pubs.acs.

org/acsaelm Article

The High-Adhesion Die-Cleaning Tape with Dual-Functional Groups

for Die to Wafer (D2W) Hybrid Bonding
Yu Zhang,* Ziqi Lian, Fei Ding, Anqi Zhou, Renxi Jin, Yudong Yang, and Qidong Wang*

Cite This: ACS Appl. Electron. Mater. 2024, 6, 3545−3553 Read Online
Downloaded via NATL INST OF TECH SARDAR VALLABHBHAI on November 7, 2024 at 08:56:39 (UTC).

ACCESS Metrics & More Article Recommendations

See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: With the rapid development of artificial intelligence, high performance

computing will break through Moore’s law and move toward three-dimensional (3D)
package integration of chips. Hybrid bonding has become a potential solution for advanced
packaging, as the interconnect pitch shrinks to less than 1 μm. Die to wafer (D2W) hybrid
bonding consists of chemical mechanical polishing (CMP), blading, plasma activation,
prebonding, and annealing processes. The diced dies that are attached to tape will undergo
several steps such as cleaning and activation; therefore, the tape should present a high
adhesion and stability property. This paper innovatively studies the adhesion and stability of
tapes in a variety of chemicals and theoretically reveals the influence of the composition and
structure of the tape adhesive layer on performance. Two types of tapes (A and B) from
different brands have been tested and their average peeling strength on the dies was
measured; the results present that tape A remains stable no matter in the acidic or alkaline
solution, even with megasonics applied for a long time. The adhesive layer of tape A is
closely connected to the substrate layer, and the surface roughness of 1.1 nm is much smaller than that of tape B. Besides, the XPS
analysis verifies that there are abundant carboxyl and ester groups on the surface of the tape A adhesive layer, which contributes to
the outstanding adhesion of dies on the tape and stability performance in a variety of chemical environments. This research reveals
the reasons for the adhesion and stability performance of the tape, which will provide a theoretical basis for the selection of die
cleaning tape and support the development of the D2W hybrid bonding process and high-density packaging technology.
KEYWORDS: die to wafer, hybrid bonding, cleaning tape, functional groups, adhesion property

1. INTRODUCTION Different from the W2W, D2W displays higher flexibility,

The demand for high-performance computing based on higher yield rate and more application scenarios.21,22 The
artificial intelligence continues to break out in diverse process of D2W hybrid bonding contains wafer chemical
applications.1−6 Increasing the size of transistors is the mechanical polishing (CMP), die blading, plasma activation,
traditional way to improve computing power; however, it is prebonding and annealing.23 The wafer after CMP is cut into
limited by the lithographic size and device miniaturization. The specific-size dies by blade or laser stealth dicing. After
advanced three-dimensional interconnect package is a novel expansion, the dies adhered on the tape with a stainless steel
approach to enhance computational capability.7,8 From the ring are cleaned to remove the particles and contamination in
traditional liquid metal welding technique to the fan-out the chemicals. The dielectric-to-dielectric will be prebonded at
package, the I/O interconnect pitch has greatly shrunk to the room temperature, and this process is very sensitive to the
smallest 36 μm.9−11 Unfortunately, due to the solder volume surface defects. Therefore, the dies on the tape frame should be
and deformation problem, it is difficult to satisfy the advanced cleaned to nearly no residuals.24 Besides, the dies may be
process for a smaller pitch. activated by the wet chemical treatment; the tape should
Hybrid bonding, as a novel package technology, refers to a exhibit not only high adhesion to support the dies but also
direct bonding of dielectric-to-dielectric and Cu-to-Cu excellent corrosion resistance to endure harsh chemical
submicron pad at the same time.12−15 This technology environments.
eliminates traditional copper pillars, increases the interconnect
bandwidth and reduces the interconnect pitch and loss (Figure Received: February 19, 2024
1).16−19 Recently, hybrid bonding has expanded from image Revised: April 18, 2024
sensors to a semiconductor advanced package of NAND flash Accepted: April 23, 2024
memory and DRAM products. Published: May 2, 2024
Depending on the sample type, hybrid bonding can be
classified as die to wafer (D2W) and wafer to wafer (W2W).20

© 2024 American Chemical Society https://doi.org/10.1021/acsaelm.4c00308

3545 ACS Appl. Electron. Mater. 2024, 6, 3545−3553
ACS Applied Electronic Materials pubs.acs.org/acsaelm Article

Figure 1. Performance contrast of copper pillar and hybrid bonding connection.16−19

Figure 2. (a) Process flow diagram of die to wafer (D2W) hybrid bonding; (b) the structure of tape; (c) average peeling strength test for tape on
the die; (d−f) the chemical structure of (d) polyacrylic ester, (e) polyethylene glycol terephthalate, (f) poly(acrylic acid), and (g) poly(propylene
oxide).

