Microscopic and Non-Site-Specific Measurement Method for Bond Strength of Direct Bonded Wafers and Dies

Microscopic and Non-Site-Specific Measurement Method for Bond Strength of Direct Bonded Wafers and Dies | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Microscopic and Non-Site-Specific Measurement Method for Bond Strength of Direct Bonded Wafers and Dies Fumihiro Inoue, Junya Fuse, Yuki Yoshihara, Yusuke Kondo, Daiki Kobayashi, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6298479/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The demand for 3D integration and advanced packaging technologies is increasing. Direct bonding technologies are gaining popularity as methods to achieve further increases of the input/output densities. Although accurate evaluations of bond strength are essential for reliable wafer-level and die-level integration, the local bond strength is not measurable because of the limitations of conventional measurement methods. Here, we used a nanoindentation method for both wafer-to-wafer and die-to-wafer samples. These measurements provide insights into local bond strength and a wide range of uniformities. The obtained statistic value is equivalent to that obtained by the double-cantilever-beam method in an anhydrous atmosphere, implying that the value obtained by the nanoindentation method is not affected by the measurement environment or the true bond strength. The developed method also enabled the measurement of bond strength in die-to-wafer bonded samples, revealing bond-strength variations within individual dies and identifying potential improvements in the die-level bonding process. Physical sciences/Engineering/Mechanical engineering Physical sciences/Engineering/Electrical and electronic engineering Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Introduction Three-dimensional (3D) integration technology enhances the performance, integrity, and energy efficiency compared with those of monolithic two-dimensional (2D) devices in diverse device applications. In addition, 2.5D/3D chiplet integration provides design flexibility and cost efficiency while enabling high yield through known-good-die (KGD) selection 1 – 14 . Plasma-activated direct wafer-to-wafer (W2W) bonding is being implemented in wafer-level 3D integration because it is compatible with front-end-of-line processes (e.g., backside power delivery network, complementary FET (CFET), and 3D DRAM) 15 – 20 . In addition, Cu hybrid bonding is used for image sensors and 3D NAND memories for higher-performance electrical connections, where it provides a greater density of input/output (I/O) connections while minimizing inductance, capacitance, and resistance, enabling faster and more energy-efficient signal transmission 21 , 22 . Die-level hybrid bonding is also crucial for chiplet architectures and high-bandwidth memory, where solder-based microbumps are currently the conventional integration method. Die-level integration is mandatory as the conceptional background for these integration schemes (e.g., KGD selection for high yield) 23 – 26 . However, direct bonding technology, which uses inorganic dielectric films as bonding interfaces instead of organic adhesives, faces challenges with respect to reliably evaluating the bond strength. Previous studies have explored measurement methods based on the double-cantilever-beam (DCB) test commonly used to evaluate adherence energy (i.e., bond strength) at direct bonding interfaces. Such studies have revealed that the method provides robust and true bond strength when it is conducted in an anhydrous atmosphere like that in a glove box filled with inert gas 27 , 28 . However, the method is limited in its ability to provide fundamental understanding of microscopic bonding dynamics. The DCB method can be used to calculate the adherence energy on the basis of the length of delamination caused by inserting a blade at the interface. The delamination caused by blades is relatively wide (e.g., a few square centimeters), which makes the measurement of microscopic bond strength difficult. In addition, variations in the within-wafer (WiW) uniformity of bond strength, which are caused by process variations, cannot be measured by the DCB method, because the method requires the insertion of a blade at the wafer edge/bevel. The data obtained by DCB are site-specific at the wafer edge and macroscopic. Furthermore, the method is not applicable for die-to-wafer (D2W) bonded samples, because the blade cannot be inserted from the die edge. Pull tests or share tests can measure the mechanical strength; however, the results of the measurements cannot be directly compared to the adherence energy obtained by DCB. These methods also often cause breakage of the bulk Si and therefore cannot provide interfacial information. Consequently, chiplet integration with direct/hybrid bonding necessitates a re-evaluation of adherence energy using an alternative approach. The present study aims to establish alternative bond-strength measurement methods for both W2W and D2W integration in direct bonding interfaces using a microscopic and non-site-specific method. We investigate the WiW uniformity of bond strength in wafer-level integration, which is crucial for assessing the robustness of 3D integration. Nanoindentation (NI) tests are widely used to measure mechanical properties such as the hardness and Young's modulus of thin films. In addition, recent studies have examined NI testing for evaluating thin-film adherence 29 – 32 . However, comprehensive studies on adherence energy measurements for direct bonding interfaces remain sparse and novel. In the present study, we carry out a detailed mechanistic examination of the NI method by comparing robust DCB measurement results as shown in Fig. 1 . In addition, we conduct NI tests to confirm their effectiveness for evaluating WiW uniformity and within-die (WiD) uniformity of bond strength to promote robust 3D/chiplet integration. Results and discussion Verification and optimization of measurement conditions The load–displacement curves acquired during the NI tests are shown in Fig. 2 . The indentation was conducted at depths ranging from 70 nm to 150 nm. The structure of the samples consisted of two thermally oxidized Si layers, each 100 nm thick (bonding interface at 100 nm depth) on the bulk Si substrate as the bottom Si. Controlling the indentation depth was necessary to avoid delamination between the substrate and the lower bonding film (thermal Si oxide) layer. Therefore, the indentation depth was carefully controlled, and appropriate indentation conditions were determined from the acquired Scanning Probe Microscope (SPM ) images. Below the film thickness of the bonding dielectric as the indentation depth, no delamination or shape change was observed around the indentation mark. Near the top film thickness (~ 100 nm) as the indentation depth, delamination was observed only at one edge of the triangular indentation. At the indentation depth of 130 nm (1.3× deeper than the original top bonding dielectric), a shape change was observed around the indentation but the delamination was not uniform. At the indentation depth of 150 nm, delamination occurred around all three edges of the indentation, with a roughly uniform pattern. In load–displacement curves corresponding to conventional indentation tests with delamination, a region where the load becomes constant upon delamination is expected to appear. However, no such phenomenon was observed in the samples tested in the present study. This lack of a constant-load region might be due to the change of the load on the thin film used in this experiment being below the detection limit of the indentation system. Thus, to minimize variations, we conducted indentation tests to a depth of 150 nm, where peeling occurred uniformly. We conducted an interface analysis using cross-sectional transmission electron microscopy (TEM) to observe the delamination shape in greater detail. Figure 3 a and b show top-view scanning electron microscopy (SEM) images of indentations made during nanoindentation testing of samples with strong and weak bond strengths, as evaluated using the DCB method. Neither sample surface was chipped or broken. In addition, we observed that the change in the shape of the periphery was small in the sample with high bond strength. The micrographs of the cross sections of these indentations are shown in Fig. 3 c and d. The horizontal length of the delamination at the SiO 2 interface was 260 nm in the case of low bond strength (Fig. 3 c), and the delamination length at the SiO 2 interface was 176 nm in case of high bond strength (Fig. 3 d). No delamination was observed between the SiO 2 layer and the Si substrate in either sample. Irrespective of the bond strength, the delamination was only observed at the SiO 2 –SiO 2 interface, implying that the bond strength can be quantitatively calculated. In addition, because the bonded interface after the NI test was not exposed, we reasonably assumed that it did not react with humidity in the air during the indentation. Thus, the progress of interfacial delamination due to water stress corrosion (WSC), observed in the DCB test, can be reasonably assumed to not affect the NI test 27 , 33 . Consequently, the results of the DCB test in an anhydrous atmosphere should with compared those of the NI test under ambient atmosphere. Comparison between DCB and NI To confirm the compatibility between the DCB test and the NI test, we evaluated the bond strength at the edge using the DCB method. The bond-strength evaluation using NI was calculated from the delamination area after the indentation. This method is described in detail elsewhere 34 – 36 . After the indentation, the sample's indentation state was observed using SPM. The corresponding 3D plots (Fig. 4 a) show a bulge around the indentation mark. The image at exactly the same indentation mark expressed in two dimensions is shown in Fig. 4 b. The white part represents the bulged area, and we calculated the bond strength using this area as the delamination region. We analyzed the acquired SPM images to calculate the delamination area. The threshold value used in this analysis matched the value shown in Fig. 4 b; the analysis results are shown in Fig. 4 c. Using the peel volume ( V = area × film thickness) determined from these analyses, we express the bonding energy ( G c ) by the following formula 34 : $$\:\begin{array}{c}{G}_{\text{c}}=2\gamma\:=\frac{{E}_{\text{f}}h{V}_{\text{o}}^{2}}{4(1-\nu\:){V}_{\text{c}}^{2}}\:[J/m²]\#\left(1\right)\end{array}$$ In the NI test, because a thin film was approached, the Young's modulus E and Poisson's ratio ν were used as the values for a thermal oxide film. Reference data were acquired using the commonly used DCB test to obtain reliable data. Measurement variations were minimized using a semi-automatic blade insertion machine and a glove box filled with inert gas, which enabled measurements under an anhydrous atmosphere. Details of the measurement environment and conditions are available elsewhere. 