness for thin film systems. The proposed test incorporates permanent magnets surface mounted onto micro-fabricated released thin film structures. An applied external magnetic field induces noncontact loading to initiate delamination along the interface between the thin film and underlying substrate. Characterization of the critical peel force and peel angle is accomplished through in situ deflection measurements, from which the fracture toughness can be inferred. The test method was used to obtain interfacial fracture strength of 0.8-1.9 J/m 2 for 1.5-1.7 μm electroplated copper on natively oxidized silicon substrates. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Microelectronics, microelectromechanical systems, photovoltaic, and other devices consist of several thin film layers that are made of dissimilar materials. Such multilayered thin film systems are often prone to interfacial delamination under various loading conditions resulting in premature failure of the devices. Thus, interfacial delamination is an important reliability concern for a wide range of applications, and numerous experimental test techniques have been proposed to characterize the fracture resistance of an interface. Materials and the interfaces they share often have different mechanical properties at the micro/nano scale, and thus appropriate test methods are needed to characterize interfacial fracture between thin films. The traditional sandwich specimen tests (e.g. four-point bend, Brazil-nut) generally require non-standard fabrication and do not correlate to films as deposited in production. More recently developed techniques such as the scratch, indentation, and blister tests require either (1) direct contact with the test sample near the crack tip resulting in complicated stress fields, and/or (2) sophisticated fixturing and experimental setup. The stressed superlayer technique, proposed by Bagchi [1] and later extended by Modi [2] and Zheng [3], is a convenient fixtureless and noncontact method to characterize interfacial fracture. However, like the previously ⁎ Corresponding author. Tel.: + 1 404 894 3405; fax: + 1 404 894 9342. E-mail address: [email protected] (S.K. Sitaraman). 0040-6090/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2012.01.042
mentioned techniques, the load driving delamination is monotonically applied and cannot be easily modified to provide cyclical loading for a fatigue crack propagation study. An innovative delamination test is thus proposed that addresses the above highlighted shortcomings of alternative techniques. In this approach, a partially released thin film structure is gradually displaced out-of-plane until the peel force exerted on the adherend is sufficient to initiate delamination. The actuation force is supplied by the magnetic interaction between a permanent magnet assembled onto the film structure and a nearby electromagnet. With precise control of the current supplied to the electromagnet (and thus force exerted on the assembled permanent magnet), a noncontact, load-controlled peel test is performed. The proposed test has several key elements: (1) electromagnetic actuation provides the driving force needed for delamination propagation, and thus does not require any fixtures to apply loads; (2) an optical profiler monitors the delamination propagation in situ; (3) a wide range of loading amplitude and frequency is obtainable by adjusting the current supplied to the electromagnet generating the external magnetic field; (4) sample preparation involves only two photolithography steps and results in representative as-deposited films; (5) tens to hundreds of test samples can be batch fabricated onto a single wafer; (6) applied loading and delamination detection methods are suitable for both monotonic and fatigue crack characterization; (7) samples can be placed within an environment of excessive temperature and humidity while the load is modulated by an external magnetic field.
3988
G.T. Ostrowicki, S.K. Sitaraman / Thin Solid Films 520 (2012) 3987–3993
Fig. 1. (a) 3D illustration of MAPT specimen with sacrificial layer intact. (b) After removal of sacrificial layer, specimen is placed above electromagnet at fixed distance δ, where magnetic repulsive force lifts released features and imposes a peel force along the anchoring film strips.
One proposed test design is the Magnetically Actuated Peel Test (MAPT) shown in Fig. 1. Here, three strips of thin film meet at a central pad on which a permanent magnet is attached. The area underneath the central pad is released after removal of a thin intermediate sacrificial layer of diameter 2a (Fig. 2-i). The specimen is then situated above an electromagnet of opposing magnetic alignment (Fig. 1b). As sufficient electrical power is supplied, the repulsive magnetic force will cause the central pad to rise and impose a peel force along the three anchoring film strips (Fig. 2-ii). Once the peel force reaches a critical value, delamination along the film strip/substrate interface will progress (Fig. 2-iii). One significant advantage of the MAPT design is that the permanent magnet maintains its orientation during actuation and thus the applied
Fig. 2. Stages of MAPT: (i) at rest after removal of sacrificial layer; (ii) initial displacement due to application of magnetic force; (iii) onset of peel at critical load.
force is uniform and strictly controlled by the electromagnet current. As the area above the specimen is free of obstruction, an optical profiler can be used to capture the out-of-plane deflection and thus determine both the strain in the released film and the progression of delamination. 2. Peel test analogy The traditional peel test has been a widely studied technique to characterize interfacial fracture strength. As shown in Fig. 3a, a flexible film is peeled off an underlying substrate by a force F acting along an angle θ. However, when applied to very thin films with thickness less than several tens of microns, the peel test can result in rupture of the film due to severe deformation either at the crack tip or near the mechanical grips. This kind of cohesive failure can be mitigated by using a modification of the peel test called the pull-off test introduced by Gent and Kaang [4], where a vertical load P applied at the center of a debonded strip imposes a simultaneous peel force on both adhered ends of the film (Fig. 3b). Here the low angle of peel limits the bending stresses near the crack tip, and the load is typically applied by a horizontal pin underneath the debonded strip rather
Fig. 3. (a) Peel test. (b) Pull-off test.