The investigation of tape adhesion in the D2W process is of subsequent bonding. Consequently, the dies on the tape need
critical importance due to the high precision and reliability to be cleaned with various chemical solutions and the tape
demanded by D2W technology. It is inevitable to generate should exhibit sufficient adhesive strength to ensure that the
organic matter residues and particles during the dicing process, tiny dies do not shift or detach during the cleaning process.
which will contaminate the die surface and destroy the Besides, a megasonic is usually applied to help further clean the
3546 https://doi.org/10.1021/acsaelm.4c00308
ACS Appl. Electron. Mater. 2024, 6, 3545−3553
ACS Applied Electronic Materials pubs.acs.org/acsaelm Article

Table 1. Etch Phenomenon and Average Peeling Strengths of Tape A and B under Different Chemical Conditions
tape A tape B
chemicals temperature phenomenon peeling strength mN/25 mm phenomenon peeling strength mN/25 mm
control 25 °C / 2856 / 1965
SC3 25 °C no change 2860 no change 1989
60 °C no change 3087 no change 1821
HF 25 °C no change 2851 no change 1973
60 °C no change 3150 no change 1803
ICS7000 25 °C no change 2834 edge separation /
60 °C no change 3025 edge separation /
SC1 25 °C no change 2848 failed /
60 °C failed / failed /
IPA 25 °C no change 2862 no change 1977
MSC-RV 25 °C no change 2832 no change 1936

dies, which requires good adhesion between the die and the The tape B is poly(acrylic acid) (PAA) (Figure 2f) and
tape to withstand the dual effects of mechanical shock and poly(propylene oxide) (PO) (Figure 2g), respectively.
chemical corrosion. At present, there is little research on tape In order to research the performance in the chemical
cleaning in D2W hybrid bonding and no mature solution in corrosion conditions, tape A and tape B were immersed in SC3
the market for cleaning tape. Therefore, the peeling strength (H2SO4:H2O2:H2O), HF(HF:H2O), ICS 7000 (an aqueous
between the die and the tape under different cleaning mixture of organic amines), SC1(NH4OH:H2O2:H2O),
conditions was measured to evaluate the stability and adhesion isopropyl alcohol (IPA), MSC-RV (main component: n-
of the tape. dodecene) at room temperature and high temperature (60
In this research, two-type tapes from different manufacturers °C). Table 1 displays the etch phenomenon of tapes under
were utilized as the supporting substrates of the dies for D2W different chemical conditions. Tape A and tape B show obvious
hybrid bonding. The corrosion resistances of the tapes under morphological differences in the corrosion experiments. Except
acid, alkali, megasonic, and high-temperature conditions were for in the SC1 solution at 60 °C, tape A with dies can remain
investigated, and the peel strengths between the dies and the the original morphology without significant change in the
tapes were measured and summarized. In addition, the analysis chemicals. Tape B and the dies stay well in the HF, SC3, IPA
of structure, composition, and surface state of the tape and the MSC-RV chemicals. However, tape B and the die edge
adhesion layer reveals the key factors affecting the corrosion are partially separated when in the ICS7000 solution, and they
resistance and adhesion performance. An ideal die-cleaning are completely peeled off under the SC1 condition no matter
tape has been selected for D2W hybrid bonding, which will at room temperature and higher temperature. This may be
promote three-dimension packaging and accelerate the because the solution penetrated the tape along the edge of the
development of computing power. die, causing break down of the adhesive layer. It will be
discussed in detail in the following sections. It is worth
2. RESULTS AND DISCUSSION mentioning that both SC1 and the ICS7000 are alkaline
solutions, while the HF and SC3 solutions show acid, which
2.1. Adhesion and Stability Property. For D2W hybrid means that tape B reveals relatively good acid-resistance and
bonding, the cleaned wafer should be diced into small dies of a poor alkali-resistance, but tape A is able to resist both acid and
certain size. These dies are supported by a special tape adhered alkali corrosion.
on the stainless steel ring, followed by cleaning with/without The average peeling strengths of tapes on the dies before
megasonic in several chemical solutions, such as SC1, HF, DI and after chemical corrosion were measured and summarized
water, and so on to remove the particles and contamination on in Table 1 and Figure 3. Tape A shows a higher original
the surface. Therefore, the tape must display excellent peeling strength of 2856 mN/25 mm than tape B of 1965 mN/
corrosion resistance and adhesion properties. In this research, 25 mm at room temperature. When at room temperature, the
the 12-in. wafer was diced as the size of 10 mm × 50 mm, then values of peeling strength slightly fluctuate around the initial
the obtained dies were attached on the two different tapes A data after the chemical treatment, without remarkable
and B with a stainless steel ring, as shown in Figure 2. The tape variation. For tape B in the SC1 and the ICS7000 alkaline
consists of three sections: covering layer, adhesion layer, and solutions, the chemicals destroyed the adhesive layer PAA and
substrate layer. As indicated in Figure 2a, the stainless steel invaded between the tape and the die, resulting in almost no
ring primarily functions to support the periphery of the circular adhesive force. When the temperature rises to 60 °C, the
tape membrane. The dies are adhered to a circular piece of peeling strength of tape A increases from ∼2800 to ∼3100
tape through the inherent adhesive properties of the tape itself. mN/25 mm except for the SC1, while tape B displays a little
The entire assembly, with the dies attached to the tape, is decrease of peeling strengths. It is obvious that tape A is more
secured within a stainless steel ring. This configuration ensures stable than tape B when applied as the cleaning tape of dies.
the structural stability and positioning accuracy required for To further verify the possibility of tape A as the cleaning
subsequent processing steps. After peeling off the covering film, the dies that pasted on tape A were placed and cleaned in
layer, the dies were stuck with the adhesion layer on the DI water, SC1 and MSC-RV under megasonic conditions for
substrate layer. The components of tape A adhesive and different durations. The phenomenon after tape is put into the
substrate layer are polyacrylic ester (Figure 2d) and poly- solution has been recorded and the measure results are shown
ethylene glycol terephthalate (PET) (Figure 2e), respectively. in Table 2. Even under 20 min of megasonic cleaning, tape A
3547 https://doi.org/10.1021/acsaelm.4c00308
ACS Appl. Electron. Mater. 2024, 6, 3545−3553
ACS Applied Electronic Materials pubs.acs.org/acsaelm Article