27 , 28 , 37 , 38 We here present only the formula for the energy release rate, G , calculated by the DCB test. In the present study, G c is treated as bond strength. The bonding energy G c and γ [J/m²] are expressed as follows: $$\:{G}_{\text{c}}=2\gamma\:=\frac{3{t}_{\text{b}}^{2}E{t}_{\text{w}}^{3}}{16{L}^{4}}\left[\text{J}/{\text{m}}^{2}\right]$$ where t b [m] is the thickness of the blade, E [Pa] is the Young's modulus, t w [m] is the thickness of the wafer, and L [m] is the delamination length 38 – 47 . The G c value calculated from the exact same sample with the image in Fig. 2 is 6.08 J/m 2 . To verify the compatibility between DCB testing and NI testing, we prepared W2W samples using various plasma activation powers during the bonding process. The plasma exposure time was fixed, and the plasma intensity was set at 30 W or 100 W. Previous studies have reported that increasing the plasma activation intensity enhances the bond strength 48 . The W2W samples were post-bond annealed at 350°C for 1 h to stabilize their bond strength. Scanning acoustic microscopy (SAM) observations confirmed that the W2W samples were void-free before the evaluation of their bond strength, thereby ensuring that the measurements were conducted under ideal conditions. Bond strength evaluations by the DCB test were conducted under two conditions—an anhydrous atmosphere and ambient atmosphere—using the same wafer but at different locations. Symbol plots show the DCB test results under an anhydrous atmosphere for each plasma condition in the W2W samples. The bond-strength reduction due to the influence of WSC induced by moisture in the ambient atmosphere is plotted on the same axis. NI testing is susceptible to variations due to minor external factors. To account for this variation, we conducted more than 30 measurements on each W2W sample within a 15 µm × 15 µm measurement area. Box plots show the NI test results. The bonding energy measured by the DCB test for the W2W sample bonded at a plasma intensity of 30 W was 6.16 J/m². The average bonding energy measured by the NI test at the same location was 6.46 J/m². The bonding energy obtained by the DCB test fell within the quartile range of the NI test results. A comparison shows a 4.8% difference in bonding energy between the DCB test and the NI test. The bonding energy measured by the DCB test for the W2W sample bonded at a plasma intensity of 100 W was 8.2 J/m², and the average bonding energy measured by the NI test was 7.92 J/m². The bonding energy obtained by the DCB test fell within the quartile range of the NI test results, and a comparison shows a 3.3% difference in bonding energy between the DCB test and the NI test. Therefore, the bond strength measured by the DCB test in the glove box was found to be equivalent and comparable to the results obtained by the NI test. This correlation also indicates that the bonding interface remained protected from exposure to air during the NI process. As a result, the potential influence of WSC on the bond strength was minimized. Therefore, we concluded that the NI test is a reliable and effective method for measuring wafer bond strength. The sharp cube corner tip angle used in this study had an internal angle of 90° and an angle of 68° between the central axis and the plane, oriented perpendicular to the thin film during peeling. No plastic deformation was observed at the delamination tip, suggesting that mode I fracture predominated in this NI test 34 , 37 . In the thermal grown SiO 2 film, which had a Poisson's ratio of 0.17, minimal lateral deformation occurred under the indenter pressure during testing, concentrating the force in the vertical direction. As a brittle material, Th-SiO 2 fractures more easily under tensile stress than under shear stress. The pressing of the indenter generates tensile stress inside the film, making interfacial delamination more likely to progress 35 , 36 . Thus, delamination growth in the NI test, a mixed mode of I and II, is largely influenced by mode I. Therefore, comparisons with DCB testing, where delamination growth is dominated by mode I, are highly plausible. One of the origins of the difference in measured bonding strength between the DCB test and the NI test is likely the influence of mode II. However, the DCB test exhibited a coefficient of variation (CV) of less than 2% as a result of the stable measurement environment. By contrast, the NI test was more susceptible to minor external factors, resulting in a CV of approximately 10%. Although measuring the bonding energy at a single point was difficult, statistical estimation from multiple measurement points yielded results compatible with those of the DCB test. Several factors may affect the results of NI testing. Surface roughness of the film can influence measurement variation, especially for extremely thin films. However, the root means square roughness ( Rq ) for the 30 W and 100 W samples after wet TMAH etching was 0.15 nm and 0.19 nm, respectively, indicating a negligible influence. In addition, the precision of the mechanical control of the equipment is critical because NI testing is a destructive test. The root mean square error of the depth calibration was 5.21×10 − 6 µm/V², with an indentation depth error of less than 0.0035%. The CV value of the indentation depth error in actual measurements was as small as 0.46%, confirming the minimal effect of calibration error. In addition, there are variability factors that are difficult to deal with in NI tests, such as local density variations, nonuniform residual stress and vibration, and differences in measurement processes. Therefore, we measured the bond strength by statistically summarizing these variabilities in the following NI tests. Verification of the repeatability using wafers subjected to different post-bond annealing conditions Figure 6 shows the bond strength of W2W after post-bond annealing, as measured using both the DCB and NI tests. The horizontal axis represents the post-bond annealing temperature. As the post-bond annealing temperature increases, the bonding energy increases in both the DCB and NI tests. These results suggest that the post-bond annealing treatment enhances chemical bonding at the interface, thereby improving the bond strength 28 . For the W2W samples that were not annealed, the bond strength measured by the DCB test was 3.46 J/m², whereas that measured by the NI test was 2.92 J/m². However, when post-bond annealing was conducted at 150°C, 250°C, and 350°C, the bond strengths from the two tests were similar. Because the NI test measures local bond strength and the DCB test measures macroscopic bond strength, we speculate that the difference in localized bond strength contributes to the variability in bond strength at low temperatures. In the DCB test, the bond strength showed minimal variability from the non-annealed sample to the sample post-bond annealed at 350°C, with an average standard deviation of ~ 0.16 J/m². By contrast, the NI test showed suppressed variability at higher annealing temperatures, likely because weak bonds result in larger peeling areas, with mechanical control errors contributing to differences in the peeling area. In addition, because the NI test is destructive, bulk fracture of the film can occur in samples with low bond strength. Nevertheless, the validity of the NI test was reaffirmed because the bond strength calculated from both tests fell within the measurement variation for all of the post-bond annealing ranges. These results suggest that the NI test is effective for evaluating WiW uniformity and could be useful for quantitatively assessing the effect of post-bond annealing treatments in improving WiW uniformity of bond strength. WiW uniformity of bond strength evaluated by non-site-specific measurement The W2W samples were annealed at 350°C for 1 h to quantitatively evaluate the WiW uniformity of bond strength. Forty-nine points were measured on a 300 mm Si wafer by the NI test. More than 30 measurements were performed at each measurement point, and the average value excluding outliers was used as the bonding energy at each measurement point. Figure 7 a shows a histogram of the bonding energy measured at 49 points by the NI test. Overlaying a normal distribution on this graph reveals that the data roughly follow a normal distribution, indicating that the variation in the bonding energy within the wafer surface is low. Figure 7 b shows a color map plotting the bonding energy at each measurement point. The average bonding energy across the wafer surface was 3.81 J/m². Because the mass production range of variation in the semiconductor industry is generally ~ 5% or less, the range of the CV value within 5% was mapped to the same color. When the notch part of the wafer was set to 0 degrees, the bonding energy showed a maximum value of 8.6 J/m² at 270 degrees and a minimum value of 6.94 J/m² at 30 degrees. The CV value of this sample was 5.46%. In this measurement, the bonded sample was diced into coupon-sized pieces, then ground, wet etched, and mounted onto the measurement stage. Sources of variation may have been present in these multiple sample preparation processes. However, given the measurement error of the NI test, the bonding energy within the wafer surface can be regarded as approximately uniform. Comparison of D2W and W2W bond strength W2W and D2W samples were bonded under the same plasma activation and cleaning conditions, followed by a post-bond annealing treatment at 150°C, 250°C, or 350°C for 1 h to evaluate the bond strength. A comparison between W2W and D2W samples prepared under the same bonding conditions validated the bond-strength evaluation of the D2W samples. An advanced flip-chip bonder with high alignment accuracy was used to carry out the bonding (Fig. 8 a). Equipment with mass-production-equivalent technology was used for dicing and grinding to minimize bonding defects 3 , 37 . Figure 8 b shows an SAM image of a D2W sample immediately after bonding, where no voids are observed. Figure 8 c shows a SAM image of the same sample after annealing at 350°C for 1 h. No noticeable increase in voids was visually observed between the two images. The void ratios, including microvoids, were 3% and 4%, respectively, which are within the measurement error range, suggesting that annealing did not lead to an increase in voids. Figure 9 shows a comparison of the bond strength obtained for W2W and D2W samples. W2W is the result of the DCB test, which has the least variability, whereas D2W is the bond strength calculated from the NI test. As previously described, the bonding energy of the W2W sample increases linearly with increasing annealing temperature, showing a trend similar as the D2W sample. The CV values for samples post-bond annealed at 150°C, 250°C, and 350°C are 7.81%, 5.05%, and 5.5%, respectively. These values also show that samples with lower bonding energy exhibit greater variation, as observed in the previous section. The D2W sample exhibited a rapid increase in bonding energy between post-bond annealing temperatures of 150°C and 250°C, indicating that a large difference in bond strength occurred between D2W and W2W at an annealing temperature of 150°C. The observed differences are likely attributable to variations in the W2W and D2W bonding processes. The procedures up to the DIW cleaning step are the same for both the W2W and D2W processes. However, a difference exists in the queue time ( Q -time) between cleaning and bonding. In the W2W process, bonding was performed immediately after plasma activation and DIW cleaning using a cluster tool. By contrast, the D2W process requires additional Q -time for sample transport from plasma activation and DIW cleaning. In addition, the sequential (direct placement) bonding step creates a time lag between each die bonding, causing changes in the surface/interface status. Fusion bonding requires sufficient moisture on the bonding film surface to achieve high bond strength. During the Q -time, moisture at the top surface may evaporate at the interface, potentially reducing the bond strength in the D2W samples. In addition, the delay between cleaning and bonding can introduce organic contaminants and residual particles. In low-temperature treatments, silanol groups at the interface may not be sufficiently activated, resulting in a substantial reduction in bond strength. At temperatures between 0°C and 200°C, only hydrogen bonds between silanol groups are active, and a low number of silanol groups due to particle presence can further reduce bond strength. However, at temperatures above 200°C, dehydration condensation reactions are more likely to influence the chemical properties of the interface, leading to similar bond strengths for W2W and D2W samples 49 , 50 . This finding suggests that bond strength in D2W samples can be measured, which may contribute to assessing verification of integration flow for D2W hybrid bonding. WiD uniformity NI tests at 17 points within the die of the D2W sample annealed at 350°C for 1 h confirmed that the WiD uniformity of bond strength in the post-bond annealing process to similar to that in the W2D process. At each measurement point, more than 30 measurements were conducted and the average value, excluding outliers, was considered the bonding energy at each point. The resulting data were plotted in a color map (Fig. 10 ). The center of the die exhibited a high bonding energy of 9.57 J/m 2 , whereas the bonding energy decreased toward the outer regions. In particular, the bonding energy at the edge was 6.78 J/m 2 , or approximately 20% lower than that at the center. This trend is consistent along both the x - and y -axes of the die, indicating that the variation is not due to measurement discrepancies but rather a substantial WiD variation in bond strength. In