Nickel–silicon interfacial adhesion strength measured by laser spallation Open Access

The increasing use of renewable energy sources has led to a pressing need for high-energy-density battery technologies. Lithium-ion (Li-ion) batteries have become the most widely used power source for portable electronic devices, electric vehicles (EVs), and stationary energy storage systems. However, it is widely acknowledged that current Li-ion technology will not meet future energy storage needs in terms of energy density, performance, durability, and longevity.^1,2

The energy density of Li-ion batteries is primarily determined by the anode materials utilized. Anodes made of amorphous silicon (a-Si) increase the specific capacity by an order of magnitude compared to conventional carbon-based anodes, achieving a specific capacity of up to 4200 mA h g⁻¹.³ A common feature in many a-Si anodes^4–7 is the use of a-Si as the active material and a metal such as nickel (Ni) as the current collector. This configuration makes the Ni/a–Si interface a crucial component of the overall battery design. Failure at this interface, induced by the significant volumetric expansion of a-Si during charge–discharge cycling,^3,8 can result in the loss of contact between a-Si and Ni and, thus, the degradation of battery capacity and performance.⁹ 

Despite the crucial role of the Ni/a–Si interface, little work has been done in the battery community to study Ni/a–Si adhesion strength. Although a number of conventional adhesion test methods¹⁰ have been used to characterize the interfacial strength between the active materials and current collector, including peeling,^11,12 pulling,^13,14 bending,^15,16 and cutting,¹⁷ and among others,^18–21 these methods have significant limitations. The data interpretation is complicated by poorly defined stress states during the test, plastic deformation, unwanted damage of the film under investigation, or complicated sample preparation, which leads to inconsistency or inaccuracy in the measured adhesion strength.²² In contrast, laser spallation (LS)^23–30 is a non-contact testing method that alleviates the intrinsic disadvantages of the aforementioned methods, by providing repeatable control over the stress state at the interface, access to high strain rates (∼10⁷ s⁻¹) with minimal inelastic deformation processes,²⁹ for a range of film thicknesses.²⁶ Due to these advantages, LS has been used on a wide range of film thicknesses, from nanometers to micrometers,²⁵ for bimaterial systems such as ceramic/ceramic,³¹ metal/metal,³² metal/ceramic,²⁵ and ceramic/polymer.³³ 

Here, we adapt LS induced delamination to make the first measurement of thin-film Ni/a–Si interfacial adhesion strength. The a-Si film is deposited via chemical vapor deposition (CVD) on an e-beam deposited Ni film to form the interface. By controlling the laser pulse fluence in combination with finite element analysis (FEA) simulations, the tensile adhesion strength of the Ni/a–Si interface is obtained. Additional experimental measurements further reveal design guidelines for the multilayer samples. The tensile LS test performed here provides a conservative baseline in system level design and provides an efficient and accurate method to simulate structured Ni/a–Si anodes.

The optical setup used for LS experiments is shown in Fig. 1.² The laser used to induce stress waves was an Nd: YAG (New Wave Tempest 10 Hz) Q-switched pulsed laser with a wavelength of 1064 nm, a pulse width of 5 ns, and a nominal beam diameter of 5 mm. The laser pulse was focused through a 150 mm focal length lens (CVI Laser, BICX-50.8–153.4-C-1064) to a 2 mm-diameter spot on the sample. An attenuator was coupled to the Nd: YAG laser to tune the energy output in the range of 0–190 mJ, corresponding to a fluence of 0–55 mJ/mm² as the attenuation factor decreases from 1000 to 0 (Fig. S1 in the supplementary material). The sample was held in a custom holder and was translatable on all three principal axes by micrometer screws. Prior to testing, the laser output power was calibrated using a Gentec power meter (QE25LP-H-MB). The Michaelson interferometer consisted of a 532 nm mode locked laser (Laser Quantum, Torus 532) that was first collimated and then focused through a 200 mm lens to bring the beam to a sharp point on the sample. After the focusing lens, the beam was passed through a non-polarizing beam splitting cube and split into two perpendicular beams. The first beam was directed toward the stationary reference mirror and the second beam toward the front face of the sample. The interferometric beam was carefully aligned to be colinear with the pulsed laser to ensure strong reading and accurate measurements. The high-speed photodetector used was an Electro-Optics Technology ET-2030. The voltage readings from the photodetector were collected using a LeCroy WaveRunner 640Zi oscilloscope with a sampling rate of 40 GHz. A built-in smoothing function was used to automatically smooth data in the oscilloscope. The smoothing function reduced noise but did not affect fringe maxima or periodicity and, thus, did not impede the accuracy of the test results.² 