Figure 4. Average peeling strengths of tape A under different

chemicals with/without ultraphonic.
Figure 3. Peeling strengths of tapes A and B under different chemical
conditions.

still retains excellent adhesion property without separation. All

the peeling strengths between tape A and the dies are focused
on about 2900 mN/25 mm, approaching the original data
(Figure 4). It can be seen that tape A exhibits prominent
stability even though, in the very harsh cleaning environment,
it can be considered as an ideal tape material for die cleaning.
For an ideal tape, it not only needs an excellent adhesion to
dies but also is easy to pick up after UV light exposure. As
shown in Figure 5 and Table 3, tape A exhibits a sharp drop of
peeling strength with almost no stickiness after being
illuminated under UV for 30 s with a power of 800 mJ, Figure 5. Average peeling strengths of tape A and tape B before and
which is very conducive to pick-up and transfer of dies. Above after UV exposure.
all, tape A is an ideal option to be used as the cleaning tape for
dies. Table 3. Average Peeling Strengths of Tape A and Tape B
2.2. Morphology and Structure Characterization. In before and after UV Exposure
order to explore the reason for superior performance for tape peeling strength mN/25 mm
A, scanning electron microscopy (SEM) was characterized on tape before UV after UV
the top view and cross-section of two tapes (Figure 6). Tape A
Tape A 2856 35
presents quite flat surface topography and cross-sectional
Tape B 1965 50
morphology without identifiable protrusions or depressions. In
addition, the adhesive layer polyacrylic ester and the substrate
layer PET are closely combined, which is beneficial to the B is significantly lower than that of tape A, especially in the
improvement of adhesive property. However, there are alkali solution.
numerous small protuberances on the adhesive layer surface The white light interference measurement results further
of tape B, and the relatively rough adhesion surface leads to the prove the difference in surface morphology between two tapes.
not tight fitness between die and the tape (Figure 6b). Besides, Several areas on the tape surface were chosen for measurement
a significant boundary can be distinguished between the and the representative results are shown in Figure 7 and Figure
adhesive layer and substrate layer from the cross-section SEM 8. The roughness values of tape A are 12.7 and 1.1 nm on the
image (Figure 6d). Hence the chemicals are easily permeated area of 3000 μm × 2500 and 300 μm × 250 μm, respectively.
into the space between the interfaces of die/adhesive layer and In contrast, tape B presents the values of 47.4 and 3.0 nm,
adhesive/substrate layer, therefore, the adhesion stress of tape much higher than tape A. The roughness measurement is in