addition, with a standard deviation of 1.01 J/m 2 and a CV of 12.3%, the WiD uniformity is low, revealing a substantial lack of consistency. The graph also shows that the bonding energy distribution is skewed to the right, suggesting that the pressure during bonding or the heat propagation during annealing may have been applied more on the right side. It is also possible that warping of the wafer or die occurred after the bonding and annealing treatment, which could have contributed to this variation. Conclusion WiW and WiD bond strength in both W2W and D2W samples were examined by measuring the bond energy through NI testing. The widely used DCB test cannot measure the WiW uniformity of bond strength because blade insertion is only possible at the edge. In addition, there is no clear method for quantitatively measuring bond strength at the D2W interface because the blade cannot be inserted. Therefore, NI testing was performed to measure and confirm WiW uniformity of bond strength. Our test result suggests that the process accuracy is sufficiently high for NI testing to be reliable. In addition, the ability to measure D2W bond strength revealed differences in the bonding energy between D2W and W2W at certain annealing temperature ranges, indicating potential differences in bonding mechanisms. Furthermore, the measurement of WiD uniformity of bond strength is expected to contribute to the maturation of the D2W processes. A challenge in NI testing is the large measurement variations. Future efforts should focus on identifying the causes of these errors and establishing a highly reproducible measurement method. The introduction of NI testing will likely contribute to the advancement of D2W bonding for chiplet integration by providing valuable bond-strength measurements. Thus, this technique could be important for the development of D2W hybrid bonding in a chiplet integration, which is currently an active research area for semiconductors for AI. Standardizing bond-strength measurement methods will enhance process reliability in semiconductor manufacturing. Evaluating WiW and WiD uniformity will enable further optimization of processes and will support the development of advanced semiconductor devices. Methods Wafer-to-Wafer (W2W) and Die-to-Wafer (D2W) Bonding The samples used in this study were all prepared from 300 mm Si (100) wafers with a thickness of 775 µm. The W2W and D2W samples were prepared from the same batch of Si wafers. The bonding dielectric film was SiO 2 (thermal Si oxide film) with a thickness of 100 nm ± 5%, which was formed on the Si wafer through steam oxidation. For W2W, a N 2 plasma treatment was applied to the samples, followed by a water rinse; direct bonding was then carried out under ambient conditions. These processes were conducted using high-volume-production-compatible fully automated cluster bonding equipment to minimize the variation among samples (within each batch) in the W2W process. For D2W, the top sample, which serves as the die, was thinned to 200 µm and singulated into 10 mm × 10 mm before the bonding sequence. After the pre-assembly processes, N 2 plasma activation and water rinse treatments like those used in W2W were applied. The bottom substrate remained at 300 mm and was subjected to the same plasma and water rinse treatments as the top dies. Bonding was then carried out using a flip-chip method with hybrid-bonding-compatible die bonder equipment. Bond strength measurement methods The NI tests were performed using a Bruker TS77. The measurements were conducted at room temperature (18–25°C) and at a relative humidity of 40 ± 10% (ambient atmosphere) using a diamond cube corner indenter. A dedicated bulk Si removal process was required to access the NI indenter at the bonding interface. After bonding or post-bond annealing, the W2W samples were diced. The diced coupon samples were ground/polished to remain ~ 100 µm of one side of the bulk Si. Selective wet etching was subsequently performed using 25% tetramethyl ammonium hydroxide (TMAH) at 70°C to expose the bonding interface 51 – 53 , which allowed the indenter to induce stress, causing delamination at the bonding interface. For the D2W samples, a similar process was applied to expose the bonding layer after sample preparation; however, the coupon grinding step was omitted because the Si bulk had already been removed before bonding. Declarations Acknowledgements This work was supported by the JSPS KAKENHI Grant Number 23K26388 and the JST PRESTO Grant Number JPMJPR22B3, Japan. Author contributions J.F. and Y.Y. were primarily responsible for the evaluations, measurements, and analyses in this study. Y.K. and D.K. handled sample preparation and process experiment management. M.S. contributed by developing Python-based analysis code for the nanoindentation test. F.I. supervised all aspects of this study. Competing interests The authors declare no competing interests. Data availability The data that support the findings of this study are available from the corresponding author upon reasonable request. References J. H. Lau, “Recent Advances and Trends in Advanced Packaging,” in IEEE Transactions on Components, Packaging and Manufacturing Technology, vol. 12, no. 2, pp. 228–252, 2022. O¨. Vallin, K. Jonsson, U. Lindberg, “Adhesion quantification methods for wafer bonding,” Materials Science and Engineering, vol.50 pp. 109–165, 2005. F. Inoue, et al. “Inorganic Temporary Direct Bonding for Collective Die to Wafer Hybrid Bonding,” 2023 IEEE 73rd Electronic Components and Technology Conference (ECTC), 2023. Y. Kagawa, et al. “Novel stacked CMOS image sensor with advanced Cu2Cu hybrid bonding,” 2016 IEEE International Electron Devices Meeting (IEDM), 2016. C.-T. Ko, et al. “Low temperature bonding technology for 3D integration,” Microelectronics Reliability, 52, 302–311, 2012. Y. Kagawa, et al. “Development of face-to-face and face-to-back ultra-fine pitch Cu-Cu hybrid bonding,” 2022 IEEE 72nd Electronic Components and Technology Conference (ECTC), 2022. E. Beyne, et al. “Scalable, sub 2μm Pitch, Cu/SiCN to Cu/SiCN Hybrid Wafer-to-Wafer Bonding Technology,” 2017 IEEE International Electron Devices Meeting (IEDM), 2017. T. Uhrmann, et al. “Heterogeneous Integration by Collective Die-To-Wafer Bonding,” 2018 International Wafer Level Packaging Conference (IWLPC), 2018. F. Inoue, et al. “Advanced Dicing Technologies for Combination of Wafer to Wafer and Collective Die to Wafer Direct Bonding,” 2019 IEEE 69th Electronic Components and Technology Conference (ECTC), 2019. F. Inoue, et al. “Protective layer for collective die to wafer hybrid bonding,” 2019 International 3D Systems Integration Conference (3DIC), 2019. Debendra Das Sharma “Universal Chiplet Express (UCle)®: Building an open chiplet ecosystem” 2022. Lau, J. H. “Recent Advances and Trends in Advanced Packaging,” IEEE Transactions on Components, Packaging and Manufacturing Technology, 12 (2), 228-252, 2022. A. Elsherbini, et al. “Enabling Next Generation 3D Heterogeneous Integration Architectures on Intel Process,” 2022 International Electron Devices Meeting (IEDM), 2022. Thomas Uhrmann, et al, “Accelerating 3D and heterogeneous integration with high-volume D2W hybrid bonding”, Chip Scale Review the Future of Semiconductor Packaging, Volume 27, Number 2, 2023. W. Hafez et al., "Intel Power Via Technology: Backside Power Delivery for High Density and High-Performance Computing," 2023 IEEE Symposium on VLSI Technology and Circuits, Kyoto, Japan, 2023. M. Liao, D. H. Zhou, P. Wang and M. Chen, “Power Systems on Chiplet: Inductor-Linked Multi-Output Switched-Capacitor Multi-Rail Power Delivery on Chiplets,” 2023 Fourth International Symposium on 3D Power Electronics Integration and Manufacturing (3D-PEIM), Miami, FL, USA, 2023, pp. 1-7. T. Sherwood et al., "300 nm Pitch W2W HBI for CFET and 3D DRAM Through Module Co-optimization," 2024 IEEE 74th Electronic Components and Technology Conference (ECTC), Denver, CO, USA, 2024, pp. 1968-1972. X. Ai et al ., "First Demonstration of 4-Layer Stacked Planar Channel-All-Around (P-CAA) IGZO FETs with Cost-Effective Process for High-Density 1T1C 3D DRAM," 2024 IEEE International Electron Devices Meeting (IEDM) , San Francisco, CA, USA, 2024, pp. 1-4. M. Faure et al ., "Ultra Low Temperature Hybrid Bonding: Morphological and Electrical Characterizations," 2024 IEEE 10th Electronics System-Integration Technology Conference (ESTC) , Berlin, Germany, 2024, pp. 1-6. T. Sherwood et al ., "300 nm Pitch W2W HBI for CFET and 3D DRAM Through Module Co-optimization," 2024 IEEE 74th Electronic Components and Technology Conference (ECTC) , Denver, CO, USA, 2024, pp. 1968-1972. M. Oda et al ., "Superior Scalability of Advanced Horizontal Channel Flash for Future Generations of 3D Flash Memory," 2024 IEEE International Electron Devices Meeting (IEDM) , San Francisco, CA, USA, 2024, pp. 1-4. S. H. Hahn, W. Kim, K. Lim, B. Moon and M. Rhee, "Achieving Sub-nm Copper Recess Controllability for Advanced 3D Integration: An Experimental and Atomic-scale Simulation Study on Wet Atomic Layer Etching Process," 2024 IEEE 74th Electronic Components and Technology Conference (ECTC) , Denver, CO, USA, 2024, pp. 1322-1328. T. Komljenovic et al., “Photonic Integrated Circuits Using Heterogeneous Integration on Silicon,” in Proceedings of the IEEE, vol. 106, no. 12, pp. 2246-2257, Dec. 2018 X. Luo, Y. Cheng, J. Song, T. -Y. Liow, Q. J. Wang and M. Yu, “Wafer-Scale Dies-Transfer Bonding Technology for Hybrid III/V-on-Silicon Photonic Integrated Circuit Application,” in IEEE Journal of Selected Topics in Quantum Electronics, vol. 22, no. 6, pp. 443-454, Nov.-Dec. 2016. Uhrmann, Thomas, Peter Urban, Markus Wimplinger, Boris Povazay, Julian Bravin, and Bernd Thallner. 2023. “Novel IR Laser Debonding for Heterogeneous and 3D Integration.” IMAP Source Proceedings 2023 (DPC): 354–76. D. Zehua, H. Kikuchi, H. Hishinuma, M. Murugesan, T. Tanaka and T. Fukushima, “Liquid Surface Tension-Driven Chip Self-Assembly Technology with Cu-Cu Hybrid Bonding for High-Precision and High-Throughput 3D Stacking of DRAM,” 2024 IEEE 74th Electronic Components and Technology Conference (ECTC), Denver, CO, USA, 2024, pp. 1335-1341. Junya Fuse, Tomoya, Iwata, Sodai Ebiko, Fumihiro Inoue, “Robust Measurement of Bonding Strength for Wafer-to-Wafer 3D Integration”, International Conference on Electronics Packaging 2023, Kumamoto, April 20, 20223. Tomoya Iwata, Junya Fuse, Yuki Yoshihara, Yusuke Kondo, Marie Sano, Fumihiro Inoue, “Water stress corrosion at wafer bonding interface during bond strength evaluation,” Materials Science in Semiconductor Processing, Volume 184, 2024, W. C. Oliver and G. M. Pharr “An Improved Technique for Determining Hardness and Elastic Modulus Using Load and Displacement Sensing Indentation Experiments,” J. Mater. Res., 7–6, 1564/1583 (1992). Ming Yuan He et al., “Crack deflection at an interface between dissimilar elastic materials: Role of residual stresses,” International Journal of Solids and Structures, Volume 31, Issue 24, 1994. D. B. Marshall and A.G. Evans, “Measurement of adherence of residually stressed thin films by indentation. I. Mechanics of interface delamination,” Journal of Applied Physics, Volume 56, Issue 10, 1984. A. A. Volinsky et al., “Interfacial toughness measurements for thin films on substrates,” Acta Materialia, Volume 50, Issue 3, 8, P441-466, 2002. F. Fournel, C. Martin-Cocher, D. Radisson, V. Larrey, E. Beche, C. Morales, P. A. Delean, F. Rieutord, H. Moriceau, Water stress corrosion in bonded structures, ECS J. Solid State Sci. Technol. 4 (2015) 124. https://iopscience.iop.org/article/10 .1149/2.0031505jss. Jinju Chen and S J Bull, “Approaches to investigate delamination and interfacial toughness in coated systems: an overview,” 2011 J. Phys. D: Appl. Phys. 44 034001. Yeap KB, Zeng KY, Chi DZ., “Determining the interfacial toughness of low-k films on Si substrate by wedge indentation: Further studies,” Acta Materialia 2008;56:977. Yeap KB, Zeng KY, Jiang HY, Shen L, Chi DZ., “Determining interfacial properties of submicron low-k films on Si substrate by using wedge indentation technique,” Journal of Applied Physics 2007;101. Weizhou Li, Thomas Siegmund, “An analysis of the indentation test to determine the interface toughness in a weakly bonded thin film coating – substrate system,” Acta Materialia, Volume 52, Issue 10, 2004, 2989-2999, K. Onishi, H. Kitagawa, S. Teranishi, A. Uedono, F. Inoue, “Temporary Direct Bonding by Low Temperature Deposited SiO2 for Chiplet Applications” DOI 10.1021/acsaelm.4c00114 (Accepted in ACS Applied Electronic Materials). W. Maszara, G. Goetz, A. Caviglia, L.B. Mckitterick, Bonding of silicon wafers for silicon-on-insulator, J. Appl. Phys. 64 (1988) 4943–4950, https://doi.org/ 10.1063/1.96768. K.T. Turner, S.M. Spearing, Accurate characterization of wafer bond toughness with the double cantilever specimen, J. Appl. Phys. 103 (2008) 013514, https:// doi.org/10.1063/1.2828156. A.B de Morais, M.F de Moura, J.P.M. Gonçalves, P.P. Camanho, Analysis of crack propagation in double cantilever beam tests of multidirectional laminates, Mech. Mater. 