Choi_2019.md

JSS FOCUS ISSUE ON CMP FOR Sub-10 NM TECHNOLOGIES

Effect of Ceria Abrasive Synthesized by Supercritical Hydrothermal Method for Chemical Mechanical Planarization

Jinhak Choi, 1 Cheolmin Shin, 1 Jichul Yang, 2 Seung-Ki Chae, 3 and Taesung Kim 1,4,z

• $^{1}$School of Mechanical Engineering, Sungkyunkwan University, Suwon-si, Gyeonggi-do 16419, Korea
• $^{2}$Head of CMP & Cleaning Innovation Team, Hynix Semiconductor Inc., Icheon-si, Gyeonggi-do 17336, Korea
• $^{3}$Research & Business Foundation, Sungkyunkwan University, Suwon-si, Gyeonggi-do 16419, Korea
• $^{4}$SKKU Advanced Institute of Nano Technology (SAINT), Sungkyunkwan University, Suwon-si, Gyeonggi-do 16419, Korea

Scratches on the wafer surface are one of the significant issues during chemical mechanical planarization process. In order to satisfy wiring width shrinking with time, after polishing process, the presence of small scratches on the wafer is considerable issue. To reduce mechanically induced scratches, small-sized ceria particles are synthesized by a supercritical hydrothermal method. Spherical shaped nano ceria primary abrasive size is about 10 nm and secondary particles are uniformly distributed around 40 nm. The SiO$_{2}$ wafer removal rate is improved through increasing Ce 3 + concentration by metal doping during the synthesis. The detailed principle of the material removal is introduced. The polishing test and its removal rates by the supercritical-hydrothermal-synthesized ceria slurries (SHC) are compared to a commercial ceria slurry. Removal rate of SHC catches up about 85% of the commercial one. The Ce 3 + concentration of the slurry samples is compared, and the wafer surfaces are scanned for morphology characteristics. Scratch and roughness results of SHC are significantly improved against the results of commercial ceria slurry.

© The Author(s) 2019. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/2.0221905jss]

Manuscript submitted February 14, 2019; revised manuscript received March 18, 2019. Published April 12, 2019. This paper is part of the JSS Focus Issue on CMP for Sub-10 nm Technologies.

With the advancement in technology, electronic devices and their semiconductors are becoming smaller and more compact. According to the roadmap for semiconductors, wiring width of semiconductor has reduced less than 20 nm since 2015. 1 Manufacturing processes have to follow up the design rule. Therefore, it is inevitable for integrated nano-scale circuit to require the planarization process. 2 Chemical mechanical planarization (CMP) allows the circuit to be compact and well-integrated. The main objective of CMP process is a polished wafer with high quality of the surface finish. The main consumables, which are pad, slurry, and conditioner, determine the results of a surface quality during the CMP process. 3 Especially, the slurry that contacts both wafer and pad under the high pressure has potentials that can control various specifications such as scratches or material removal rate (MRR) and so on. Thus, the slurry is an important part for these research fields.

Requirements of satisfactory CMP results include less scratch and roughness. They are significant criteria for designing the next level semiconductor. Scratch and bad wafer roughness can cause functional failure on electronic structure and also affect its yield, reliability etc. 4 Commonly, a SiO$_{2}$ wafer is polished by a silica slurry; however, they are prone to leave scratches which is not suitable for next semiconductor generation that requires even a smaller and compact structure. Those defects are usually brought from mechanical forces by the abrasive. In this study, it is proposed that improvement of chemical force in the slurry makes roughness and scratches decrease during the CMP. Thus, a ceria slurry whose chemical force is considerable than other slurries is investigated as an alternative to existing common slurry.