G.T. Ostrowicki, S.K. Sitaraman / Thin Solid Films 520 (2012) 3987–3993
3989
than mechanical grips. The MAPT design is most analogous to the pull-off test, except that there are three debonded strips rather than two, and these strips are coupled to the bottom surface of a cylindrical permanent magnet supplying the vertical load. Assuming symmetry, the individual peel force F is related to the total magnetic load P by the relation P ¼ 3F sin θ:
ð1Þ
The thermodynamic approach to fracture characterization requires performing an energy balance during a crack growth process, and many theoretical models of interfacial fracture specific to the peel and pull-off tests can be found in the literature. A particularly simple and relevant model as described by Williams [5] assumes a flexible peel strip under uniform tension on a rigid substrate, and results in an expression for the interfacial fracture energy rate Γ0 as
Fig. 4. Fabrication process of MAPT specimen.
ε
F Γ 0 ¼ ð1− cos θ þ εÞ−h ∫ σ ⋅dε; b
ð2Þ
0
where F is the peel force, b is the strip width, θ is the peel angle, ε is the tensile strain, h is the strip thickness, and σ is the tensile stress. When an elastic constitutive model is used for the peeling film, the integral in (2) represents the stored elastic strain energy, in which case Γ0 is equivalent to the critical energy release rate Gc. However, when an elastic–plastic constitutive model is used for the peeling film, the integral includes the contributions of both the elastic strain energy and the plastic work. A nonlinear elastic–plastic constitutive model for 2 μm thick Cu film was used in this work [6]. Calculations of the critical peel force, peel stress, and fracture energy rate were done with respect to θavg, the average peel angle amongst the three strips per MAPT specimen. Since the focus of this work is the development and demonstration of an experimental test technique, an evaluation and comparison of alternative peel test models, including both analytical and finite element approaches, will be published elsewhere. 3. Experimental procedure 3.1. Sample preparation The initial proof-of-concept MAPT specimens detailed in this work feature electroplated copper film strips peeled off a natively oxidized silicon substrate. The microfabrication processes used to define the MAPT test film pattern are described in detail in [7] and are summarized below. A thin sacrificial polymer was patterned to define the initial pre-crack between the substrate and subsequently deposited metal film (Fig. 4a). Cu was then sputtered 500 nm thick and electroplated to total thickness of 1.5-1.7 μm. The encapsulated polymer was reflowed to achieve a dome shape, and after a second photolithography step the metal film was patterned via wet etch into the MAPT strip design (Fig. 4b). After removal of the sacrificial polymer (Fig. 4c), uncured Epibond® 7275 surface mount epoxy was then printed though a 3 mil stainless steel stencil onto the released central Cu pad. Finally, a custom alignment setup was used to precisely place a nickel-plated neodymium (NdFeB-42SH) magnetic cylinder (1/16”D × 1/16”H) onto the central pad and the assembly is completed with a 100 °C cure (Fig. 4d). The high temperature grade neodymium showed no loss of magnetism after the cure bake. The poles of the permanent magnet are at the ends of the cylinder such that they are magnetically aligned out-of-plane after assembly. Batch assembly of magnets is shown in Fig. 5. In the mask designs used in this research, b varies from 25 to 200 μm, and initial delaminated strip length α0 (i.e. distance from Cu central pad edge to outer sacrificial edge) varies from 250 to 1000 μm.