35 (7) (2003) 641–652, https://doi.org/10.1016/S0167-6636(02)00289-2. K. Takeuchi, T. Suga, Quantification of wafer bond strength under controlled atmospheres, Jpn. J. Appl. Phys. 61 (2022) SF1010. https://iopscience.iop.org/arti cle/10.35848/1347-4065/ac5e49. Y. Bertholet, F. Iker, J.P. Raskin, T. Pardoen, Steady-state measurement of wafer bonding cracking resistance, Sensor Actuator Phys. 110 (Issues 1–3) (2004) 157–163, https://doi.org/10.1016/j.sna.2003.09.004. F. Fournel, L. Continni, C. Morales, J. Da Fonseca, H. Moriceau, F. Rieutord, A. Barthelemy, I. Radu, Measurement of bonding energy in an anhydrous nitrogen atmosphere and its application to silicon direct bonding technology, J. Appl. Phys. 111 (2012) 104907, https://doi.org/10.1063/1.4716030. P.P. Gillis, J.J. Gilman, Double-cantilever cleavage mode of crack propagation, J. Appl. Phys. 35 (1964) 647–658, https://doi.org/10.1063/1.1713430. K. Pantzas, F. Fournel, A. Talneau, G. Patriarche, E. Le Bourhis, Measuring the surface bonding energy: a comparison between the classical double-cantilever beam experiment and its nanoscale analog, AIP Adv. 10 (2020) 045006, https:// doi.org/10.1063/1.5143843. M.S. El-Zein, K.L. Reifsnider, Evaluation of GIC of a DCB specimen usin an anisotropic solution, J. Compos. Technol. Res. 10 (4) (1988) 151, https://doi.org/ 10.1520/CTR10278J. Ryosuke Gösele, U.; Tong, “SEMICONDUCTOR WAFER BONDING,” Q.-Y. ANNUAL REVIEW OF MATERIALS SCIENCE, 01 Aug 1998, Vol. 28, Issue 1, pages 215 – 241. T. Plach, K. Hingerl, S. Tollabimazraehno, G. Hesser, V. Dragoi, M. Wimplinger “Mechanisms for room temperature direct wafer bonding.” J. Appl. Phys. 7 March 2013; 113 (9): 094905. K. Biswas, S. Kal, “Etch characteristics of KOH, TMAH and dual doped TMAH for bulk micromachining of silicon,” Microelectronics Journal, Volume 37, Issue 6, 2006, Pages 519-525, Ping-Hei Chen, Hsin-Yah Peng, Chia-Ming Hsieh, Minking K Chyu, “The characteristic behavior of TMAH water solution for anisotropic etching on both Silicon substrate and SiO2 layer,” Sensors and Actuators A: Physical, Volume 93, Issue 2, 2001, Pages 132-137, Jae Sung You, Donghwan Kim, Joo Youl Huh, Ho Joon Park, James Jungho Pak, Choon Sik Kang, “Experiments on anisotropic etching of Si in TMAH,” Solar Energy Materials and Solar Cells, Volume 66, Issues 1–4, 2001, Pages 37-44. Additional Declarations There is NO Competing Interest. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6298479","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":435325614,"identity":"1804a3a7-3b38-48c7-920e-7fb2e3f5c1eb","order_by":0,"name":"Fumihiro Inoue","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABEElEQVRIiWNgGAWjYBACNgglx8DAzMNwAMxmb4BJHsCnxRiuRYKBB64SuxYGuBYGHjBLgkEiAb/D+KSbj0l8YDBIXNvOe/BwRQVDncHNN2aPeRjs5BkYz2K1hk3mWJrkDKCWbYf5Eg6eOcMgYXA7x9yYhyHZsIHhHFb72CRyzKR5//0BauExONjY9h+kxUyah4EZqPyMAU4tf8C2gLUAbbl5BqSlHr8WBhQtN3hAWg7j0ZKWbNnDYGAM1tJwhkFy5pm0Msk5BscN23D4RX5G8sEbPxgMZLedP2P8saGCgZ/v+OFtEm8qquX5JbCHGCZQOMABdI8ByAFniNPBIN/A/gDC4u8hUssoGAWjYBQMcwAAhvFZUOgaUwYAAAAASUVORK5CYII=","orcid":"","institution":"Yokohama National University","correspondingAuthor":true,"prefix":"","firstName":"Fumihiro","middleName":"","lastName":"Inoue","suffix":""},{"id":435325615,"identity":"b91ce996-628e-4569-8dbc-92eb962df696","order_by":1,"name":"Junya Fuse","email":"","orcid":"","institution":"Yokohama National University","correspondingAuthor":false,"prefix":"","firstName":"Junya","middleName":"","lastName":"Fuse","suffix":""},{"id":435325616,"identity":"836a377b-e6e0-4743-a48d-64b16e9ff135","order_by":2,"name":"Yuki Yoshihara","email":"","orcid":"","institution":"Yokohama National University","correspondingAuthor":false,"prefix":"","firstName":"Yuki","middleName":"","lastName":"Yoshihara","suffix":""},{"id":435325617,"identity":"3a045b4e-04ef-4681-9e4c-e53a3d470528","order_by":3,"name":"Yusuke Kondo","email":"","orcid":"","institution":"Yokohama National University","correspondingAuthor":false,"prefix":"","firstName":"Yusuke","middleName":"","lastName":"Kondo","suffix":""},{"id":435325618,"identity":"c20bcafe-424f-42ec-8d9b-ec75e39f9641","order_by":4,"name":"Daiki Kobayashi","email":"","orcid":"","institution":"Yokohama National University","correspondingAuthor":false,"prefix":"","firstName":"Daiki","middleName":"","lastName":"Kobayashi","suffix":""},{"id":435325619,"identity":"dd6f13fd-58fd-4fde-b704-2f52716b40de","order_by":5,"name":"Marie Sano","email":"","orcid":"","institution":"Yokohama National University","correspondingAuthor":false,"prefix":"","firstName":"Marie","middleName":"","lastName":"Sano","suffix":""}],"badges":[],"createdAt":"2025-03-24 22:20:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6298479/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6298479/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":80733468,"identity":"c35a51f7-1016-4cd6-a4eb-22076f19c677","added_by":"auto","created_at":"2025-04-16 13:04:44","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1200818,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSchematic of the adherence energy measurement by nanoindentation on direct bonding interfaces.\u003c/strong\u003eBird’s view on entire area and cross-sectional view at around indentation.\u003c/p\u003e","description":"","filename":"Fig.1Nanoindentation.png","url":"https://assets-eu.researchsquare.com/files/rs-6298479/v1/61839f81c24d48bd67322745.png"},{"id":80733473,"identity":"2778f6ba-2360-44b1-9e34-de87021b1cf9","added_by":"auto","created_at":"2025-04-16 13:04:44","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":2330083,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLoad–indentation depth curves.\u003c/strong\u003e The data were acquired during the nanoindentation tests of SiO\u003csub\u003e2\u003c/sub\u003e–SiO\u003csub\u003e2\u003c/sub\u003e films at several depths ranging from 70 nm to 150 nm. The film thickness on one side of the SiO\u003csub\u003e2\u003c/sub\u003e was 100 nm.\u003c/p\u003e","description":"","filename":"Fig.2PL70150nm.png","url":"https://assets-eu.researchsquare.com/files/rs-6298479/v1/8fee45945f26ca795e4a2d17.png"},{"id":80734665,"identity":"5d169a32-25d9-441e-9e5d-09d252102f1a","added_by":"auto","created_at":"2025-04-16 13:12:44","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2828538,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eElectron microscopy observations of bond strength.\u003c/strong\u003e Weak bond strength was observed by \u003cstrong\u003ea \u003c/strong\u003eSEM and \u003cstrong\u003ec\u003c/strong\u003e TEM, and strong bond strength was also observed by \u003cstrong\u003eb\u003c/strong\u003e SEM and \u003cstrong\u003ed\u003c/strong\u003e TEM.\u003c/p\u003e","description":"","filename":"Fig.3SEMTEM.png","url":"https://assets-eu.researchsquare.com/files/rs-6298479/v1/b0ea988fec93402f36c77fb1.png"},{"id":80734666,"identity":"125aedde-241a-4f3d-a5f8-8663d995aacf","added_by":"auto","created_at":"2025-04-16 13:12:44","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":2840814,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDetails of the nanoindentation test. \u003c/strong\u003eAn indentation caused by the NI test at the bonding interface, as represented in \u003cstrong\u003ea\u003c/strong\u003e a 3D plot,\u003cstrong\u003e b\u003c/strong\u003e an SPM image, and\u003cstrong\u003e c\u003c/strong\u003ean analysis image.\u003c/p\u003e","description":"","filename":"Fig.4NI3DplotsSPM.png","url":"https://assets-eu.researchsquare.com/files/rs-6298479/v1/46567ae3d7ba13be3e9b8f71.png"},{"id":80733475,"identity":"861a3328-0c35-49fc-95a2-ae139b0f898a","added_by":"auto","created_at":"2025-04-16 13:04:44","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":609783,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eComparison of bonding strength measured by DCB and NI testing. \u003c/strong\u003eSymbol plots show the DCB test value under an anhydrous atmosphere in the W2W samples and box plots indicate the value obtained by NI test. The blue at left side of the graph is at plasma power 30W, the green at right side of the graph is 100 W.\u003c/p\u003e","description":"","filename":"Fig.5DCBvsNI.png","url":"https://assets-eu.researchsquare.com/files/rs-6298479/v1/86ad7f05642a3c711d20e002.png"},{"id":80734690,"identity":"21781b27-1b5a-45b1-b5e6-882939d15f66","added_by":"auto","created_at":"2025-04-16 13:12:47","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":167408322,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe bonding energy of W2W samples annealed at various post-bond annealing temperatures.\u003c/strong\u003eThe bonding energy was measured using DCB and NI testing.\u003c/p\u003e","description":"","filename":"Fig.6ThW2WPBA.png","url":"https://assets-eu.researchsquare.com/files/rs-6298479/v1/841b0345918cf38948795533.png"},{"id":80733471,"identity":"c865ef52-dbf6-4d63-acc2-98fb85121fc7","added_by":"auto","created_at":"2025-04-16 13:04:44","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":3079997,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eWiW uniformity of bonding energy. a\u003c/strong\u003e A histogram of the bonding energy measured at 49 points, overlaid with a Gaussian distribution and \u003cstrong\u003eb \u003c/strong\u003ethe corresponding color map of the bonding energy at each measurement point.\u003c/p\u003e","description":"","filename":"Fig.7WiW.png","url":"https://assets-eu.researchsquare.com/files/rs-6298479/v1/aa06c575053be777ed0cbc4d.png"},{"id":80734667,"identity":"6f06d4ad-7522-437f-a38a-f918881c0bda","added_by":"auto","created_at":"2025-04-16 13:12:44","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":2470264,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eD2W samples.\u003c/strong\u003e \u003cstrong\u003ea\u003c/strong\u003e Sample images, \u003cstrong\u003eb\u003c/strong\u003e a SAM image of a sample prepared without annealing, and \u003cstrong\u003ec\u003c/strong\u003e a SAM image of a sample annealed at 350°C for 1 h.\u003c/p\u003e","description":"","filename":"Fig.8D2Wsample.png","url":"https://assets-eu.researchsquare.com/files/rs-6298479/v1/03ef3d9625374034a4b925da.png"},{"id":80734679,"identity":"c872714d-0f05-4091-8f0d-630695ed4fc1","added_by":"auto","created_at":"2025-04-16 13:12:45","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":77729361,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eComparison of the bonding energy between W2W and D2W samples annealed at various post-bond annealing temperatures range from 150 – 350 °C. \u003c/strong\u003eSymbol plots show the value obtained by DCB under an anhydrous atmosphere on W2W samples. Box plots show the value obtained by NI test on D2W samples.\u003c/p\u003e","description":"","filename":"Fig.9ThW2WD2W.png","url":"https://assets-eu.researchsquare.com/files/rs-6298479/v1/e192d6f36155401e4037f943.png"},{"id":80733529,"identity":"d1577de7-65f6-4526-a9d1-ea50ef4b7690","added_by":"auto","created_at":"2025-04-16 13:04:47","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":179357059,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eColor map for WiD uniformity of the bonding energy.\u003c/strong\u003e The corresponding color map of the bonding energy at each measurement points with D2W. The line graphs represent the WiD uniformity of bonding energy along each axis direction.