The ceria slurry cannot be analyzed without the chemical effect. Many studies have attempted to explain the chemical mechanism of oxide layer removal. 5 Both silanol groups of the SiO$_{2}$ wafer and the ceria abrasive take part in the chemical reaction and mechanical tearing. 6 Ce 3 + and Ce 4 + are present in the ceria particle. 7 The key element that reacts with the SiO$_{2}$ surface is Ce 3 $^{+}$, which reaches the Si-O structure and weakens a bonding between Si and O. 8 Mechanical friction from the polishing pad tears the wafer film off under the soften condition. Ce 3 + is formed by an oxygen vacancy in the ceria molecules. 7 According to the structural characteristics of ceria particle, some oxygen

$^{z}$E-mail: [email protected]

ions are unoccupied between the molecules. Among them, Ce 3 + is formed around the vacancy where the area is reactive with SiO$_{2}$. 9,10 In this study, the concentration of Ce 3 $^{+}$is increased, which influences the wafer MRR. A conventional ceria slurry is prepared by calcination. Its abrasive particles are formed in different sizes and shapes, even sharp fragments. This mechanical component of abrasive is expected that the existing CMP acquire a high removal rate but leave the coarse surface. 11 In this research, nano sized and spherical ceria abrasive is synthesized to improve scratches and roughness. The manufacturing of 10 nm sized abrasive is conducted by supercritical hydrothermal synthesis. However, the mechanical impact is also reduced and lower MRR is followed. 12 Thus, increasing chemical impact of Ce 3 $^{+}$is studied to improve the issue. Figures of the supercritical-hydrothermalsynthesized ceria slurry (SHC) and a commercial calcined ceria slurry are compared. The MRR of SHC was almost kept up with that of the commercial slurry's result. Forming Ce 3 + on the ceria particle is introduced, and its concentration in the slurry is correlated by UV intensity.

Experimental

The ceria abrasive whose primary particle size is about 10 nm is prepared by supercritical hydrothermal synthesis. In this process, it is necessary to maintain the water pressure above 221 bar and temperature at 374°C. A precursor, aqueous cerium nitride, is dissolved in deionized (DI) water and it decomposes as follows. 13

Ce(NO$_{3}$)$_{3}$ + H$_{2}$ O → Ce(OH)$_{4}$ + HNO$_{3}$

This solution is mixed with the water that is in supercritical condition. The synthesis is performed in a mixer-T as shown in Fig. 1. The hydroxide solution is dehydrated in supercritical condition. The solution is rapidly formed as an oxide which is the ceria nanoparticle.

Ce(OH)$_{4}$ → CeO$_{2}$ + H$_{2}$ O

Additionally, as indicated in Fig. 2, a reactant (metal ion) is added at the mixing point to increase the oxygen vacancy as well as concentration of Ce 3 $^{+}$. This formation mechanism is discussed on the later

Figure 1. Schematic of Supercritical hydrothermal synthesis system.

section. The particles are allowed to pass through the cooling jacket and filtered to remove the unnecessary residual before the output. In order to control the concentration of Ce 3 ± on ceria particle surface, concentration of metal reactant is varied from 0 to 4 wt%. SHC 0-4 (lower to higher concentration of reactant added) and the calcined commercial ceria slurry (KCS-1200MS, KCTech) which particle size is distributed around 150 nm are prepared to compare a polishing experiment.

The ceria particles are observed through a transmission electron microscope (TEM; JEM-3010, JEOL). The secondary particle size distribution is analyzed by a scanning mobility particle sizer (SMPS;

Figure 2. Schematic of metal ion added on hydrothermal synthesis method in a mixer-T.

NANO-5416, GRIMM Aerosol Technik). The polishing experiment is performed by a tablet tribometer (CP-4, Bruker). The polishing time is 120 s under 3 psi fixed head pressure and the 40 × 40 mm thermal oxide wafer coupon is polished with the polyurethane pad (IC1000, Dow Chemical). Ceria abrasive in all the slurry samples is diluted to 10 wt% with deionized water and the slurry is spread by the flow rate of 85ml/min. After planarization, MRR is measured by a reflectometer (ST5000-DLX, K-Mac). 15 different locations on the center of thermal oxide wafer that are divided by 2 mm from each other points are measured. Surface scratch and roughness are investigated through the tapping mode of an atomic force microscope (AFM; Innova, Bruker). Correlating the MRR with the concentration of Ce 3 $^{+}$, a radiation intensity is measured by an ultraviolet visible spectrophotometer (UV-Vis; UV-3600, Shimadzu).