Table 2. Phenomenon and Average Peeling Strengths of Tape A under DI Water, SC1, and MSC-RV Chemicals with
Megasonic Cleaning

no megasonic megasonic 10 min megasonic 20 min

condition phenomenon peeling strength mN/25 mm phenomenon peeling strength mN/25 mm phenomenon peeling strength mN/25 mm
DI water no change 2865 no change 2858 no change 2875
SC1 no change 2848 no change 2850 no change 2798
MSC-RV no change 2832 no change 2846 no change 2842

3548 https://doi.org/10.1021/acsaelm.4c00308
ACS Appl. Electron. Mater. 2024, 6, 3545−3553
ACS Applied Electronic Materials pubs.acs.org/acsaelm Article

respectively. Except for the two characteristic peaks mentioned

above, the C 1s XPS of tape A also presents two other
characteristic peaks at 286.2 eV (C−O) and 288.8 eV (O�
C�OH) binding energy.27,28 What’s more, no N 1s XPS
spectra were detected on tape B, while tape A shows a strong
N 1s spectrum that can be split into two characteristic peaks
that were attributed to N−C at 399.0 eV and N�C�O at
399.9 eV, respectively.29,30
According to the XPS results, the main component of the
tape B adhesive layer is mostly the straight alkane polymer long
chain, with only a small part of the carboxyl group (C�O).
Figure 6. Scanning Electron Microscopy (SEM) top view (a, b) and
On the surface of tape A were detected much more C−O, C�
cross-section images (c, d) of tape A (a, c) and tape B (b, d).
O and COOH functional groups that hung outside the main
alkane chain.31 These chemical groups that contain an oxygen
accord with the SEM morphology results. It can be seen from
element and double bonds present a large dipole moment and
Figure 8b that the surface of tape B is not smooth, forming a
strong polarity, which promotes the intermediate layer of tape
large undulating adhesive layer. This will seriously reduce the
A having a high binding force with both the surface die and the
adhesion of the die and tape B, further having a negative
base layer (Figure 11). In particular, the PET substrate
impact on the cleaning process.
contains π bond of benzene ring and double bond, and the
Considering the differences of peeling strength under
strong intermolecular force between polyacrylic ester and PET
different temperatures, thermal gravimetric analysis (TGA) of
guarantees the substrate and adhesive layer tightly bound
two tapes were characterized from room temperature to 800
together, so tape A displays better stability and heat resistance
°C (Figure 9). They both present three weight change stages.
than tape B. In addition, the N 1s XPS spectra in Figure 10b
The first is observed at 250 °C, which can be attributed to a
and Figure 10d imply that there are not only carboxyl groups
little elimination of water molecules adsorbed on the tape. The
but also ester groups (N�C�O) on the surface of tape A,
second stage between 250 and 500 °C presents a fast weight
while only a few carboxyl groups for tape B. The introduction
loss that is assigned to thermal decomposition of the tape.25
of ester groups will improve the corrosion resistance to alkaline
The third stage higher than 500 °C trends to be stable because
solutions, which explains why tape B is easily destroyed in
of the almost complete decomposition of tape polymer.26 The
alkaline solutions. It can be concluded that the synergistic
weight percentage of tape A between 300 and 450 °C is
working of carboxyl groups and ester groups on the adhesive
apparently larger than that of tape B, meaning tape A displays a
layer ensures the high adhesion property and corrosion
better thermal stability. Because the benzene ring structure of
resistance to a variety of chemical environments.
PET on the tape A substrate is much more stable than the
long-chain alkane of PO for tape B. The adhesion property of
tape A under high temperature is better than that of tape B. 3. CONCLUSIONS
The X-ray photoelectron spectroscopy (XPS) of two tapes Hybrid bonding is regarded as the most promising integrated
were characterized to analyze the surface electronic states and circuit packaging technology beyond Moore’s Law. The Die to
chemical compositions. The C 1s and N 1s high-resolution Wafer (D2W) hybrid bonding process, which includes dicing
XPS spectra of tape A and tape B are remarkably different as wafers and cleaning dies on tape using chemicals to remove
shown in Figure 10. There are two C 1s characteristic peaks of particles, is a critical step in ensuring the integrity and
tape B: a significant one at 284.8 eV and a small one at ∼287.0 performance of semiconductor devices. In this study, two
eV, which are attributed to C−C/C−H and C�O, distinct tapes were utilized for the die-cleaning process,