\u003c/p\u003e","description":"","filename":"Fig.10ThD2WWiD.png","url":"https://assets-eu.researchsquare.com/files/rs-6298479/v1/c5c41e606a871ea8062ce278.png"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Microscopic and Non-Site-Specific Measurement Method for Bond Strength of Direct Bonded Wafers and Dies","fulltext":[{"header":"Introduction","content":"\u003cp\u003eThree-dimensional (3D) integration technology enhances the performance, integrity, and energy efficiency compared with those of monolithic two-dimensional (2D) devices in diverse device applications. In addition, 2.5D/3D chiplet integration provides design flexibility and cost efficiency while enabling high yield through known-good-die (KGD) selection\u003csup\u003e\u003cspan additionalcitationids=\"CR2 CR3 CR4 CR5 CR6 CR7 CR8 CR9 CR10 CR11 CR12 CR13\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Plasma-activated direct wafer-to-wafer (W2W) bonding is being implemented in wafer-level 3D integration because it is compatible with front-end-of-line processes (e.g., backside power delivery network, complementary FET (CFET), and 3D DRAM)\u003csup\u003e\u003cspan additionalcitationids=\"CR16 CR17 CR18 CR19\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. In addition, Cu hybrid bonding is used for image sensors and 3D NAND memories for higher-performance electrical connections, where it provides a greater density of input/output (I/O) connections while minimizing inductance, capacitance, and resistance, enabling faster and more energy-efficient signal transmission\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e. Die-level hybrid bonding is also crucial for chiplet architectures and high-bandwidth memory, where solder-based microbumps are currently the conventional integration method. Die-level integration is mandatory as the conceptional background for these integration schemes (e.g., KGD selection for high yield)\u003csup\u003e\u003cspan additionalcitationids=\"CR24 CR25\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. However, direct bonding technology, which uses inorganic dielectric films as bonding interfaces instead of organic adhesives, faces challenges with respect to reliably evaluating the bond strength.\u003c/p\u003e \u003cp\u003ePrevious studies have explored measurement methods based on the double-cantilever-beam (DCB) test commonly used to evaluate adherence energy (i.e., bond strength) at direct bonding interfaces. Such studies have revealed that the method provides robust and true bond strength when it is conducted in an anhydrous atmosphere like that in a glove box filled with inert gas\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e,\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. However, the method is limited in its ability to provide fundamental understanding of microscopic bonding dynamics. The DCB method can be used to calculate the adherence energy on the basis of the length of delamination caused by inserting a blade at the interface. The delamination caused by blades is relatively wide (e.g., a few square centimeters), which makes the measurement of microscopic bond strength difficult. In addition, variations in the within-wafer (WiW) uniformity of bond strength, which are caused by process variations, cannot be measured by the DCB method, because the method requires the insertion of a blade at the wafer edge/bevel. The data obtained by DCB are site-specific at the wafer edge and macroscopic. Furthermore, the method is not applicable for die-to-wafer (D2W) bonded samples, because the blade cannot be inserted from the die edge. Pull tests or share tests can measure the mechanical strength; however, the results of the measurements cannot be directly compared to the adherence energy obtained by DCB. These methods also often cause breakage of the bulk Si and therefore cannot provide interfacial information. Consequently, chiplet integration with direct/hybrid bonding necessitates a re-evaluation of adherence energy using an alternative approach.\u003c/p\u003e \u003cp\u003eThe present study aims to establish alternative bond-strength measurement methods for both W2W and D2W integration in direct bonding interfaces using a microscopic and non-site-specific method. We investigate the WiW uniformity of bond strength in wafer-level integration, which is crucial for assessing the robustness of 3D integration.\u003c/p\u003e \u003cp\u003eNanoindentation (NI) tests are widely used to measure mechanical properties such as the hardness and Young's modulus of thin films. In addition, recent studies have examined NI testing for evaluating thin-film adherence\u003csup\u003e\u003cspan additionalcitationids=\"CR30 CR31\" citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e. However, comprehensive studies on adherence energy measurements for direct bonding interfaces remain sparse and novel. In the present study, we carry out a detailed mechanistic examination of the NI method by comparing robust DCB measurement results as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. In addition, we conduct NI tests to confirm their effectiveness for evaluating WiW uniformity and within-die (WiD) uniformity of bond strength to promote robust 3D/chiplet integration.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"Results and discussion","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eVerification and optimization of measurement conditions\u003c/h2\u003e \u003cp\u003eThe load\u0026ndash;displacement curves acquired during the NI tests are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. The indentation was conducted at depths ranging from 70 nm to 150 nm. The structure of the samples consisted of two thermally oxidized Si layers, each 100 nm thick (bonding interface at 100 nm depth) on the bulk Si substrate as the bottom Si. Controlling the indentation depth was necessary to avoid delamination between the substrate and the lower bonding film (thermal Si oxide) layer. Therefore, the indentation depth was carefully controlled, and appropriate indentation conditions were determined from the acquired Scanning Probe Microscope (SPM\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e)\u003c/span\u003e images.\u003c/p\u003e \u003cp\u003eBelow the film thickness of the bonding dielectric as the indentation depth, no delamination or shape change was observed around the indentation mark. Near the top film thickness (~\u0026thinsp;100 nm) as the indentation depth, delamination was observed only at one edge of the triangular indentation. At the indentation depth of 130 nm (1.3\u0026times; deeper than the original top bonding dielectric), a shape change was observed around the indentation but the delamination was not uniform. At the indentation depth of 150 nm, delamination occurred around all three edges of the indentation, with a roughly uniform pattern.\u003c/p\u003e \u003cp\u003eIn load\u0026ndash;displacement curves corresponding to conventional indentation tests with delamination, a region where the load becomes constant upon delamination is expected to appear. However, no such phenomenon was observed in the samples tested in the present study. This lack of a constant-load region might be due to the change of the load on the thin film used in this experiment being below the detection limit of the indentation system. Thus, to minimize variations, we conducted indentation tests to a depth of 150 nm, where peeling occurred uniformly.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWe conducted an interface analysis using cross-sectional transmission electron microscopy (TEM) to observe the delamination shape in greater detail. Figure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea and b show top-view \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003escanning electron microscopy (SEM)\u003c/span\u003e images of indentations made during nanoindentation testing of samples with strong and weak bond strengths, as evaluated using the DCB method. Neither sample surface was chipped or broken. In addition, we observed that the change in the shape of the periphery was small in the sample with high bond strength. The micrographs of the cross sections of these indentations are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec and d. The horizontal length of the delamination at the SiO\u003csub\u003e2\u003c/sub\u003e interface was 260 nm in the case of low bond strength (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec), and the delamination length at the SiO\u003csub\u003e2\u003c/sub\u003e interface was 176 nm in case of high bond strength (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ed). No delamination was observed between the SiO\u003csub\u003e2\u003c/sub\u003e layer and the Si substrate in either sample.\u003c/p\u003e \u003cp\u003eIrrespective of the bond strength, the delamination was only observed at the SiO\u003csub\u003e2\u003c/sub\u003e\u0026ndash;SiO\u003csub\u003e2\u003c/sub\u003e interface, implying that the bond strength can be quantitatively calculated. In addition, because the bonded interface after the NI test was not exposed, we reasonably assumed that it did not react with humidity in the air during the indentation. Thus, the progress of interfacial delamination due to water stress corrosion (WSC), observed in the DCB test, can be reasonably assumed to not affect the NI test\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e,\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e. Consequently, the results of the DCB test in an anhydrous atmosphere should with compared those of the NI test under ambient atmosphere.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eComparison between DCB and NI\u003c/h3\u003e\n\u003cp\u003eTo confirm the compatibility between the DCB test and the NI test, we evaluated the bond strength at the edge using the DCB method. The bond-strength evaluation using NI was calculated from the delamination area after the indentation. This method is described in detail elsewhere\u003csup\u003e\u003cspan additionalcitationids=\"CR35\" citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. After the indentation, the sample's indentation state was observed using SPM. The corresponding 3D plots (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea) show a bulge around the indentation mark. The image at exactly the same indentation mark expressed in two dimensions is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb. The white part represents the bulged area, and we calculated the bond strength using this area as the delamination region. We analyzed the acquired SPM images to calculate the delamination area. The threshold value used in this analysis matched the value shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb; the analysis results are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eUsing the peel volume (\u003cem\u003eV\u003c/em\u003e\u0026thinsp;=\u0026thinsp;area \u0026times; film thickness) determined from these analyses, we express the bonding energy (\u003cem\u003eG\u003c/em\u003e\u003csub\u003ec\u003c/sub\u003e) by the following formula\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:\\begin{array}{c}{G}_{\\text{c}}=2\\gamma\\:=\\frac{{E}_{\\text{f}}h{V}_{\\text{o}}^{2}}{4(1-\\nu\\:){V}_{\\text{c}}^{2}}\\:[J/m\u0026sup2;]\\#\\left(1\\right)\\end{array}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eIn the NI test, because a thin film was approached, the Young's modulus \u003cem\u003eE\u003c/em\u003e and Poisson's ratio \u003cem\u003eν\u003c/em\u003e were used as the values for a thermal oxide film.\u003c/p\u003e \u003cp\u003eReference data were acquired using the commonly used DCB test to obtain reliable data. Measurement variations were minimized using a semi-automatic blade insertion machine and a glove box filled with inert gas, which enabled measurements under an anhydrous atmosphere. Details of the measurement environment and conditions are available elsewhere.