3.2. Magnetic force calibration Delamination in the MAPT design is driven by the force of magnetic repulsion, and this force is characterized using the experimental setup illustrated in Fig. 6. Here a Magnetic Sensor Systems E-66-38 tubular electromagnet is fixed at a specified distance δ from the NdFeB magnet resting on a high precision scale with ~ 1 μN sensitivity. Accounting for the weight of the permanent magnet, the downward force is directly captured by the scale. In this setup, the two critical parameters are the separation δ and the voltage potential V applied to the electromagnet. A voltage range of 0–13 V was applied at separation distances from 1.0 to 2.2 mm as shown in Fig. 7a, with maximum force achieved to be ~ 16 mN. Positive values of force indicate magnetic repulsion, whereas negative values indicate magnetic attraction. At low voltage, the ferromagnetic core of the electromagnet has a weak native polarity and is thus attracted to the permanent NdFeB magnet regardless of the direction of alignment. However, once ample current is supplied in the proper direction, the electromagnet will have a stable opposing polarity and repel the NdFeB magnet when at sufficiently large separation distances. It was interesting to observe that at very close distances the magnets would experience a net attractive force even at high voltage levels. This occurs because of the size effects of the magnetic field interaction between the small NdFeB magnet and the larger electromagnet. The separation gap is fixed during the peel test, and δ = 1.8 mm was chosen as an optimized distance due to high force potential, large operable voltage range, and low sensitivity to variations in the gap distance. At this distance, the magnetic force was found to be a linear function of V, transitioning from magnetic attraction to repulsion at ~3 V (see Fig. 7b). Each MAPT specimen magnet was calibrated individually in this manner with near identical results. For the peel test, the actual gap between the permanent magnet and electromagnet is δ + Δ, where Δ is the vertical displacement of the central Cu pad. However, the magnitude of Δ was less than 100 μm in the experiments, and so remained within the calibration's range of accuracy. 3.3. Peel test execution During the peel test, an electromagnet of opposing magnetic alignment is placed underneath the MAPT specimen such that actuation is a result of a repulsive magnetic force (Fig. 1b). Monotonic peel test experiments were conducted while a Veeco Wyko® NT-2000 optical profiler was used to measure the sample surface topography in situ. Surface images capturing the deflection of the film and
3990
G.T. Ostrowicki, S.K. Sitaraman / Thin Solid Films 520 (2012) 3987–3993
Fig. 5. Assembled MAPT specimens.
delamination propagation were taken at several discrete intervals of applied voltage. Shown in Fig. 8 are example images from one test specimen taken at significant points during the delamination
test. At 3.0 V the sample experiences essentially no force and is in the initial at rest state. At 6.0 V the strips are raised and pulled taught but no significant delamination is observed. At 7.0 V delamination is
G.T. Ostrowicki, S.K. Sitaraman / Thin Solid Films 520 (2012) 3987–3993
3991
the interfacial crack length and peel angle were obtained as a function of the applied magnetic force. Each MAPT specimen provides displacement data for three peel strips simultaneously. Assuming evenly distributed force amongst the three strips, the critical peel force and associated peel angle are determined once delamination of at least one peel strip reaches steady-state. Since no images could be taken during the dynamic peeling process, it is assumed that any incrementally higher force beyond the last stable deflection measurement would induce steady state delamination. Consequently, the highest sustained peel force is reported here as a conservative estimate for the critical peel force. Peel test results for five MAPT specimens are presented in this work. Due to two different fabrication techniques to remove the sacrificial material, two specimens were heated within an inert environment at 250 °C for 2 hours before magnet assembly while the remaining three specimens were not heat treated.
Fig. 6. Experimental setup for magnetic force calibration.
observed and slowly propagates until 9.1 V, after which steady state delamination results in total failure of the peel test. The slope of each peel strip along with the location of the crack front was determined from each of these profiler images. Thus, along with the previous force calibration results, measurements of
4. Results and discussion Processing of the optical profiler images resulted in the typical peel angle vs. magnetic force relationship shown in Fig. 9. Since the measured peel angles of the three strips may be slightly different from one another at a given voltage, the average peel angle was used to estimate the peel force. In all but one case, steady-state peeling was reached immediately following delamination initiation. In the one exception, a heat-treated MAPT specimen was observed to exhibit unstable crack growth over a range of applied force values until steady-state peeling was reached at the critical load (Fig. 10). The results of the MAPT experiments are summarized in Table 1. In order to appropriately compare peel strips of varying dimensions, the magnetic pull-off force and tangential peel force are expressed per strip cross section as P/bh and σ = F/bh, respectively. The critical fracture energy rates were calculated by inputting the average measured peel angle and average peel force (and stress) at fracture into Eq. (2). Using the nonlinear constitutive model for Cu, the interfacial fracture energy rate of the heated specimens (samples 1–2) was calculated to be ~0.8 J/m 2, whereas the non-heat-treated specimens (samples 3–5) exhibited slightly higher values of 0.9-1.9 J/m 2. If instead the Cu film is assumed to be linear elastic with a modulus of 110 GPa, the reported interfacial fracture energy rate would increase by ~ 10-20%. These values are comparable to the reported critical energy release rates of ~ 1-10 J/m 2 for ~ 1 μm Cu on SiO2 using other techniques [8,9]. Despite the limited sample size, there appears to be a trend of interface weakening due to heat treatment. While it is suspected that thermal expansion mismatch and Cu film grain coarsening is causing damage at the interface, the conclusive reasoning for this weakened interface, including the testing of a larger sample set, is the subject of ongoing work. 5. Concluding remarks
Fig. 7. (a) Typical force vs. magnet separation gap δ for varying applied voltage to the electromagnet. (b) Typical force vs. electromagnet voltage at fixed δ = 1.8 mm.