Figure 7. White light interference topography (a, c) and roughness (b, d) of tape A at 5× (a, b) and 50× (c, d) magnifications

3549 https://doi.org/10.1021/acsaelm.4c00308
ACS Appl. Electron. Mater. 2024, 6, 3545−3553
ACS Applied Electronic Materials pubs.acs.org/acsaelm Article

Figure 8. White light interference topography (a, c) and roughness (b, d) of tape B at 5× (a, b) and 50× (c, d) magnifications.

The tape sample that has not been subjected to any chemical
treatments is marked as “control”. The control samples serve as a
blank experimental group, providing a baseline for comparing the
effects of various chemical exposures on the adhesive properties and
structural integrity of the tapes. After die expansion and
reconstruction, two tapes A and B from different branches were
applied as the die supporting layer on the tape frame for cleaning. In
order to measure the adhesion property, the tapes were cut to a small
size of 90 mm × 10 mm that is longer than the die, which facilitates
the peel strength testing.
4.2. Corrosion Resistance Test. The die-attached tapes were
placed in a series of cleaning chemicals at room temperature and a
high temperature of 60 °C, such as SC3 (H2SO4:H2O2:H2O),
HF(HF:H2O), ICS 7000 (an aqueous mixture of organic amines),
SC1(NH4OH:H2O2:H2O), isopropyl alcohol (IPA), and MSC-RV
Figure 9. Thermal gravimetric analysis (TGA) of tapes A and B from (main component: n-dodecene). The cleaning chemicals were chosen
room temperature to 800 °C. to simulate the actual D2W cleaning process. Followed by the
deionized water cleaning and nitrogen drying, the samples were
placed on the universal testing machine to test the adhesion between
focusing on their chemical resistance and durability under the dies and the tapes. In addition, tape A with dies was immersed in
rigorous conditions. Tape A with a polyacrylic ester adhesive DI water, SC1 and MSC-RV under megasonic cleaning for 0, 10 and
layer and PET substrate layer presents excellent adhesive 20 min. This experiment simulated the cleaning process of the tape in
different solutions during the D2W bonding process. To mimic the
strength and corrosion resistance even under long-time
rotation encountered during the cleaning process, the tape was
megasonic cleaning, which is closely related to its surface immersed in various solutions and stirring was employed as a means
structure and composition. The carboxyl groups and ester to simulate the rotational motion. The peeling strength was tested and
groups on the polyacrylic ester adhesive layer exhibit strong compared with the control tape. Besides, the tapes also were placed
adsorption forces not only for dies but also for tape substrate under UV illumination for 30 s with a power of 800 mJ, then the
and can resist the acidic and alkaline environments. This paper peeling strength was measured by the universal testing machine.
concludes the adhesion and stability of tapes in a variety of 4.3. Peeling Strength Test. The peeling strength test for the tape
chemicals and theoretically reveals the influence of the was conducted on SHIMADZU Corporation AG-X plus equipment.
composition and structure of the tape adhesive layer on The tapes were cut at a size of 90 mm × 10 mm and pasted on the
performance. This research provides theoretical support for the clean dummy die surface. The tape is longer than the die to facilitate
the peeling strength test. The die was horizontally secured on a jig,
selection of die cleaning tape, supports the rapid development
and the tape was gradually lifted in the vertical direction at a constant
of the D2W hybrid bonding process, and further accelerates rate. The displacement-load curves were recorded, and then the
the breakthrough of chip 3D integration. average peeling strength is calculated based on several test curves.
4.4. Characterization. The field-emission scanning electric
4. EXPERIMENTAL SECTION microscopy (SEM, HITACHI 4800) was used to observe the top
4.1. Sample Preparation. 300 mm Si (100) p-type wafers were and cross-sectional morphology of tapes. The white light interference
diced by a blade at the size of 50 mm × 10 mm with a fully automatic microscope was applied to measure the surface roughness. The
dicing saw (Disco DFD6362, Japan) machine. To circumvent the surface chemical states of tapes were checked by X-ray photoelectron
uncontrollable impact on the tape’s adhesiveness resulting from spectrometer (XPS) with an Al Kα radiation source (hυ = 1486.6 eV),
reactions between metal structures on the die and the cleaning and the carbon peak (C 1s) located at 284.6 eV is used to calibrate.
materials, the dies employed in this experiment were all dummy dies. The thermal stability of samples was measured by a HITACHI