Results and Discussion

The ceria particles were observed to investigate their shape and size. Their TEM images are shown in Fig. 3. Due to the rapid reaction by supercritical hydrothermal synthesis, 10 nm sized particles were possible to be synthesized. Analyzing through the SMPS (Fig. 4)and

Figure 3. TEM images of SHC slurries and commercial calcined ceria slurry.

SHC 3

SHC 4

Commercial ceria slurry

Figure 4. Particle Size distribution of SHC slurries and commercial ceria slurry (3 times repeatability).

TEM results, the primary particle size is 10 nm and secondary particle size is agglomerated as 41 nm. Comparing SHC 0 and other SHC slurries, adding metal ion during the synthesis hardly effects on particle configurations such as size or distribution but the surface chemical formation. The SHC abrasives are uniform and spherical which illustrates a positive result on the wafer morphology. On the other hand, the calcined commercial ceria slurry particles are irregular and sharp.

Due to a milling process in the synthesis of abrasives, their particle sizes vary from 50 to 200 nm or even larger. These factors reach the insufficient result after CMP process.

After the evaluation of the SHC particles' figurations, the CMP experiment was performed with SHC 0-4 slurries. The results were compared to the polishing result of the commercial ceria slurry. The measured data is shown in Fig. 5. For the SHC slurry without any

Figure 5. Material removal rate results of SHC and commercial ceria slurries.

Figure 6. Ultra violet absorbance comparison of SHC samples.

reactant during the synthesis (SHC 0), the MRR was 2132 Å/min. The highest MRR was found to be 4344 Å/min, which was obtained for SHC 3. The MRR of SHC slurries increased in proportion to the reactant concentration, but there seems to be a critical point around SHC 3. This indicates that a certain concentration of the reactant in the synthesis improves adhesion between the ceria particles and the SiO$_{2}$ wafer. Planarization result of SHC was almost reached to the result of commercial ceria slurry by 85%. It is possible that smaller particle slurry can reach comparable rank with larger particle slurry by chemical modification.

Surface chemistry has a significant impact in this field. Thus, ultraviolet (UV) absorption spectroscopy was carried out for SHC slurries. The absorbance peaks were measured to correlate with CMP results. The absorbance peak of Ce 3 + is located at around 295 nm. 14 As seen from Fig. 6, SHC 3 has the highest Ce 3 + absorbance (it also achieved the maximum MRR). The lowest peak was found for SHC 0, which was synthesized without a reactant. It can be concluded that MRR are possibly correlated to the concentration of Ce 3 $^{+}$. All the ceria absorbance results correspond to each MRR.

Table I. Roughness results of each SiO$_{2}$ wafer after CMP.

Samples	Roughness (RMS)
SHC 0	2.35 ˚ A
SHC 1	4.32 ˚ A
SHC 2	3.79 ˚ A
SHC 3	4.11 ˚ A
SHC 4	5.40 ˚ A
Commercial ceria slurry	21.81 ˚ A

Fig. 7 explains how Ce 3 $^{+}$is formed by the substitution of the metal ion (denoted by M 3 $^{+}$). During the supercritical hydrothermal synthesis, the reactant is added to modify the ceria particle surface. The metal ion is used as the reactant during the synthesis. During the reaction in the mixer-T, Ce 4 + is substituted by M 3 + on the ceria surface. The number of bridged structures is reduced from 4 + to 3 + . It can be interpreted that an available oxygen link is reduced. Therefore, oxygen is prone to be deoxidized from the surface structure. At last, the oxygen vacancy is developed on the ceria surface with the chemical formation of Ce$_{1-x}$M$_{x}$O$_{2}$. 15,16 It is assumed that if more than 3 wt% reactant was added in the synthesis, it does not react more with the ceria surface. Hence, the critical point of interaction between the ceria ions and metal ions exists.