The magnetically actuated peel test concept was devised and demonstrated to reliably measure interfacial fracture strength between a thin film and underlying substrate. A sample specimen design featuring pull-off of three film strips allowed for precise and stable loading while an optical profiler captured film deflection and crack propagation. Unlike most conventional techniques, the MAPT design does not require fixtures or physical contact with the test specimen. Interfacial fracture toughness values were obtained for a thin electroplated Cu film on a natively oxidized Si wafer, with preliminary results indicating a weakened interface for heat-treated samples. Interfacial characterization was demonstrated for single-material peeling films with ~ 1.5 μm thickness. It is expected that the MAPT technique can accommodate single-material peeling films as thin as
3992
G.T. Ostrowicki, S.K. Sitaraman / Thin Solid Films 520 (2012) 3987–3993
Fig. 8. Optical profiler images at select intermittent applied voltage during delamination test. Black color indicates areas that are missing out-of-plane height data. Delamination is observed at 7.0 V and propagates until 9.1 V, after which steady-state peeling commenced.
a few hundred nm with no change in the as described fabrication process or test setup. However, in order to test even thinner films (b100 nm), it may be required to use a bi-material peeling film, where the interface of interest is between the substrate and the bottom peel strip material, and the top peel strip material provides structural support. Using such a bi-material design would allow for interfacial characterization of films that are a few nm thick, but would also likely require more sophisticated mechanics analysis to account for the effect of the supporting film on the extracted interfacial fracture parameters.
Fig. 9. Typical peel angle vs. magnetic force.
Although only monotonic fracture is demonstrated in this work, the MAPT test design is ideally suited to conduct interfacial crack propagation experiments under fatigue loading, and is the subject of ongoing research. In such a fatigue test, a cyclic load can be applied to the test specimen by supplying an AC current with desired amplitude and frequency to the driving electromagnet, while optical profiler measurements can be taken intermittently after a set number of cycles in order to capture the delamination propagation. Thus, the crack growth rate can be observed as a
Fig. 10. Non-typical unstable crack growth exhibited for one heat treated MAPT specimen (No. 2).
G.T. Ostrowicki, S.K. Sitaraman / Thin Solid Films 520 (2012) 3987–3993 Table 1 MAPT results for electroplated Cu on natively oxidized Si substrate. Cu width Pull-off Sample Cu Average Average peel FER Γ0 force (P/bh)c peel angle stress σavg number thickness b (μm) (J/m2)* h (μm) θavg (deg) (MPa) (μN/μm2) 1 2 3 4 5
1.7 1.7 1.5 1.5 1.5
100 200 100 100 100
33 36 47 53 74
4.1 3.8 4.0 5.5 5.5
154 184 229 187 255
0.8 0.8 0.9 1.4 1.9
Note: * Fracture Energy Rate (FER) calculated from Eq. (2) using nonlinear Cu constitutive model.
function of the loading amplitude, from which Paris Law parameters can be determined. Acknowledgments This work is supported by the National Science Foundation under grant No. CMMI-0800037 and also by Grant No. ECCS-0901679. The
3993
authors gratefully acknowledge Nathan Fritz and Dr. Paul Kohl (School of Chemical and Biomolecular Engineering, Georgia Institute of Technology) for assistance and helpful discussion regarding sample preparation.
References [1] [2] [3] [4] [5] [6] [7]
A. Bagchi, G.E. Lucas, Z. Suo, A.G. Evans, J. Mater. Res. 9 (1994) 1734. M.B. Modi, S.K. Sitaraman, Eng. Fract. Mech. 71 (2004) 1219. J.T. Zheng, S.K. Sitaraman, Thin Solid Films 515 (2007) 4709. A.N. Gent, S. Kaang, J. Appl. Polym. Sci. 32 (1986) 4689. J.G. Williams, Int. J. Fract. 87 (1997) 265. C.A.O. Henning, F.W. Boswell, J.M. Corbett, Acta Metall. 23 (1975) 177. G.T. Ostrowicki, N.T. Fritz, R.I. Okereke, P.A. Kohl, S.K. Sitaraman, IEEE Trans. Device Mater. Reliab. (2012), doi:10.1109/TDMR.2011.2175927. [8] A.A. Volinsky, N.I. Tymiak, M.D. Kriese, W.W. Gerberich, J.W. Hutchinson, MRS Symp. Proc. 539 (1999) 277. [9] A. Bagchi, A.G. Evans, Thin Solid Films 286 (1996) 203.