\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e,\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e,\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e We here present only the formula for the energy release rate, \u003cem\u003eG\u003c/em\u003e, calculated by the DCB test. In the present study, \u003cem\u003eG\u003c/em\u003e\u003csub\u003ec\u003c/sub\u003e is treated as bond strength. The bonding energy \u003cem\u003eG\u003c/em\u003e\u003csub\u003ec\u003c/sub\u003e and \u003cem\u003eγ\u003c/em\u003e [J/m\u0026sup2;] are expressed as follows:\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:{G}_{\\text{c}}=2\\gamma\\:=\\frac{3{t}_{\\text{b}}^{2}E{t}_{\\text{w}}^{3}}{16{L}^{4}}\\left[\\text{J}/{\\text{m}}^{2}\\right]$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003ewhere \u003cem\u003et\u003c/em\u003e\u003csub\u003eb\u003c/sub\u003e [m] is the thickness of the blade, \u003cem\u003eE\u003c/em\u003e [Pa] is the Young's modulus, \u003cem\u003et\u003c/em\u003e\u003csub\u003ew\u003c/sub\u003e [m] is the thickness of the wafer, and \u003cem\u003eL\u003c/em\u003e [m] is the delamination length\u003csup\u003e\u003cspan additionalcitationids=\"CR39 CR40 CR41 CR42 CR43 CR44 CR45 CR46\" citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u003c/sup\u003e. The \u003cem\u003eG\u003c/em\u003e\u003csub\u003ec\u003c/sub\u003e value calculated from the exact same sample with the image in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e is 6.08 J/m\u003csup\u003e2\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eTo verify the compatibility between DCB testing and NI testing, we prepared W2W samples using various plasma activation powers during the bonding process. The plasma exposure time was fixed, and the plasma intensity was set at 30 W or 100 W. Previous studies have reported that increasing the plasma activation intensity enhances the bond strength\u003csup\u003e\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e. The W2W samples were post-bond annealed at 350\u0026deg;C for 1 h to stabilize their bond strength. Scanning acoustic microscopy (SAM) observations confirmed that the W2W samples were void-free before the evaluation of their bond strength, thereby ensuring that the measurements were conducted under ideal conditions.\u003c/p\u003e \u003cp\u003eBond strength evaluations by the DCB test were conducted under two conditions\u0026mdash;an anhydrous atmosphere and ambient atmosphere\u0026mdash;using the same wafer but at different locations. Symbol plots show the DCB test results under an anhydrous atmosphere for each plasma condition in the W2W samples. The bond-strength reduction due to the influence of WSC induced by moisture in the ambient atmosphere is plotted on the same axis.\u003c/p\u003e \u003cp\u003eNI testing is susceptible to variations due to minor external factors. To account for this variation, we conducted more than 30 measurements on each W2W sample within a 15 \u0026micro;m \u0026times; 15 \u0026micro;m measurement area. Box plots show the NI test results.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe bonding energy measured by the DCB test for the W2W sample bonded at a plasma intensity of 30 W was 6.16 J/m\u0026sup2;. The average bonding energy measured by the NI test at the same location was 6.46 J/m\u0026sup2;. The bonding energy obtained by the DCB test fell within the quartile range of the NI test results. A comparison shows a 4.8% difference in bonding energy between the DCB test and the NI test.\u003c/p\u003e \u003cp\u003eThe bonding energy measured by the DCB test for the W2W sample bonded at a plasma intensity of 100 W was 8.2 J/m\u0026sup2;, and the average bonding energy measured by the NI test was 7.92 J/m\u0026sup2;. The bonding energy obtained by the DCB test fell within the quartile range of the NI test results, and a comparison shows a 3.3% difference in bonding energy between the DCB test and the NI test. Therefore, the bond strength measured by the DCB test in the glove box was found to be equivalent and comparable to the results obtained by the NI test. This correlation also indicates that the bonding interface remained protected from exposure to air during the NI process. As a result, the potential influence of WSC on the bond strength was minimized. Therefore, we concluded that the NI test is a reliable and effective method for measuring wafer bond strength. The sharp cube corner tip angle used in this study had an internal angle of 90\u0026deg; and an angle of 68\u0026deg; between the central axis and the plane, oriented perpendicular to the thin film during peeling. No plastic deformation was observed at the delamination tip, suggesting that mode I fracture predominated in this NI test\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e,\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. In the thermal grown SiO\u003csub\u003e2\u003c/sub\u003e film, which had a Poisson's ratio of 0.17, minimal lateral deformation occurred under the indenter pressure during testing, concentrating the force in the vertical direction. As a brittle material, Th-SiO\u003csub\u003e2\u003c/sub\u003e fractures more easily under tensile stress than under shear stress. The pressing of the indenter generates tensile stress inside the film, making interfacial delamination more likely to progress\u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e,\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. Thus, delamination growth in the NI test, a mixed mode of I and II, is largely influenced by mode I. Therefore, comparisons with DCB testing, where delamination growth is dominated by mode I, are highly plausible. One of the origins of the difference in measured bonding strength between the DCB test and the NI test is likely the influence of mode II. However, the DCB test exhibited a coefficient of variation (CV) of less than 2% as a result of the stable measurement environment. By contrast, the NI test was more susceptible to minor external factors, resulting in a CV of approximately 10%. Although measuring the bonding energy at a single point was difficult, statistical estimation from multiple measurement points yielded results compatible with those of the DCB test. Several factors may affect the results of NI testing. Surface roughness of the film can influence measurement variation, especially for extremely thin films. However, the root means square roughness (\u003cem\u003eRq\u003c/em\u003e) for the 30 W and 100 W samples after wet TMAH etching was 0.15 nm and 0.19 nm, respectively, indicating a negligible influence. In addition, the precision of the mechanical control of the equipment is critical because NI testing is a destructive test. The root mean square error of the depth calibration was 5.21\u0026times;10\u003csup\u003e\u0026minus;\u0026thinsp;6\u003c/sup\u003e \u0026micro;m/V\u0026sup2;, with an indentation depth error of less than 0.0035%. The CV value of the indentation depth error in actual measurements was as small as 0.46%, confirming the minimal effect of calibration error. In addition, there are variability factors that are difficult to deal with in NI tests, such as local density variations, nonuniform residual stress and vibration, and differences in measurement processes. Therefore, we measured the bond strength by statistically summarizing these variabilities in the following NI tests.\u003c/p\u003e\n\u003ch3\u003eVerification of the repeatability using wafers subjected to different post-bond annealing conditions\u003c/h3\u003e\n\u003cp\u003eFigure \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e shows the bond strength of W2W after post-bond annealing, as measured using both the DCB and NI tests. The horizontal axis represents the post-bond annealing temperature. As the post-bond annealing temperature increases, the bonding energy increases in both the DCB and NI tests. These results suggest that the post-bond annealing treatment enhances chemical bonding at the interface, thereby improving the bond strength\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFor the W2W samples that were not annealed, the bond strength measured by the DCB test was 3.46 J/m\u0026sup2;, whereas that measured by the NI test was 2.92 J/m\u0026sup2;. However, when post-bond annealing was conducted at 150\u0026deg;C, 250\u0026deg;C, and 350\u0026deg;C, the bond strengths from the two tests were similar. Because the NI test measures local bond strength and the DCB test measures macroscopic bond strength, we speculate that the difference in localized bond strength contributes to the variability in bond strength at low temperatures.\u003c/p\u003e \u003cp\u003eIn the DCB test, the bond strength showed minimal variability from the non-annealed sample to the sample post-bond annealed at 350\u0026deg;C, with an average standard deviation of ~\u0026thinsp;0.16 J/m\u0026sup2;. By contrast, the NI test showed suppressed variability at higher annealing temperatures, likely because weak bonds result in larger peeling areas, with mechanical control errors contributing to differences in the peeling area. In addition, because the NI test is destructive, bulk fracture of the film can occur in samples with low bond strength.\u003c/p\u003e \u003cp\u003eNevertheless, the validity of the NI test was reaffirmed because the bond strength calculated from both tests fell within the measurement variation for all of the post-bond annealing ranges. These results suggest that the NI test is effective for evaluating WiW uniformity and could be useful for quantitatively assessing the effect of post-bond annealing treatments in improving WiW uniformity of bond strength.\u003c/p\u003e\n\u003ch3\u003eWiW uniformity of bond strength evaluated by non-site-specific measurement\u003c/h3\u003e\n\u003cp\u003eThe W2W samples were annealed at 350\u0026deg;C for 1 h to quantitatively evaluate the WiW uniformity of bond strength. Forty-nine points were measured on a 300 mm Si wafer by the NI test. More than 30 measurements were performed at each measurement point, and the average value excluding outliers was used as the bonding energy at each measurement point. Figure\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003ea shows a histogram of the bonding energy measured at 49 points by the NI test.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eOverlaying a normal distribution on this graph reveals that the data roughly follow a normal distribution, indicating that the variation in the bonding energy within the wafer surface is low. Figure\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eb shows a color map plotting the bonding energy at each measurement point. The average bonding energy across the wafer surface was 3.81 J/m\u0026sup2;. Because the mass production range of variation in the semiconductor industry is generally\u0026thinsp;~\u0026thinsp;5% or less, the range of the CV value within 5% was mapped to the same color. When the notch part of the wafer was set to 0 degrees, the bonding energy showed a maximum value of 8.6 J/m\u0026sup2; at 270 degrees and a minimum value of 6.94 J/m\u0026sup2; at 30 degrees. The CV value of this sample was 5.46%.\u003c/p\u003e \u003cp\u003eIn this measurement, the bonded sample was diced into coupon-sized pieces, then ground, wet etched, and mounted onto the measurement stage. Sources of variation may have been present in these multiple sample preparation processes. However, given the measurement error of the NI test, the bonding energy within the wafer surface can be regarded as approximately uniform.\u003c/p\u003e\n\u003ch3\u003eComparison of D2W and W2W bond strength\u003c/h3\u003e\n\u003cp\u003eW2W and D2W samples were bonded under the same plasma activation and cleaning conditions, followed by a post-bond annealing treatment at 150\u0026deg;C, 250\u0026deg;C, or 350\u0026deg;C for 1 h to evaluate the bond strength. A comparison between W2W and D2W samples prepared under the same bonding conditions validated the bond-strength evaluation of the D2W samples. An advanced flip-chip bonder with high alignment accuracy was used to carry out the bonding (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ea). Equipment with mass-production-equivalent technology was used for dicing and grinding to minimize bonding defects\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e,\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. Figure\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eb shows an SAM image of a D2W sample immediately after bonding, where no voids are observed. Figure\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003ec shows a SAM image of the same sample after annealing at 350\u0026deg;C for 1 h. No noticeable increase in voids was visually observed between the two images. The void ratios, including microvoids, were 3% and 4%, respectively, which are within the measurement error range, suggesting that annealing did not lead to an increase in voids.\u003c/p\u003e \u003cp\u003eFigure 9 shows a comparison of the bond strength obtained for W2W and D2W samples. W2W is the result of the DCB test, which has the least variability, whereas D2W is the bond strength calculated from the NI test.