To enable the interfacial adhesion strength measurement, we designed and fabricated multi-layered samples for the LS test as shown in Fig. 2. The samples consist of a 1.6 mm-thick and 25.4 mm-diameter fused silica substrate (GM Associates, 7500–01), an energy-absorbing layer and a confining layer on the backside of the fused silica, a Ni layer, and an a-Si layer on the frontside [Fig. 2(a)]. Fused silica was used as the substrate due to its non-linear stress–strain relationship that shapes a Gaussian pulse into a wave with a linear ramp and large substrate compressive stress followed by a very sharp fall time.^34,35 At 1.6 mm thickness, the fused silica substrate (GM Associates, 7500-01) generates a compressive stress with limited geometric attenuation while maintaining a planar one-dimensional (1D) wave propagation.^25,36 A 400 nm-thick Al layer was deposited as the energy-absorbing layer via e-beam evaporation (Temescal FC-2000). The Al layer is much thicker than the critical penetration depth (∼10 nm) of the laser, allowing for the laser energy to be completely absorbed within the Al layer.²⁵ Waterglass (sodium silicate, Fisher Scientific SS338-1) solution (1:1 by weight) was spin-coated (3000 rpm for 180 s) on the Al absorbing layer to form a 7 μm-thick solidified layer as a confining layer (thickness measured by stylus profilometry). On the front side, a 5 nm-thick chromium (Cr) adhesion-enhancing intermediate layer followed by a 50 nm-thick Ni layer was deposited via e-beam evaporation (Temescal FC-2000). The deposition of Ni on Cr was performed after the sample was naturally cooled down in the evaporator. The thickness of the Ni layer (50 nm) is close to the Ni thicknesses in a scaffold of a well-designed Si anode⁴ and is beyond the cut-off distance of molecular interactions (<10 nm) to represent an infinite body.³⁷ The Ni layer is thin enough to reduce the substrate/Ni interfacial stress and allows for varying thicknesses of a-Si. Last, the a-Si layer with a thickness of 125–146 nm was deposited via static chemical vapor deposition (CVD) using a customized setup (Fig. S2 in the supplementary material).³⁸ Air plasma cleaning (PDC001-HP, Harrick Plasma, medium power, 10–20 min) of the substrate was performed immediately before thin-film deposition to remove adventitious contamination (e.g., adsorbed volatile organic compounds or VOCs).^39,40 The same plasma cleaning procedures were conducted whenever the surfaces were exposed to the atmosphere for 10 min or more, to minimize the effects of VOCs. During e-beam coating, a relatively low deposition rate (<0.15 nm/s) was adopted to limit thermal and growth stresses during the deposition of the thin films. In addition to the test samples, calibration samples were also fabricated following the same procedure, except for the absence of the a-Si layer on the front side. The exposed Ni layer acts as the reflective layer to enable reflection and interference of the diagnosis laser.

Laser spallation tests were performed by directing laser pulses of varying fluences (or extent of attenuation) to the backside of the as-fabricated samples. Upon laser ablation of the absorbing layer, the expansion was confined by the water glass (WG) to generate a compressive stress wave propagating through the fused silica substrate toward the Ni/a–Si interface [Fig. 2(b)]. The laser fluence was increased by tuning the attenuation factor (AT, ranging from 1000 to 0) to initiate a-Si film spallation. A large attenuation factor represents a lower laser energy fluence. After each laser shot, the sample was translated by 3 mm to align a new location for laser pulses. The 3 mm spacing was large enough to ensure planar 1D wave propagation³⁶ for mechanics analysis. As shown in Fig. 3, spallation was first observed at a laser fluence of ψ = 15.87 mJ/mm² (or attenuation factor AT = 850), as confirmed by the optical and digital microscope. A lower laser fluence (e.g., ψ = 11.96 mJ/mm², or AT = 900) was not able to induce a-Si film failure, while larger scale delamination was observed at higher fluences (e.g., ψ = 19.52 mJ/mm² or AT = 800). Under the digital microscope (Keyence VHX-5000), the a-Si surface appears green while the Ni surface looks pink, confirming the delamination of the upmost a-Si from the underlying Ni surface. To ensure the repeatability of the measurement, a minimum of three spallation tests were conducted at each fluence to obtain the critical laser fluence above which the a-Si film delaminated.