Contents lists available at SciVerse ScienceDirect
Thin Solid Films journal homepage: www.elsevier.com/locate/tsf
Magnetically actuated peel test for thin films G.T. Ostrowicki, S.K. Sitaraman ⁎ George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, Georgia, USA
a r t i c l e
i n f o
Article history: Received 9 June 2011 Received in revised form 24 January 2012 Accepted 26 January 2012 Available online 1 February 2012 Keywords: Peel test Pull-off test Interfacial fracture Thin film Magnetic actuation
a b s t r a c t Delamination along thin film interfaces is a prevalent failure mechanism in microelectronic, photonic, microelectromechanical systems, and other engineering applications. Current interfacial fracture test techniques specific to thin films are limited by either sophisticated mechanical fixturing, physical contact near the crack tip, or complicated stress fields. Moreover, these techniques are generally not suitable for investigating fatigue crack propagation under cyclical loading. Thus, a fixtureless and noncontact experimental test technique with potential for fatigue loading is proposed and implemented to study interfacial fracture toughness for thin film systems. The proposed test incorporates permanent magnets surface mounted onto micro-fabricated released thin film structures. An applied external magnetic field induces noncontact loading to initiate delamination along the interface between the thin film and underlying substrate. Characterization of the critical peel force and peel angle is accomplished through in situ deflection measurements, from which the fracture toughness can be inferred. The test method was used to obtain interfacial fracture strength of 0.8-1.9 J/m 2 for 1.5-1.7 μm electroplated copper on natively oxidized silicon substrates. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Microelectronics, microelectromechanical systems, photovoltaic, and other devices consist of several thin film layers that are made of dissimilar materials. Such multilayered thin film systems are often prone to interfacial delamination under various loading conditions resulting in premature failure of the devices. Thus, interfacial delamination is an important reliability concern for a wide range of applications, and numerous experimental test techniques have been proposed to characterize the fracture resistance of an interface. Materials and the interfaces they share often have different mechanical properties at the micro/nano scale, and thus appropriate test methods are needed to characterize interfacial fracture between thin films. The traditional sandwich specimen tests (e.g. four-point bend, Brazil-nut) generally require non-standard fabrication and do not correlate to films as deposited in production. More recently developed techniques such as the scratch, indentation, and blister tests require either (1) direct contact with the test sample near the crack tip resulting in complicated stress fields, and/or (2) sophisticated fixturing and experimental setup. The stressed superlayer technique, proposed by Bagchi [1] and later extended by Modi [2] and Zheng [3], is a convenient fixtureless and noncontact method to characterize interfacial fracture. However, like the previously ⁎ Corresponding author. Tel.: + 1 404 894 3405; fax: + 1 404 894 9342. E-mail address: [email protected] (S.K. Sitaraman). 0040-6090/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2012.01.042
mentioned techniques, the load driving delamination is monotonically applied and cannot be easily modified to provide cyclical loading for a fatigue crack propagation study. An innovative delamination test is thus proposed that addresses the above highlighted shortcomings of alternative techniques. In this approach, a partially released thin film structure is gradually displaced out-of-plane until the peel force exerted on the adherend is sufficient to initiate delamination. The actuation force is supplied by the magnetic interaction between a permanent magnet assembled onto the film structure and a nearby electromagnet. With precise control of the current supplied to the electromagnet (and thus force exerted on the assembled permanent magnet), a noncontact, load-controlled peel test is performed. The proposed test has several key elements: (1) electromagnetic actuation provides the driving force needed for delamination propagation, and thus does not require any fixtures to apply loads; (2) an optical profiler monitors the delamination propagation in situ; (3) a wide range of loading amplitude and frequency is obtainable by adjusting the current supplied to the electromagnet generating the external magnetic field; (4) sample preparation involves only two photolithography steps and results in representative as-deposited films; (5) tens to hundreds of test samples can be batch fabricated onto a single wafer; (6) applied loading and delamination detection methods are suitable for both monotonic and fatigue crack characterization; (7) samples can be placed within an environment of excessive temperature and humidity while the load is modulated by an external magnetic field.
3988
G.T. Ostrowicki, S.K. Sitaraman / Thin Solid Films 520 (2012) 3987–3993
Fig. 1. (a) 3D illustration of MAPT specimen with sacrificial layer intact. (b) After removal of sacrificial layer, specimen is placed above electromagnet at fixed distance δ, where magnetic repulsive force lifts released features and imposes a peel force along the anchoring film strips.
One proposed test design is the Magnetically Actuated Peel Test (MAPT) shown in Fig. 1. Here, three strips of thin film meet at a central pad on which a permanent magnet is attached. The area underneath the central pad is released after removal of a thin intermediate sacrificial layer of diameter 2a (Fig. 2-i). The specimen is then situated above an electromagnet of opposing magnetic alignment (Fig. 1b). As sufficient electrical power is supplied, the repulsive magnetic force will cause the central pad to rise and impose a peel force along the three anchoring film strips (Fig. 2-ii). Once the peel force reaches a critical value, delamination along the film strip/substrate interface will progress (Fig. 2-iii). One significant advantage of the MAPT design is that the permanent magnet maintains its orientation during actuation and thus the applied
Fig. 2. Stages of MAPT: (i) at rest after removal of sacrificial layer; (ii) initial displacement due to application of magnetic force; (iii) onset of peel at critical load.