\u003c/p\u003e\n\u003cp\u003eAs previously described, the bonding energy of the W2W sample increases linearly with increasing annealing temperature, showing a trend similar as the D2W sample. The CV values for samples post-bond annealed at 150\u0026deg;C, 250\u0026deg;C, and 350\u0026deg;C are 7.81%, 5.05%, and 5.5%, respectively. These values also show that samples with lower bonding energy exhibit greater variation, as observed in the previous section. The D2W sample exhibited a rapid increase in bonding energy between post-bond annealing temperatures of 150\u0026deg;C and 250\u0026deg;C, indicating that a large difference in bond strength occurred between D2W and W2W at an annealing temperature of 150\u0026deg;C.\u003c/p\u003e \u003cp\u003eThe observed differences are likely attributable to variations in the W2W and D2W bonding processes. The procedures up to the DIW cleaning step are the same for both the W2W and D2W processes. However, a difference exists in the queue time (\u003cem\u003eQ\u003c/em\u003e-time) between cleaning and bonding. In the W2W process, bonding was performed immediately after plasma activation and DIW cleaning using a cluster tool. By contrast, the D2W process requires additional \u003cem\u003eQ\u003c/em\u003e-time for sample transport from plasma activation and DIW cleaning. In addition, the sequential (direct placement) bonding step creates a time lag between each die bonding, causing changes in the surface/interface status. Fusion bonding requires sufficient moisture on the bonding film surface to achieve high bond strength. During the \u003cem\u003eQ\u003c/em\u003e-time, moisture at the top surface may evaporate at the interface, potentially reducing the bond strength in the D2W samples.\u003c/p\u003e \u003cp\u003eIn addition, the delay between cleaning and bonding can introduce organic contaminants and residual particles. In low-temperature treatments, silanol groups at the interface may not be sufficiently activated, resulting in a substantial reduction in bond strength. At temperatures between 0\u0026deg;C and 200\u0026deg;C, only hydrogen bonds between silanol groups are active, and a low number of silanol groups due to particle presence can further reduce bond strength. However, at temperatures above 200\u0026deg;C, dehydration condensation reactions are more likely to influence the chemical properties of the interface, leading to similar bond strengths for W2W and D2W samples\u003csup\u003e\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e,\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e. This finding suggests that bond strength in D2W samples can be measured, which may contribute to assessing verification of integration flow for D2W hybrid bonding.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eWiD uniformity\u003c/h2\u003e \u003cp\u003eNI tests at 17 points within the die of the D2W sample annealed at 350\u0026deg;C for 1 h confirmed that the WiD uniformity of bond strength in the post-bond annealing process to similar to that in the W2D process. At each measurement point, more than 30 measurements were conducted and the average value, excluding outliers, was considered the bonding energy at each point. The resulting data were plotted in a color map (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe center of the die exhibited a high bonding energy of 9.57 J/m\u003csup\u003e2\u003c/sup\u003e, whereas the bonding energy decreased toward the outer regions. In particular, the bonding energy at the edge was 6.78 J/m\u003csup\u003e2\u003c/sup\u003e, or approximately 20% lower than that at the center. This trend is consistent along both the \u003cem\u003ex\u003c/em\u003e- and \u003cem\u003ey\u003c/em\u003e-axes of the die, indicating that the variation is not due to measurement discrepancies but rather a substantial WiD variation in bond strength. In addition, with a standard deviation of 1.01 J/m\u003csup\u003e2\u003c/sup\u003e and a CV of 12.3%, the WiD uniformity is low, revealing a substantial lack of consistency.\u003c/p\u003e \u003cp\u003eThe graph also shows that the bonding energy distribution is skewed to the right, suggesting that the pressure during bonding or the heat propagation during annealing may have been applied more on the right side. It is also possible that warping of the wafer or die occurred after the bonding and annealing treatment, which could have contributed to this variation.\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eWiW and WiD bond strength in both W2W and D2W samples were examined by measuring the bond energy through NI testing. The widely used DCB test cannot measure the WiW uniformity of bond strength because blade insertion is only possible at the edge. In addition, there is no clear method for quantitatively measuring bond strength at the D2W interface because the blade cannot be inserted. Therefore, NI testing was performed to measure and confirm WiW uniformity of bond strength. Our test result suggests that the process accuracy is sufficiently high for NI testing to be reliable. In addition, the ability to measure D2W bond strength revealed differences in the bonding energy between D2W and W2W at certain annealing temperature ranges, indicating potential differences in bonding mechanisms. Furthermore, the measurement of WiD uniformity of bond strength is expected to contribute to the maturation of the D2W processes. A challenge in NI testing is the large measurement variations. Future efforts should focus on identifying the causes of these errors and establishing a highly reproducible measurement method. The introduction of NI testing will likely contribute to the advancement of D2W bonding for chiplet integration by providing valuable bond-strength measurements. Thus, this technique could be important for the development of D2W hybrid bonding in a chiplet integration, which is currently an active research area for semiconductors for AI. Standardizing bond-strength measurement methods will enhance process reliability in semiconductor manufacturing. Evaluating WiW and WiD uniformity will enable further optimization of processes and will support the development of advanced semiconductor devices.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eWafer-to-Wafer (W2W) and Die-to-Wafer (D2W) Bonding\u003c/h2\u003e \u003cp\u003eThe samples used in this study were all prepared from 300 mm Si (100) wafers with a thickness of 775 \u0026micro;m. The W2W and D2W samples were prepared from the same batch of Si wafers. The bonding dielectric film was SiO\u003csub\u003e2\u003c/sub\u003e (thermal Si oxide film) with a thickness of 100 nm\u0026thinsp;\u0026plusmn;\u0026thinsp;5%, which was formed on the Si wafer through steam oxidation.\u003c/p\u003e \u003cp\u003eFor W2W, a N\u003csub\u003e2\u003c/sub\u003e plasma treatment was applied to the samples, followed by a water rinse; direct bonding was then carried out under ambient conditions. These processes were conducted using high-volume-production-compatible fully automated cluster bonding equipment to minimize the variation among samples (within each batch) in the W2W process.\u003c/p\u003e \u003cp\u003eFor D2W, the top sample, which serves as the die, was thinned to 200 \u0026micro;m and singulated into 10 mm \u0026times; 10 mm before the bonding sequence. After the pre-assembly processes, N\u003csub\u003e2\u003c/sub\u003e plasma activation and water rinse treatments like those used in W2W were applied. The bottom substrate remained at 300 mm and was subjected to the same plasma and water rinse treatments as the top dies. Bonding was then carried out using a flip-chip method with hybrid-bonding-compatible die bonder equipment.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eBond strength measurement methods\u003c/h2\u003e \u003cp\u003eThe NI tests were performed using a Bruker TS77. The measurements were conducted at room temperature (18\u0026ndash;25\u0026deg;C) and at a relative humidity of 40\u0026thinsp;\u0026plusmn;\u0026thinsp;10% (ambient atmosphere) using a diamond cube corner indenter. A dedicated bulk Si removal process was required to access the NI indenter at the bonding interface. After bonding or post-bond annealing, the W2W samples were diced. The diced coupon samples were ground/polished to remain\u0026thinsp;~\u0026thinsp;100 \u0026micro;m of one side of the bulk Si. Selective wet etching was subsequently performed using 25% tetramethyl ammonium hydroxide (TMAH) at 70\u0026deg;C to expose the bonding interface\u003csup\u003e\u003cspan additionalcitationids=\"CR52\" citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e\u003c/sup\u003e, which allowed the indenter to induce stress, causing delamination at the bonding interface. For the D2W samples, a similar process was applied to expose the bonding layer after sample preparation; however, the coupon grinding step was omitted because the Si bulk had already been removed before bonding.\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the JSPS KAKENHI Grant Number 23K26388 and the JST PRESTO Grant Number JPMJPR22B3, Japan.\u003cem\u003e\u0026nbsp;\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJ.F. and Y.Y. were primarily responsible for the evaluations, measurements, and analyses in this study. Y.K. and D.K. handled sample preparation and process experiment management. M.S. contributed by developing Python-based analysis code for the nanoindentation test. F.I. supervised all aspects of this study.\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003cem\u003e\u0026nbsp;\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data that support the findings of this study are available from the corresponding author upon reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eJ. H. Lau, \u0026ldquo;Recent Advances and Trends in Advanced Packaging,\u0026rdquo; in IEEE Transactions on Components, Packaging and Manufacturing Technology, vol. 12, no. 2, pp. 228\u0026ndash;252, 2022.\u003c/li\u003e\n\u003cli\u003eO\u0026uml;. Vallin, K. Jonsson, U. Lindberg, \u0026ldquo;Adhesion quantification methods for wafer bonding,\u0026rdquo; Materials Science and Engineering, vol.50 pp. 109\u0026ndash;165, 2005.\u003c/li\u003e\n\u003cli\u003eF. Inoue, et al. \u0026ldquo;Inorganic Temporary Direct Bonding for Collective Die to Wafer Hybrid Bonding,\u0026rdquo; 2023 IEEE 73rd Electronic Components and Technology Conference (ECTC), 2023.\u003c/li\u003e\n\u003cli\u003eY. Kagawa, et al. \u0026ldquo;Novel stacked CMOS image sensor with advanced Cu2Cu hybrid bonding,\u0026rdquo; 2016 IEEE International Electron Devices Meeting (IEDM), 2016.\u003c/li\u003e\n\u003cli\u003eC.-T. Ko, et al. \u0026ldquo;Low temperature bonding technology for 3D integration,\u0026rdquo; Microelectronics Reliability, 52, 302\u0026ndash;311, 2012.\u003c/li\u003e\n\u003cli\u003eY. Kagawa, et al. \u0026ldquo;Development of face-to-face and face-to-back ultra-fine pitch Cu-Cu hybrid bonding,\u0026rdquo; 2022 IEEE 72nd Electronic Components and Technology Conference (ECTC), 2022.\u003c/li\u003e\n\u003cli\u003eE. Beyne, et al. \u0026ldquo;Scalable, sub 2\u0026mu;m Pitch, Cu/SiCN to Cu/SiCN Hybrid Wafer-to-Wafer Bonding Technology,\u0026rdquo; 2017 IEEE International Electron Devices Meeting (IEDM), 2017.\u003c/li\u003e\n\u003cli\u003eT. Uhrmann, et al. \u0026ldquo;Heterogeneous Integration by Collective Die-To-Wafer Bonding,\u0026rdquo; 2018 International Wafer Level Packaging Conference (IWLPC), 2018. \u003c/li\u003e\n\u003cli\u003eF. Inoue, et al. \u0026ldquo;Advanced Dicing Technologies for Combination of Wafer to Wafer and Collective Die to Wafer Direct Bonding,\u0026rdquo; 2019 IEEE 69th Electronic Components and Technology Conference (ECTC), 2019. \u003c/li\u003e\n\u003cli\u003eF. Inoue, et al. \u0026ldquo;Protective layer for collective die to wafer hybrid bonding,\u0026rdquo; 2019 International 3D Systems Integration Conference (3DIC), 2019.\u003c/li\u003e\n\u003cli\u003eDebendra Das Sharma \u0026ldquo;Universal Chiplet Express (UCle)\u0026reg;: Building an open chiplet ecosystem\u0026rdquo; 2022.\u003c/li\u003e\n\u003cli\u003eLau, J. H. \u0026ldquo;Recent Advances and Trends in Advanced Packaging,\u0026rdquo; IEEE Transactions on Components, Packaging and Manufacturing Technology, 12 (2), 228-252, 2022.\u003c/li\u003e\n\u003cli\u003eA. Elsherbini, et al. \u0026ldquo;Enabling Next Generation 3D Heterogeneous Integration Architectures on Intel Process,\u0026rdquo; 2022 International Electron Devices Meeting (IEDM), 2022.\u003c/li\u003e\n\u003cli\u003eThomas Uhrmann, et al, \u0026ldquo;Accelerating 3D and heterogeneous integration with high-volume D2W hybrid bonding\u0026rdquo;, Chip Scale Review the Future of Semiconductor Packaging, Volume 27, Number 2, 2023.