As shown in Fig. 4, a lower attenuation factor (or a higher laser fluence) results in a faster-oscillating fringe [Fig. 4(a)], indicating a larger displacement of the reflective film [Fig. 4(b)] and a larger maximum compressive stress within the substrate [Fig. 4(c)]. Consistent with previously reported stress wave profiles within fused silica when using a Gaussian laser pulse,^25,30,34 the substrate stress shows a linear increase in the magnitude followed by a fast decompression, with the maximum substrate stress reaching −1.1 GPa for ψ = 19.52 mJ/mm² (AT = 800) in Fig. 4(c). At all laser fluences tested, no delamination of the reflective Ni layer was observed. Note that the negative sign denotes compressive stress.

The interfacial adhesion strength is determined using finite element analysis (FEA) and the stress profiles measured for the calibration samples. Specifically, a one-dimensional FEA was developed to study the wave propagation in the multilayer thin-film specimens and analyze the stresses developed at each interface. The non-linear fused silica substrate was not explicitly included in the analysis. Instead, we imposed the substrate stress history measured directly from calibration experiments at the same fluence, i.e., the substrate stress profile [Eq. (3)] obtained from the calibration samples at the critical laser fluence (ψ = 15.87 mJ/mm² or AT = 850) in Fig. 4(c), to calculate the stress at the Ni/a–Si interface, $σ Ni / Si(t)$⁠. The top and bottom surfaces of the multilayer film stack were assumed in plane strain with traction free boundary conditions. The specimen geometry was meshed using the elements of size $Δh/30$⁠, where $Δh$ is the thinnest layer dimension. This element size was chosen based on a convergence study. An explicit direct integration scheme with a forward difference time step was applied for transient analysis. The simulations were performed using a time step $Δt/5$⁠, where $Δt$ is the minimum of wave travel time in the elements of considered layers. Details on FEA can be found in previous studies.⁴² The peak value of the simulated $σ Ni / Si(t)$ represents the interfacial adhesions strength $σ Ni / Si ∗$⁠. To check the accuracy of the FEA, we compared the interferometric measurements of the film surface displacements to the FEA-calculated displacements. The simulated peak displacements averaged less than 0.5% deviation for all shots, validating the use of continuum 1D wave mechanics in our simulations (Fig. S3 in the supplementary material).

The thickness measurement of the a-Si film is critical for accurate calculation of $σ Ni / Si(t)$⁠. Figure 5(a) shows laser scanning microscopy (Keyence VK-X1000) and a cross-sectional profile of the spallation spot obtained at ψ = 46.18 mJ/mm² (AT = 400, inset in Fig. 3). Surface-averaging measurements indicate an a-Si thickness of $Δ h Si=125 nm$ (Fig. S4 in the supplementary material). Although the laser scanning microscope has a relatively low height resolution (20 nm), it provides a large field of view for baseline benchmarking. Atomic force microscopy (AFM, Asylum Research Cypher S) with a higher vertical resolution (∼ 0.1 nm) confirms our measurement consistency [ $Δ h Si=142 nm$⁠, Fig. 5(b)].