force is uniform and strictly controlled by the electromagnet current. As the area above the specimen is free of obstruction, an optical profiler can be used to capture the out-of-plane deflection and thus determine both the strain in the released film and the progression of delamination. 2. Peel test analogy The traditional peel test has been a widely studied technique to characterize interfacial fracture strength. As shown in Fig. 3a, a flexible film is peeled off an underlying substrate by a force F acting along an angle θ. However, when applied to very thin films with thickness less than several tens of microns, the peel test can result in rupture of the film due to severe deformation either at the crack tip or near the mechanical grips. This kind of cohesive failure can be mitigated by using a modification of the peel test called the pull-off test introduced by Gent and Kaang [4], where a vertical load P applied at the center of a debonded strip imposes a simultaneous peel force on both adhered ends of the film (Fig. 3b). Here the low angle of peel limits the bending stresses near the crack tip, and the load is typically applied by a horizontal pin underneath the debonded strip rather
Fig. 3. (a) Peel test. (b) Pull-off test.
G.T. Ostrowicki, S.K. Sitaraman / Thin Solid Films 520 (2012) 3987–3993
3989
than mechanical grips. The MAPT design is most analogous to the pull-off test, except that there are three debonded strips rather than two, and these strips are coupled to the bottom surface of a cylindrical permanent magnet supplying the vertical load. Assuming symmetry, the individual peel force F is related to the total magnetic load P by the relation P ¼ 3F sin θ:
ð1Þ
The thermodynamic approach to fracture characterization requires performing an energy balance during a crack growth process, and many theoretical models of interfacial fracture specific to the peel and pull-off tests can be found in the literature. A particularly simple and relevant model as described by Williams [5] assumes a flexible peel strip under uniform tension on a rigid substrate, and results in an expression for the interfacial fracture energy rate Γ0 as
Fig. 4. Fabrication process of MAPT specimen.
ε
F Γ 0 ¼ ð1− cos θ þ εÞ−h ∫ σ ⋅dε; b
ð2Þ
0
where F is the peel force, b is the strip width, θ is the peel angle, ε is the tensile strain, h is the strip thickness, and σ is the tensile stress. When an elastic constitutive model is used for the peeling film, the integral in (2) represents the stored elastic strain energy, in which case Γ0 is equivalent to the critical energy release rate Gc. However, when an elastic–plastic constitutive model is used for the peeling film, the integral includes the contributions of both the elastic strain energy and the plastic work. A nonlinear elastic–plastic constitutive model for 2 μm thick Cu film was used in this work [6]. Calculations of the critical peel force, peel stress, and fracture energy rate were done with respect to θavg, the average peel angle amongst the three strips per MAPT specimen. Since the focus of this work is the development and demonstration of an experimental test technique, an evaluation and comparison of alternative peel test models, including both analytical and finite element approaches, will be published elsewhere. 3. Experimental procedure 3.1. Sample preparation The initial proof-of-concept MAPT specimens detailed in this work feature electroplated copper film strips peeled off a natively oxidized silicon substrate. The microfabrication processes used to define the MAPT test film pattern are described in detail in [7] and are summarized below. A thin sacrificial polymer was patterned to define the initial pre-crack between the substrate and subsequently deposited metal film (Fig. 4a). Cu was then sputtered 500 nm thick and electroplated to total thickness of 1.5-1.7 μm. The encapsulated polymer was reflowed to achieve a dome shape, and after a second photolithography step the metal film was patterned via wet etch into the MAPT strip design (Fig. 4b). After removal of the sacrificial polymer (Fig. 4c), uncured Epibond® 7275 surface mount epoxy was then printed though a 3 mil stainless steel stencil onto the released central Cu pad. Finally, a custom alignment setup was used to precisely place a nickel-plated neodymium (NdFeB-42SH) magnetic cylinder (1/16”D × 1/16”H) onto the central pad and the assembly is completed with a 100 °C cure (Fig. 4d). The high temperature grade neodymium showed no loss of magnetism after the cure bake. The poles of the permanent magnet are at the ends of the cylinder such that they are magnetically aligned out-of-plane after assembly. Batch assembly of magnets is shown in Fig. 5. In the mask designs used in this research, b varies from 25 to 200 μm, and initial delaminated strip length α0 (i.e. distance from Cu central pad edge to outer sacrificial edge) varies from 250 to 1000 μm.