\u003c/li\u003e\n\u003cli\u003eW. Hafez et al., \u0026quot;Intel Power Via Technology: Backside Power Delivery for High Density and High-Performance Computing,\u0026quot; 2023 IEEE Symposium on VLSI Technology and Circuits, Kyoto, Japan, 2023. \u003c/li\u003e\n\u003cli\u003eM. Liao, D. H. Zhou, P. Wang and M. Chen, \u0026ldquo;Power Systems on Chiplet: Inductor-Linked Multi-Output Switched-Capacitor Multi-Rail Power Delivery on Chiplets,\u0026rdquo; 2023 Fourth International Symposium on 3D Power Electronics Integration and Manufacturing (3D-PEIM), Miami, FL, USA, 2023, pp. 1-7. \u003c/li\u003e\n\u003cli\u003eT. Sherwood et al., \u0026quot;300 nm Pitch W2W HBI for CFET and 3D DRAM Through Module Co-optimization,\u0026quot; 2024 IEEE 74th Electronic Components and Technology Conference (ECTC), Denver, CO, USA, 2024, pp. 1968-1972.\u003c/li\u003e\n\u003cli\u003eX. Ai \u003cem\u003eet al\u003c/em\u003e., \u0026quot;First Demonstration of 4-Layer Stacked Planar Channel-All-Around (P-CAA) IGZO FETs with Cost-Effective Process for High-Density 1T1C 3D DRAM,\u0026quot; \u003cem\u003e2024 IEEE International Electron Devices Meeting (IEDM)\u003c/em\u003e, San Francisco, CA, USA, 2024, pp. 1-4.\u003c/li\u003e\n\u003cli\u003eM. Faure \u003cem\u003eet al\u003c/em\u003e., \u0026quot;Ultra Low Temperature Hybrid Bonding: Morphological and Electrical Characterizations,\u0026quot; \u003cem\u003e2024 IEEE 10th Electronics System-Integration Technology Conference (ESTC)\u003c/em\u003e, Berlin, Germany, 2024, pp. 1-6.\u003c/li\u003e\n\u003cli\u003eT. Sherwood \u003cem\u003eet al\u003c/em\u003e., \u0026quot;300 nm Pitch W2W HBI for CFET and 3D DRAM Through Module Co-optimization,\u0026quot; \u003cem\u003e2024 IEEE 74th Electronic Components and Technology Conference (ECTC)\u003c/em\u003e, Denver, CO, USA, 2024, pp. 1968-1972.\u003c/li\u003e\n\u003cli\u003eM. Oda \u003cem\u003eet al\u003c/em\u003e., \u0026quot;Superior Scalability of Advanced Horizontal Channel Flash for Future Generations of 3D Flash Memory,\u0026quot; \u003cem\u003e2024 IEEE International Electron Devices Meeting (IEDM)\u003c/em\u003e, San Francisco, CA, USA, 2024, pp. 1-4.\u003c/li\u003e\n\u003cli\u003eS. H. Hahn, W. Kim, K. Lim, B. Moon and M. Rhee, \u0026quot;Achieving Sub-nm Copper Recess Controllability for Advanced 3D Integration: An Experimental and Atomic-scale Simulation Study on Wet Atomic Layer Etching Process,\u0026quot; \u003cem\u003e2024 IEEE 74th Electronic Components and Technology Conference (ECTC)\u003c/em\u003e, Denver, CO, USA, 2024, pp. 1322-1328.\u003c/li\u003e\n\u003cli\u003eT. Komljenovic et al., \u0026ldquo;Photonic Integrated Circuits Using Heterogeneous Integration on Silicon,\u0026rdquo; in Proceedings of the IEEE, vol. 106, no. 12, pp. 2246-2257, Dec. 2018\u003c/li\u003e\n\u003cli\u003eX. Luo, Y. Cheng, J. Song, T. -Y. Liow, Q. J. Wang and M. Yu, \u0026ldquo;Wafer-Scale Dies-Transfer Bonding Technology for Hybrid III/V-on-Silicon Photonic Integrated Circuit Application,\u0026rdquo; in IEEE Journal of Selected Topics in Quantum Electronics, vol. 22, no. 6, pp. 443-454, Nov.-Dec. 2016.\u003c/li\u003e\n\u003cli\u003eUhrmann, Thomas, Peter Urban, Markus Wimplinger, Boris Povazay, Julian Bravin, and Bernd Thallner. 2023. \u0026ldquo;Novel IR Laser Debonding for Heterogeneous and 3D Integration.\u0026rdquo; IMAP Source Proceedings 2023 (DPC): 354\u0026ndash;76.\u003c/li\u003e\n\u003cli\u003eD. Zehua, H. Kikuchi, H. Hishinuma, M. Murugesan, T. Tanaka and T. Fukushima, \u0026ldquo;Liquid Surface Tension-Driven Chip Self-Assembly Technology with Cu-Cu Hybrid Bonding for High-Precision and High-Throughput 3D Stacking of DRAM,\u0026rdquo; 2024 IEEE 74th Electronic Components and Technology Conference (ECTC), Denver, CO, USA, 2024, pp. 1335-1341.\u003c/li\u003e\n\u003cli\u003eJunya Fuse, Tomoya, Iwata, Sodai Ebiko, Fumihiro Inoue, \u0026ldquo;Robust Measurement of Bonding Strength for Wafer-to-Wafer 3D Integration\u0026rdquo;, International Conference on Electronics Packaging 2023, Kumamoto, April 20, 20223.\u003c/li\u003e\n\u003cli\u003eTomoya Iwata, Junya Fuse, Yuki Yoshihara, Yusuke Kondo, Marie Sano, Fumihiro Inoue, \u0026ldquo;Water stress corrosion at wafer bonding interface during bond strength evaluation,\u0026rdquo; Materials Science in Semiconductor Processing, Volume 184, 2024,\u003c/li\u003e\n\u003cli\u003eW. C. Oliver and G. M. Pharr \u0026ldquo;An Improved Technique for Determining Hardness and Elastic Modulus Using Load and Displacement Sensing Indentation Experiments,\u0026rdquo; J. Mater. Res., 7\u0026ndash;6, 1564/1583 (1992).\u003c/li\u003e\n\u003cli\u003eMing Yuan He et al., \u0026ldquo;Crack deflection at an interface between dissimilar elastic materials: Role of residual stresses,\u0026rdquo; International Journal of Solids and Structures, Volume 31, Issue 24, 1994.\u003c/li\u003e\n\u003cli\u003eD. B. Marshall and A.G. Evans, \u0026ldquo;Measurement of adherence of residually stressed thin films by indentation. I. Mechanics of interface delamination,\u0026rdquo; Journal of Applied Physics, Volume 56, Issue 10, 1984.\u003c/li\u003e\n\u003cli\u003eA. A. Volinsky et al., \u0026ldquo;Interfacial toughness measurements for thin films on substrates,\u0026rdquo; Acta Materialia, Volume 50, Issue 3, 8, P441-466, 2002.\u003c/li\u003e\n\u003cli\u003eF. Fournel, C. Martin-Cocher, D. Radisson, V. Larrey, E. Beche, C. Morales, P. A. Delean, F. Rieutord, H. Moriceau, Water stress corrosion in bonded structures, ECS J. Solid State Sci. Technol. 4 (2015) 124. https://iopscience.iop.org/article/10 .1149/2.0031505jss. \u003c/li\u003e\n\u003cli\u003eJinju Chen and S J Bull, \u0026ldquo;Approaches to investigate delamination and interfacial toughness in coated systems: an overview,\u0026rdquo; 2011 J. Phys. D: Appl. Phys. 44 034001.\u003c/li\u003e\n\u003cli\u003eYeap KB, Zeng KY, Chi DZ., \u0026ldquo;Determining the interfacial toughness of low-k films on Si substrate by wedge indentation: Further studies,\u0026rdquo; Acta Materialia 2008;56:977.\u003c/li\u003e\n\u003cli\u003eYeap KB, Zeng KY, Jiang HY, Shen L, Chi DZ., \u0026ldquo;Determining interfacial properties of submicron low-k films on Si substrate by using wedge indentation technique,\u0026rdquo; Journal of Applied Physics 2007;101.\u003c/li\u003e\n\u003cli\u003eWeizhou Li, Thomas Siegmund, \u0026ldquo;An analysis of the indentation test to determine the interface toughness in a weakly bonded thin film coating \u0026ndash; substrate system,\u0026rdquo; Acta Materialia, Volume 52, Issue 10, 2004, 2989-2999,\u003c/li\u003e\n\u003cli\u003eK. Onishi, H. Kitagawa, S. Teranishi, A. Uedono, F. Inoue, \u0026ldquo;Temporary Direct Bonding by Low Temperature Deposited SiO2 for Chiplet Applications\u0026rdquo; DOI 10.1021/acsaelm.4c00114 (Accepted in ACS Applied Electronic Materials). \u003c/li\u003e\n\u003cli\u003eW. Maszara, G. Goetz, A. Caviglia, L.B. Mckitterick, Bonding of silicon wafers for silicon-on-insulator, J. Appl. Phys. 64 (1988) 4943\u0026ndash;4950, https://doi.org/ 10.1063/1.96768. \u003c/li\u003e\n\u003cli\u003eK.T. Turner, S.M. Spearing, Accurate characterization of wafer bond toughness with the double cantilever specimen, J. Appl. Phys. 103 (2008) 013514, https:// doi.org/10.1063/1.2828156. \u003c/li\u003e\n\u003cli\u003eA.B de Morais, M.F de Moura, J.P.M. Gon\u0026ccedil;alves, P.P. Camanho, Analysis of crack propagation in double cantilever beam tests of multidirectional laminates, Mech. Mater. 35 (7) (2003) 641\u0026ndash;652, https://doi.org/10.1016/S0167-6636(02)00289-2. \u003c/li\u003e\n\u003cli\u003eK. Takeuchi, T. Suga, Quantification of wafer bond strength under controlled atmospheres, Jpn. J. Appl. Phys. 61 (2022) SF1010. https://iopscience.iop.org/arti cle/10.35848/1347-4065/ac5e49. \u003c/li\u003e\n\u003cli\u003eY. Bertholet, F. Iker, J.P. Raskin, T. Pardoen, Steady-state measurement of wafer bonding cracking resistance, Sensor Actuator Phys. 110 (Issues 1\u0026ndash;3) (2004) 157\u0026ndash;163, https://doi.org/10.1016/j.sna.2003.09.004. \u003c/li\u003e\n\u003cli\u003eF. Fournel, L. Continni, C. Morales, J. Da Fonseca, H. Moriceau, F. Rieutord, A. Barthelemy, I. Radu, Measurement of bonding energy in an anhydrous nitrogen atmosphere and its application to silicon direct bonding technology, J. Appl. Phys. 111 (2012) 104907, https://doi.org/10.1063/1.4716030. \u003c/li\u003e\n\u003cli\u003eP.P. Gillis, J.J. Gilman, Double-cantilever cleavage mode of crack propagation, J. Appl. Phys. 35 (1964) 647\u0026ndash;658, https://doi.org/10.1063/1.1713430. \u003c/li\u003e\n\u003cli\u003eK. Pantzas, F. Fournel, A. Talneau, G. Patriarche, E. Le Bourhis, Measuring the surface bonding energy: a comparison between the classical double-cantilever beam experiment and its nanoscale analog, AIP Adv. 10 (2020) 045006, https:// doi.org/10.1063/1.5143843. \u003c/li\u003e\n\u003cli\u003eM.S. El-Zein, K.L. Reifsnider, Evaluation of GIC of a DCB specimen usin an anisotropic solution, J. Compos. Technol. Res. 10 (4) (1988) 151, https://doi.org/ 10.1520/CTR10278J. \u003c/li\u003e\n\u003cli\u003eRyosuke\u003c/li\u003e\n\u003cli\u003eG\u0026ouml;sele, U.; Tong, \u0026ldquo;SEMICONDUCTOR WAFER BONDING,\u0026rdquo; Q.-Y. ANNUAL REVIEW OF MATERIALS SCIENCE, 01 Aug 1998, Vol. 28, Issue 1, pages 215 \u0026ndash; 241.\u003c/li\u003e\n\u003cli\u003eT. Plach, K. Hingerl, S. Tollabimazraehno, G. Hesser, V. Dragoi, M. Wimplinger \u0026ldquo;Mechanisms for room temperature direct wafer bonding.\u0026rdquo; J. Appl. Phys. 7 March 2013; 113 (9): 094905.\u003c/li\u003e\n\u003cli\u003eK. Biswas, S. Kal, \u0026ldquo;Etch characteristics of KOH, TMAH and dual doped TMAH for bulk micromachining of silicon,\u0026rdquo; Microelectronics Journal, Volume 37, Issue 6, 2006, Pages 519-525,\u003c/li\u003e\n\u003cli\u003ePing-Hei Chen, Hsin-Yah Peng, Chia-Ming Hsieh, Minking K Chyu, \u0026ldquo;The characteristic behavior of TMAH water solution for anisotropic etching on both Silicon substrate and SiO2 layer,\u0026rdquo; Sensors and Actuators A: Physical, Volume 93, Issue 2, 2001, Pages 132-137,\u003c/li\u003e\n\u003cli\u003eJae Sung You, Donghwan Kim, Joo Youl Huh, Ho Joon Park, James Jungho Pak, Choon Sik Kang, \u0026ldquo;Experiments on anisotropic etching of Si in TMAH,\u0026rdquo; Solar Energy Materials and Solar Cells, Volume 66, Issues 1\u0026ndash;4, 2001, Pages 37-44.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-6298479/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6298479/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe demand for 3D integration and advanced packaging technologies is increasing. Direct bonding technologies are gaining popularity as methods to achieve further increases of the input/output densities. Although accurate evaluations of bond strength are essential for reliable wafer-level and die-level integration, the local bond strength is not measurable because of the limitations of conventional measurement methods. Here, we used a nanoindentation method for both wafer-to-wafer and die-to-wafer samples. These measurements provide insights into local bond strength and a wide range of uniformities. The obtained statistic value is equivalent to that obtained by the double-cantilever-beam method in an anhydrous atmosphere, implying that the value obtained by the nanoindentation method is not affected by the measurement environment or the true bond strength. The developed method also enabled the measurement of bond strength in die-to-wafer bonded samples, revealing bond-strength variations within individual dies and identifying potential improvements in the die-level bonding process.\u003c/p\u003e","manuscriptTitle":"Microscopic and Non-Site-Specific Measurement Method for Bond Strength of Direct Bonded Wafers and Dies","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-16 13:04:39","doi":"10.21203/rs.3.rs-6298479/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"070028dd-d6e5-4a14-868a-e20eb697dc29","owner":[],"postedDate":"April 16th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":46355515,"name":"Physical sciences/Engineering/Mechanical engineering"},{"id":46355516,"name":"Physical sciences/Engineering/Electrical and electronic engineering"}],"tags":[],"updatedAt":"2025-04-30T08:51:45+00:00","versionOfRecord":[],"versionCreatedAt":"2025-04-16 13:04:39","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6298479","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6298479","identity":"rs-6298479","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}