Using the known a-Si thickness, the interfacial stress at the critical laser fluence (ψ = 15.87 mJ/mm² or AT = 850) was calculated to be $σ Ni / Si ∗=68 MPa$ [Fig. 5(c)]. To account for the uncertainty of interfacial adhesion strength, we incorporated the uncertainty of thin-film thickness and critical substrate stress into the FEA simulation by considering the lower and upper boundaries of a-Si thicknesses and substrate stresses. Given that the thicknesses of Cr and Ni layers do not directly contribute to the Ni/a–Si interfacial stress [Eq. (4)], we only considered the lower and upper boundaries of the a-Si layer $( Δ h Si − < Δ h Si < Δ h Si +)$⁠. To facilitate the analysis, $Δ h Si −$ and $Δ h Si +$ are obtained from thickness measurements by laser scanning microscopy $( Δ h Si − = 125 nm)$ and AFM $( Δ h Si + = 146 mm)$ . The boundaries of the critical interfacial stress $( σ Ni / Si − < σ Ni / Si ∗ < σ Ni / Si +)$ are calculated from the FEA using the substrate stress profiles at critical laser fluences near spallation initiation (ψ = 11.96 and 15.87 mJ/mm², corresponding to AT = 900 and 850, respectively). Figure 5(d) shows the Ni/a–Si interfacial stress as a function of a-Si thickness for varying laser fluences. It is worth noting that the FEA-calculated interfacial stress is proportional to the a-Si thickness for a given substrate stress, consistent with the theoretical analysis [Eq. (4)]. The lower boundary of the Ni/a–Si interfacial adhesion strength $( σ Ni / Si −)$ is obtained at a lower substrate stress (ψ = 11.96 mJ/mm² or AT = 900) and a thinner a-Si thickness $Δ h Si −=125 nm$ , while the upper boundary exists at a higher substrate stress (ψ = 15.87 mJ/mm² or AT = 850) and a larger a-Si thickness $( Δ h Si + = 146 mm)$⁠, i.e., $σ Ni / Si −= σ Ni / Si ∗$ (AT = 900, $Δ h Si −$⁠) and $σ Ni / Si += σ Ni / Si ∗$ (AT = 850, $Δ h Si +$⁠). From Fig. 5(d), we obtain $σ Ni / Si ∗∈[ 45.7 MPa , 72.1 MPa]$⁠.

The key to successful measurement of interfacial adhesion strength is the design of the multi-layer test sample. The properties and constitutive relationship of the substrate play a critical role in the development of a stress wave and must be well characterized. In contrast to fused silica substrates, metal plates (Al or Cu) show slower fall and rise in the compressive stress profile that leads to lower peak compressive stress and weaker interfacial stress (Fig. S5 in the supplementary material), making interfacial failure difficult. The use of Si wafers sees damage due to cleavage under the action of the tensile wave at high laser energy fluences.²⁵ 

In addition to the rational selection of the substrate, the orientation of the thin films needs to be designed to ensure spallation at the interfaces of interest.² In the current surface configuration, two heterogeneous interfaces exist: the substrate/Ni and Ni/a–Si interfaces. The delamination of the Ni/a–Si interface needs to be initiated prior to the failure of the substrate/Ni interface, or $σ sub / Ni ∗> σ Ni / Si ∗$⁠, where $σ sub / Ni ∗$ and $σ Ni / Si ∗$ are the maximum interfacial stresses at the substrate/Ni and Ni/a–Si interfaces, respectively. The intermediate Cr layer enhances the substrate/Ni adhesion strength from <100 MPa to $σ sub / Ni ∗≈233 MPa$ (Fig. S6 in the supplementary material), thus allowing for a similar interfacial adhesion strength of Ni/a–Si to be measured.

As plotted in Fig. 6, $σ sub / Ni$ and $σ Ni / Si$ are the linear functions of $Δ h Si$⁠, with the same slope [ $− ρ Sia$ in Eq. (5)] determined by the stress profile. For $Δ h Ni=50 nm$ and ψ = 15.87 mJ/mm² (AT = 850), we obtained $Δ h Si$ ranging from $Δ h Si −=90 nm$ to $Δ h Si +=340 nm$⁠, corresponding to known values of $σ ~ Ni / Si ∗=50 MPa$ and $σ sub / Ni ∗=233 MPa$ (Fig. 6). Similar calculation procedures can be used to optimize the design a multi-layer sample for laser spallation tests. Equation (5) reveals that a thick Ni layer $( Δ h Ni)$ reduces the upper boundary of a-Si thickness $( Δ h Si +)$⁠, and in the limiting cases where $Δ h Si +$ drops below $Δ h Si −$⁠, substrate/Ni interface failure becomes a critical issue preventing the precise characterization of Ni/a–Si interfacial adhesion strength.