3.2. Magnetic force calibration Delamination in the MAPT design is driven by the force of magnetic repulsion, and this force is characterized using the experimental setup illustrated in Fig. 6. Here a Magnetic Sensor Systems E-66-38 tubular electromagnet is fixed at a specified distance δ from the NdFeB magnet resting on a high precision scale with ~ 1 μN sensitivity. Accounting for the weight of the permanent magnet, the downward force is directly captured by the scale. In this setup, the two critical parameters are the separation δ and the voltage potential V applied to the electromagnet. A voltage range of 0–13 V was applied at separation distances from 1.0 to 2.2 mm as shown in Fig. 7a, with maximum force achieved to be ~ 16 mN. Positive values of force indicate magnetic repulsion, whereas negative values indicate magnetic attraction. At low voltage, the ferromagnetic core of the electromagnet has a weak native polarity and is thus attracted to the permanent NdFeB magnet regardless of the direction of alignment. However, once ample current is supplied in the proper direction, the electromagnet will have a stable opposing polarity and repel the NdFeB magnet when at sufficiently large separation distances. It was interesting to observe that at very close distances the magnets would experience a net attractive force even at high voltage levels. This occurs because of the size effects of the magnetic field interaction between the small NdFeB magnet and the larger electromagnet. The separation gap is fixed during the peel test, and δ = 1.8 mm was chosen as an optimized distance due to high force potential, large operable voltage range, and low sensitivity to variations in the gap distance. At this distance, the magnetic force was found to be a linear function of V, transitioning from magnetic attraction to repulsion at ~3 V (see Fig. 7b). Each MAPT specimen magnet was calibrated individually in this manner with near identical results. For the peel test, the actual gap between the permanent magnet and electromagnet is δ + Δ, where Δ is the vertical displacement of the central Cu pad. However, the magnitude of Δ was less than 100 μm in the experiments, and so remained within the calibration's range of accuracy. 3.3. Peel test execution During the peel test, an electromagnet of opposing magnetic alignment is placed underneath the MAPT specimen such that actuation is a result of a repulsive magnetic force (Fig. 1b). Monotonic peel test experiments were conducted while a Veeco Wyko® NT-2000 optical profiler was used to measure the sample surface topography in situ. Surface images capturing the deflection of the film and
3990
G.T. Ostrowicki, S.K. Sitaraman / Thin Solid Films 520 (2012) 3987–3993
Fig. 5. Assembled MAPT specimens.
delamination propagation were taken at several discrete intervals of applied voltage. Shown in Fig. 8 are example images from one test specimen taken at significant points during the delamination
test. At 3.0 V the sample experiences essentially no force and is in the initial at rest state. At 6.0 V the strips are raised and pulled taught but no significant delamination is observed. At 7.0 V delamination is
G.T. Ostrowicki, S.K. Sitaraman / Thin Solid Films 520 (2012) 3987–3993
3991
the interfacial crack length and peel angle were obtained as a function of the applied magnetic force. Each MAPT specimen provides displacement data for three peel strips simultaneously. Assuming evenly distributed force amongst the three strips, the critical peel force and associated peel angle are determined once delamination of at least one peel strip reaches steady-state. Since no images could be taken during the dynamic peeling process, it is assumed that any incrementally higher force beyond the last stable deflection measurement would induce steady state delamination. Consequently, the highest sustained peel force is reported here as a conservative estimate for the critical peel force. Peel test results for five MAPT specimens are presented in this work. Due to two different fabrication techniques to remove the sacrificial material, two specimens were heated within an inert environment at 250 °C for 2 hours before magnet assembly while the remaining three specimens were not heat treated.
Fig. 6. Experimental setup for magnetic force calibration.
observed and slowly propagates until 9.1 V, after which steady state delamination results in total failure of the peel test. The slope of each peel strip along with the location of the crack front was determined from each of these profiler images. Thus, along with the previous force calibration results, measurements of
4. Results and discussion Processing of the optical profiler images resulted in the typical peel angle vs. magnetic force relationship shown in Fig. 9. Since the measured peel angles of the three strips may be slightly different from one another at a given voltage, the average peel angle was used to estimate the peel force. In all but one case, steady-state peeling was reached immediately following delamination initiation. In the one exception, a heat-treated MAPT specimen was observed to exhibit unstable crack growth over a range of applied force values until steady-state peeling was reached at the critical load (Fig. 10). The results of the MAPT experiments are summarized in Table 1. In order to appropriately compare peel strips of varying dimensions, the magnetic pull-off force and tangential peel force are expressed per strip cross section as P/bh and σ = F/bh, respectively. The critical fracture energy rates were calculated by inputting the average measured peel angle and average peel force (and stress) at fracture into Eq. (2). Using the nonlinear constitutive model for Cu, the interfacial fracture energy rate of the heated specimens (samples 1–2) was calculated to be ~0.8 J/m 2, whereas the non-heat-treated specimens (samples 3–5) exhibited slightly higher values of 0.9-1.9 J/m 2. If instead the Cu film is assumed to be linear elastic with a modulus of 110 GPa, the reported interfacial fracture energy rate would increase by ~ 10-20%. These values are comparable to the reported critical energy release rates of ~ 1-10 J/m 2 for ~ 1 μm Cu on SiO2 using other techniques [8,9]. Despite the limited sample size, there appears to be a trend of interface weakening due to heat treatment. While it is suspected that thermal expansion mismatch and Cu film grain coarsening is causing damage at the interface, the conclusive reasoning for this weakened interface, including the testing of a larger sample set, is the subject of ongoing work. 5. Concluding remarks
Fig. 7. (a) Typical force vs. magnet separation gap δ for varying applied voltage to the electromagnet. (b) Typical force vs. electromagnet voltage at fixed δ = 1.8 mm.