3550 https://doi.org/10.1021/acsaelm.4c00308
ACS Appl. Electron. Mater. 2024, 6, 3545−3553
ACS Applied Electronic Materials pubs.acs.org/acsaelm Article

Figure 10. C 1s (a, c) and N 1s (b, d) X-ray Photoelectron Spectroscopy (XPS) survey spectra of tape A (a, b) and tape B (c, d) surfaces.

Figure 11. Die on tape A: the chemical component and structure of the adhesion and substrate layers.

STA200 simultaneous thermal analyzer with a heating rate of 10 °C/ Authors

min from room temperature to 800 °C at N2 atmosphere. The peeling
strength of the tapes on the dies was measured by a universal test Ziqi Lian − Institute of Microelectronics, Chinese Academy of
system (Shimadzu Corporation, AG 10kNxp). Sciences, Beijing 100029, China; School of Integrated
Circuits, University of Chinese Academy of Sciences, Beijing
■ AUTHOR INFORMATION
Corresponding Authors
100049, China
Fei Ding − Institute of Microelectronics, Chinese Academy of
Sciences, Beijing 100029, China
Yu Zhang − Institute of Microelectronics, Chinese Academy of Anqi Zhou − Institute of Microelectronics, Chinese Academy of
Sciences, Beijing 100029, China; orcid.org/0000-0003- Sciences, Beijing 100029, China; School of Integrated
4755-1099; Email: [email protected] Circuits, University of Chinese Academy of Sciences, Beijing
Qidong Wang − Institute of Microelectronics, Chinese 100049, China
Academy of Sciences, Beijing 100029, China; Renxi Jin − Institute of Microelectronics, Chinese Academy of
Email: [email protected] Sciences, Beijing 100029, China
3551 https://doi.org/10.1021/acsaelm.4c00308
ACS Appl. Electron. Mater. 2024, 6, 3545−3553
ACS Applied Electronic Materials pubs.acs.org/acsaelm Article

Yudong Yang − Institute of Microelectronics, Chinese Academy iMAPS All Asia Conference (ICEP-IAAC); April 14−17, 2015; pp 65−
of Sciences, Beijing 100029, China 68.
(15) Ong, J.-J.; Chiu, W.-L.; Lee, O.-H.; Chiang, C.-W.; Chang, H.-
Complete contact information is available at: H.; Wang, C.-H.; Shie, K.-C.; Yang, S.-C.; Tran, D.-P.; Tu, K.-N.;
https://pubs.acs.org/10.1021/acsaelm.4c00308 Chen, C. Low-Temperature Cu/SiO2 Hybrid Bonding with Low
Contact Resistance Using (111)-Oriented Cu Surfaces. Materials
Notes 2022, 15 (5), 1888.
The authors declare no competing financial interest. (16) Yu, D. C. H.; Wang, C. T.; Hsia, H. Foundry Perspectives on

■ ACKNOWLEDGMENTS
The work was supported by the National Natural Science
2.5D/3D Integration and Roadmap. In 2021 IEEE International
Electron Devices Meeting (IEDM); December 11−16, 2021; pp 3.7.1−
3.7.4.
Foundation of China (No. 62304250, No. U21A20504), (17) Liang, S. W.; Wu, G. C. Y.; Yee, K. C.; Wang, C. T.; Cui, J. J.;
Beijing Superstring Academy of Memory Technology (No. Yu, D. C. H. High Performance and Energy Efficient Computing with
Advanced SoIC Scaling. In IEEE 72nd Electronic Components And
SAMT-2022-PM02-16) and China Scholarship Council (No.
Technology Conference (ECTC 2022); IEEE, 2022; pp 1090−1094.
202304910275). (18) DBI Ultra Die-to-Wafer Hybrid Bonding. https://adeia.com/