The measured Ni/a–Si interfacial adhesion strength $( σ Ni / Si ∗ ∈ [ 45.7 MPa , 72.1 MPa ])$ is similar to Al/Si (∼60 MPa),²⁵ but smaller than that of polyimide/Si or Si₃N₄ (>140 MPa),⁴² epoxy/fused silica (>120 MPa),⁴³ and diamond/polycrystalline alumina (>140 MPa)⁴⁴ interfaces, as well as the Ni/fused silica interfaces with an intermediate Cr layer reported here $( σ sub / Ni ∗ > 200 MPa)$⁠. The relatively weak adhesion strength may result from the fabrication processes of the Ni and a-Si films. Future work is needed to provide a further understanding of the Ni/a–Si system by investigating the effects of intermediate layers, surface roughness, annealing, residual stresses, impurities, and native oxides, all of which have the potential to change the interface strength.^45,46 Furthermore, with Ni underneath the Si layer, the current surface configuration enables electrodeposition of Si to Ni in addition to CVD presented here, allowing one to explore the role of Si deposition methodology in interfacial strength. The measured Ni/a–Si interfacial strength, which has not been reported in the literature, could be employed for failure modeling⁴⁷ and to investigate the changes in electrode configuration¹⁹ in future studies.

Laser spallation (LS) coupled with finite element analysis (FEA) was used to probe the interfacial adhesion strength of the nickel and amorphous silicon (Ni/a–Si) interface. Multilayer samples were designed to enable the LS test by enabling high tensile interfacial stress and promoting delamination of the interface of interest. For a-Si fabricated via CVD, the Ni/a–Si adhesion strength was measured to be ≈46–72 MPa. Furthermore, design guidelines for multi-layer samples for LS tests were developed based on a thin-film mechanics analysis. Our results provide insight into the strength of the Ni/a–Si interface and serve as an important metric for the material design of high-energy-density anodes. The heterogeneous adhesion strength characterization procedures presented here also provide valuable methods for the design and understanding of thin films and coatings for energy,⁴ thermal,^48–50 optical,⁵¹ chemical,⁵² and microelectromechanical applications.^16,53

See the supplementary material for information about laser energy fluence as a function of attenuation factor (Fig. S1), CVD setup and procedures for a-Si deposition (Fig. S2), analysis of interfacial stress (Fig. S3), surface-averaging measurement of the a-Si thickness (Fig. S4), compressive stress profile of Al, Cu, and Si substrates (Fig. S5), and measurement of the substrate/Ni adhesion strength (Fig. S6).

N.S., P.B., X.Y. and J.M.D. acknowledge the funding support by the U.S. Army CERL under Grant No. W9132T-19-2-0008. X.Y. and N.M. acknowledge the funding support by the Office of Naval Research (ONR) under Grant No. N00014-16-1-2625. N.M. also acknowledges the funding support from the International Institute for Carbon Neutral Energy Research (No. WPI-I2CNER), sponsored by the Japanese Ministry of Education, Culture, Sports, Science and Technology. X.Y. also acknowledges support from the Innovative research group project of National Natural Science Foundation of China (No. 52021004) during writing of the manuscript. This work was carried out, in part, at the Materials Research Laboratory Central Research Facilities, University of Illinois.

The authors have no conflicts to disclose.

X.Y. and J.M.D. contributed equally to this work.

Xiao Yan: Data curation (equal); Formal analysis (equal); Investigation (equal); Writing – original draft (lead); Writing – review & editing (equal). Jacob M. Diamond: Data curation (equal); Formal analysis (equal); Investigation (equal); Writing – original draft (equal); Writing – review & editing (equal). Nathan J. Fritz: Data curation (equal); Formal analysis (equal); Writing – original draft (supporting). Satoshi Matsuo: Data curation (supporting); Writing – original draft (supporting). Kazi F. Rabbi: Data curation (supporting); Writing – original draft (supporting). Ishrat Zarin: Data curation (supporting); Writing – original draft (supporting). Nenad Miljkovic: Project administration (equal); Supervision (equal); Writing – review & editing (equal). Paul V. Braun: Funding acquisition (equal); Project administration (equal); Supervision (equal); Writing – review & editing (equal). Nancy R. Sottos: Conceptualization (lead); Funding acquisition (lead); Project administration (lead); Writing – original draft (equal); Writing – review & editing (equal).

The data that support the findings of this study are available within the article and its supplementary material.