The magnetically actuated peel test concept was devised and demonstrated to reliably measure interfacial fracture strength between a thin film and underlying substrate. A sample specimen design featuring pull-off of three film strips allowed for precise and stable loading while an optical profiler captured film deflection and crack propagation. Unlike most conventional techniques, the MAPT design does not require fixtures or physical contact with the test specimen. Interfacial fracture toughness values were obtained for a thin electroplated Cu film on a natively oxidized Si wafer, with preliminary results indicating a weakened interface for heat-treated samples. Interfacial characterization was demonstrated for single-material peeling films with ~ 1.5 μm thickness. It is expected that the MAPT technique can accommodate single-material peeling films as thin as
3992
G.T. Ostrowicki, S.K. Sitaraman / Thin Solid Films 520 (2012) 3987–3993
Fig. 8. Optical profiler images at select intermittent applied voltage during delamination test. Black color indicates areas that are missing out-of-plane height data. Delamination is observed at 7.0 V and propagates until 9.1 V, after which steady-state peeling commenced.
a few hundred nm with no change in the as described fabrication process or test setup. However, in order to test even thinner films (b100 nm), it may be required to use a bi-material peeling film, where the interface of interest is between the substrate and the bottom peel strip material, and the top peel strip material provides structural support. Using such a bi-material design would allow for interfacial characterization of films that are a few nm thick, but would also likely require more sophisticated mechanics analysis to account for the effect of the supporting film on the extracted interfacial fracture parameters.
Fig. 9. Typical peel angle vs. magnetic force.
Although only monotonic fracture is demonstrated in this work, the MAPT test design is ideally suited to conduct interfacial crack propagation experiments under fatigue loading, and is the subject of ongoing research. In such a fatigue test, a cyclic load can be applied to the test specimen by supplying an AC current with desired amplitude and frequency to the driving electromagnet, while optical profiler measurements can be taken intermittently after a set number of cycles in order to capture the delamination propagation. Thus, the crack growth rate can be observed as a
Fig. 10. Non-typical unstable crack growth exhibited for one heat treated MAPT specimen (No. 2).
G.T. Ostrowicki, S.K. Sitaraman / Thin Solid Films 520 (2012) 3987–3993 Table 1 MAPT results for electroplated Cu on natively oxidized Si substrate. Cu width Pull-off Sample Cu Average Average peel FER Γ0 force (P/bh)c peel angle stress σavg number thickness b (μm) (J/m2)* h (μm) θavg (deg) (MPa) (μN/μm2) 1 2 3 4 5
1.7 1.7 1.5 1.5 1.5
100 200 100 100 100
33 36 47 53 74
4.1 3.8 4.0 5.5 5.5
154 184 229 187 255
0.8 0.8 0.9 1.4 1.9
Note: * Fracture Energy Rate (FER) calculated from Eq. (2) using nonlinear Cu constitutive model.
function of the loading amplitude, from which Paris Law parameters can be determined. Acknowledgments This work is supported by the National Science Foundation under grant No. CMMI-0800037 and also by Grant No. ECCS-0901679. The
3993
authors gratefully acknowledge Nathan Fritz and Dr. Paul Kohl (School of Chemical and Biomolecular Engineering, Georgia Institute of Technology) for assistance and helpful discussion regarding sample preparation.
References [1] [2] [3] [4] [5] [6] [7]
A. Bagchi, G.E. Lucas, Z. Suo, A.G. Evans, J. Mater. Res. 9 (1994) 1734. M.B. Modi, S.K. Sitaraman, Eng. Fract. Mech. 71 (2004) 1219. J.T. Zheng, S.K. Sitaraman, Thin Solid Films 515 (2007) 4709. A.N. Gent, S. Kaang, J. Appl. Polym. Sci. 32 (1986) 4689. J.G. Williams, Int. J. Fract. 87 (1997) 265. C.A.O. Henning, F.W. Boswell, J.M. Corbett, Acta Metall. 23 (1975) 177. G.T. Ostrowicki, N.T. Fritz, R.I. Okereke, P.A. Kohl, S.K. Sitaraman, IEEE Trans. Device Mater. Reliab. (2012), doi:10.1109/TDMR.2011.2175927. [8] A.A. Volinsky, N.I. Tymiak, M.D. Kriese, W.W. Gerberich, J.W. Hutchinson, MRS Symp. Proc. 539 (1999) 277. [9] A. Bagchi, A.G. Evans, Thin Solid Films 286 (1996) 203.