■ REFERENCES
(1) Wu, Z.; Pan, S.; Chen, F.; Long, G.; Zhang, C.; Yu, P. S. A
dbi-die-to-wafer-hybrid-bonding/.
(19) Appelt, B.; Wang, T.; Lai, Y.; Chung, H. Next Generation
fcCSP Packaging. ICEP2011 Proc. 2011, 426−439.
Comprehensive Survey on Graph Neural Networks. IEEE Trans (20) Gao, G.; Mirkarimi, L.; Workman, T.; Fountain, G.; Theil, J.;
Neural Netw Learn Syst 2021, 32 (1), 4−24. Guevara, G.; Liu, P.; Lee, B.; Mrozek, P.; Huynh, M.; Rudolph, C.;
(2) Choquette, J. NVIDIA Hopper H100 GPU: Scaling Perform- Werner, T.; Hanisch, A. Low Temperature Cu Interconnect with
ance. IEEE MICRO 2023, 43 (3), 9−17. Chip to Wafer Hybrid Bonding. In 2019 IEEE 69th Electronic
(3) Labayen, M.; Medina, L.; Eizaguirre, F.; Flich, J.; Aginako, N. Components and Technology Conference (ECTC); May 28−31, 2019;
HPC Platform for Railway Safety-Critical Functionalities Based on pp 628−635.
Artificial Intelligence. Applied Sciences-Basel 2023, 13 (15), 9017. (21) Bourjot, E.; Bond, A.; Nadi, N.; Enot, T.; Sanchez, L.;
(4) Lee, K. L. K.; Kumar, N. Artificial Intelligence for Scientific Montmeat, P.; Martin, B.; Campo, A.; Fournel, F.; Eid, F.; Swan, J.;
Discovery at High-Performance Computing Scales. Computer 2023, Ieee, Integration And Process Challenges Of Self Assembly Applied
56 (4), 116−122.
To Die-To-Wafer Hybrid Bonding. In 2023 IEEE 73rd Electronic
(5) Perez-Cerrolaza, J. J.; Cazorla, F.; Abella, J. Uncertainty
Components And Technology Conference, ECTC; IEEE, 2023; pp
Management in Dependable and Intelligent Embedded Software.
Computer 2023, 56 (3), 114−117. 1397−1402.
(6) Usman, S.; Mehmood, R.; Katib, I.; Albeshri, A. Data Locality in (22) Bourjot, E.; Castan, C.; Nadi, N.; Bond, A.; Bresson, N.;
High Performance Computing, Big Data, and Converged Systems: An Sanchez, L.; Fournel, F.; Raynaud, N.; Metzger, P.; Cheramy, S.; Soc,
Analysis of the Cutting Edge and a Future System Architecture. I. C., Towards 5μm Interconnection Pitch with Die-to-Wafer Direct
Electronics 2023, 12 (1), 53. Hybrid Bonding. In IEEE 71st Electronic Components and Technology
(7) Park, J.; Kang, S.; Kim, M. E.; Kim, N. J.; Kim, J.; Kim, S.; Kim, Conference (ECTC 2021); IEEE, 2021; pp 470−475.
K. M. Advanced Cu/Polymer Hybrid Bonding System for Fine-Pitch (23) Lujan, A. P. Cost and Yield Analysis of Die-to-Wafer Hybrid
3D Stacking Devices. Adv. Mater. Technol. 2023, 8 (20), 2202134. Bonding. In 2022 International Conference on Electronics Packaging
(8) Fang, H.; Qi, X.; Zhou, S.; Yang, S.; Hang, C.; Tian, Y.; Wang, C. (ICEP); May 11−14, 2022; pp 129−130.
High-Efficient Vacuum Ultraviolet-Ozone Assist-Deposited Polydop- (24) Nagano, F.; Inoue, F.; Phommahaxay, A.; Peng, L.; Chancerel,
amine for Poly(lactic-co-glycolic acid)-Coated Pure Zn toward F.; Naser, H.; Beyer, G.; Uedono, A.; Beyne, E.; De Gendt, S.; Iacovo,
Biodegradable Cardiovascular Stent Applications. ACS Appl. Mater. S. Origin of Voids at the SiO2/SiO2 and SiCN/SiCN Bonding
Interfaces 2022, 14 (2), 3536−3550. Interface Using Positron Annihilation Spectroscopy and Electron Spin
(9) Stream, C.; Emile, K.; Maggie, T. Bump Shape Control on High Resonance. ECS Journal of Solid State Science and Technology 2023, 12