After polishing the SiO$_{2}$ wafers, their surface morphology was observed by AFM. The polishing results of SHC slurries show smooth surfaces (Fig. 8). On the other hand, many scratches remained on the wafer after polishing with the commercial ceria slurry. Almost scratch free morphologies were obtained in the micron scale for the SHC cases. These results illustrate that the scratch issue in SHC slurries is significantly reduced compared to that in the commercial slurry. The roughness of each wafer surface is evaluated after CMP. Table I presents their RMS roughness. It is well known that the roughness of the wafer polished by the commercial ceria slurry ranges from 10 Å to 50 Å. Here, the result of commercial ceria's planarization was 21 Å. On the other hand, the RMS roughness of the surface was improved to 2-5 Å for the SHC results.

Conclusions

The supercritical hydrothermal synthesis of ceria and the principle of increase in Ce 3 + concentration were investigated in this study. To

Figure 7. Oxygen vacancy and Ce3 + formation by substitution of the metal ion.

Figure 8. AFM images of polishing results by SHC and commercial ceria.

improve the CMP process for the SiO$_{2}$ wafer, ceria abrasive nanoparticles were synthesized by these methods. Uniform and spherical abrasive particles were observed. The polishing tests were performed with the SHC slurries and the results were compared to those of the commercial ceria slurry. A CMP experiment gave an account to determine the improvement of the MRR outcome. The MRR of SHC was caught up to 85% by the MRR of the commercial ceria slurry. The UV absorbance for each SHC sample agreed well with the MRR results. After the planarization experiments, the SiO$_{2}$ wafer surfaces were scanned by AFM. Their roughness and scratches were considerably improved than those of the commercial ceria slurry.

Acknowledgments

This work was supported by the MOTIE (Ministry of Trade, Industry & Energy (project number 10080545)) and KSRC (Korea Semiconductor Research Consortium) support program for the development of the future semiconductor device.

ORCID

Taesung Kim https://orcid.org/0000-0001-6280-7668

References

• 1. B. Hoefflinger, in Chips 2020 , p. 161, Springer (2011).
• 2. T. Doi, I. D. Marinescu, and S. Kurokawa, Advances in CMP/polishing Technologies for the Manufacture of Electronic Devices , William Andrew (2011).
• 3. N. Saka, T. Eusner, and J.-H. Chun, CIRP Annals-Manufacturing Technology , 57 , 341 (2008).
• 4. Y. Yamada, N. Konishi, J. Noguchi, and T. Jimbo, Journal of The Electrochemical Society , 155 , H485 (2008).
• 5. P. V. Dandu, B. Peethala, and S. Babu, Journal of The Electrochemical Society , 157 , H869 (2010).
• 6. M. Tomozawa, Solid State Technology , 40 , 169 (1997).
• 7. A. Migani, G. N. Vayssilov, S. T. Bromley, F. Illas, and K. M. Neyman, Chemical Communications , 46 , 5936 (2010).
• 8. T. Hoshino, Y. Kurata, Y. Terasaki, and K. Susa, Journal of Non-Crystalline Solids , 283 , 129 (2001).
• 9. C. T. Campbell and C. H. Peden, Science , 309 , 713 (2005).
• 10. H. Doi, M. Suzuki, and K. Kinuta, in Planarization/CMP Technology (ICPT), 2014 International Conference on , p. 194 (2014).
• 11. M.-H. Oh, J.-S. Nho, S.-B. Cho, J.-S. Lee, and R. K. Singh, Powder technology , 206 , 239 (2011).
• 12. D.-S. Lim, J. Ahn, H. Park, and J. Shin, Surface and Coatings Technology , 200 , 1751 (2005).
• 13. M. Hirano and E. Kato, Journal of the American Ceramic Society , 79 , 777 (1996).
• 14. A. Efimov, A. Ignat'ev, N. Nikonorov, and E. Postnikov, Optics and Spectroscopy , 111 , 426 (2011).
• 15. F. Pérez-Alonso, M. López Granados, M. Ojeda, P. Terreros, S. Rojas, T. Herranz, J. Fierro, M. Gracia, and J. Gancedo, Chemistry of Materials , 17 , 2329 (2005).
• 16. Y. Hakuta, S. Onai, H. Terayama, T. Adschiri, and K. Arai, Journal of materials science letters , 17 , 1211 (1998).