Speed Copper Pillar Plating Process in Lead-Free Wafer Level (3), 033002.
Packaging. In 2009 4th International Microsystems, Packaging, Assembly (25) Al Mahmood, A.; Hossain, R.; Sahajwalla, V. Microrecycling of
and Circuits Technology Conference; October 21−23, 2009; pp 432− the Metal-Polymer-Laminated Packaging Materials via Thermal
435. Disengagement Technology. SN Appl. Sci. 2019, 1, 1106.
(10) Miller, L. F. Controlled Collapse Reflow Chip Joining. IBM J. (26) Meghnani, D.; Gupta, H.; Singh, S.; Srivastava, N.; Mishra, R.;
Res. Dev. 1969, 13 (3), 239−250. Tiwari, R.; Patel, A.; Tiwari, A.; Singh, R. Fabrication and
(11) Yao, M.; Zhao, N.; Wang, T.; Yu, D.; Xiao, Z.; Ma, H. Study of Electrochemical Characterization of Lithium Metal Battery Using
Three-Dimensional Small Chip Stacking Using Low Cost Wafer-Level IL-based Polymer Electrolyte and Ni-rich NCA Cathode. Ionics 2020,
Micro-bump/B-Stage Adhesive Film Hybrid Bonding and Via-Last
26, 4835−4851.
TSVs. J. Electron. Mater. 2018, 47 (12), 7544−7557.
(27) He, J.; Das, C.; Yang, F.; Maibach, J. Crosslinked Poly(acrylic
(12) Brun, X. F.; Burggraf, J.; Ruxandra-Aida, B.; Mühlstätter, C.
acid) Enhances Adhesion and Electrochemical Performance of Si
Low temperature Direct Bonding of SiN and SiO Interfaces for
Packaging Applications. In 2020 IEEE 70th Electronic Components and Anodes in Li-ion Batteries. Electrochim. Acta 2022, 411, 140038.
Technology Conference (ECTC); June 3−30, 2020; pp 182−187. (28) Carton, O.; Ben Salem, D.; Bhatt, S.; Pulpytel, J.; Arefi-
(13) Gao, G.; Mirkarimi, L.; Workman, T.; Guevara, G.; Theil, J.; Khonsari, F. Plasma Polymerization of Acrylic Acid by Atmospheric
Uzoh, C.; Fountain, G.; Lee, B.; Mrozek, P.; Huynh, M.; Katkar, R. Pressure Nitrogen Plasma Jet for Biomedical Applications. Plasma
Development of Low Temperature Direct Bond Interconnect Processes and Polymers 2012, 9 (10), 984−993.
Technology for Die-To-Wafer and Die-To-Die Applications-Stacking, (29) Wu, Y.; Fang, Q.; Yi, X.; Liu, G.; Li, R.-W. Recovery of Gold
Yield Improvement, Reliability Assessment. In 2018 International From Hydrometallurgical Leaching Solution of Electronic Waste via
Wafer Level Packaging Conference (IWLPC); October 23−25, 2018; pp Spontaneous Reduction by Polyaniline. Progress in Natural Science:
1−7. Materials International 2017, 27 (4), 514−519.
(14) He, R.; Fujino, M.; Yamauchi, A.; Suga, T. Combined Surface (30) Safo, I. A.; Dosche, C.; Ö zaslan, M. Effects of Capping Agents
Activated Bonding (SAB) Approach for SiO2 Direct Wafer Bonding in on the Oxygen Reduction Reaction Activity and Shape Stability of Pt
Vacuum. In 2015 International Conference on Electronics Packaging and Nanocubes. ChemPhysChem 2019, 20 (22), 3010−3023.

3552 https://doi.org/10.1021/acsaelm.4c00308
ACS Appl. Electron. Mater. 2024, 6, 3545−3553
ACS Applied Electronic Materials pubs.acs.org/acsaelm Article

(31) Kawaguchi, D.; Nakayama, R.; Koga, H.; Totani, M.; Tanaka,
K. Improvement of Polymer Adhesion by Designing the Interface
Layer. Polymer 2023, 265, 125581.

3553 https://doi.org/10.1021/acsaelm.4c00308
ACS Appl. Electron. Mater. 2024, 6, 3545−3553