The present invention relates generally to chemical mechanical polishing of substrates.
In the process of fabricating modern semiconductor integrated circuits (IC), it is often necessary to planarize the outer surface of the substrate. For example, planarization may be needed to polish away an outer layer until a predetermined thickness of the outer layer remains or until the top surface of a patterned underlying layer is exposed. For example, in shallow trench isolation (STI), an oxide layer is deposited to fill an aperture and cover a nitride layer. The oxide layer is then polished away to expose the top surface of the nitride layer, leaving the oxide material between the raised pattern of the nitride layer to form an insulating trench on the substrate.
Chemical mechanical polishing (CMP) is one accepted method of planarization. This planarization method typically requires that a substrate be mounted on a carrier head. The exposed surface of the substrate is typically placed against a rotating polishing pad. The polishing pad can have a durable roughened surface. An abrasive polishing slurry is typically supplied to the surface of the polishing pad. The carrier head provides a controllable load on the substrate to push it against the polishing pad while the substrate and polishing pad undergo relative motion.
Abrasive polishing slurries having nanosized abrasive particles can provide improved CMP performance, for example, by reducing the numbers of defects in the polished substrates, e.g., as compared to slurries containing abrasive particles in the submicron size range. In particular, slurries containing abrasive particles that are spherical, and have controlled sizes, and size distribution, can reduce defects in the substrate and yield polished substrates having flat surfaces.
Cerium oxide (ceria) is a material suitable for use as an abrasive polishing slurry for CMP. Ceria particles produced by hydrothermal synthesis can have a more well-defined distribution of particle sizes in the nanometer range, such that a slurry including such ceria particles results in fewer defects in the substrate after polishing.
In one aspect, a slurry for chemical mechanical planarization includes a surfactant, and abrasive particles having an average diameter between 20 and 30 nm and an outer surface of ceria. The abrasive particles are formed using a hydrothermal synthesis process. The abrasive particles are between 0.1 and 3 wt % of the slurry.
In another aspect, a method of manufacturing a slurry for chemical mechanical planarization includes adding a precursor material into a solution, maintaining a pH of the solution at a pH greater than 7, subjecting the solution to a pressure greater than 100 psi and a temperature greater than 100° C. in a reaction vessel, and collecting the abrasive particles, wherein the abrasive particles have diameters of less than 30 nm.
Advantages may include optionally one or more of the following. Defect rates may be reduced. Scaling up the hydrothermal process to obtain ceria particles at full industrial scale quantities may be easy and cost effective. Hydrothermal synthesis may be a facile process for producing both thermodynamically stable and metastable state materials. For example, the reaction product may easily and effectively be controlled when sub or supercritical water is used as a solvent in the reaction. Properties of the solvent (e.g., water), such as its density, may be varied with temperature and pressure, thus enabling the control of the crystal phase, morphology, and particle size of the product. These hydrothermal processes are also relatively low temperature (<250° C.) and high pressure processes (kPa to MPa) that produce oxide materials with controlled morphology. In general, hydrothermal synthesis can be used to synthesize multicomponent materials like ceramics, BST, perovskite oxides like Ca0.8Sr0.2Ti1-xFeO3, yttria and zirconia based oxides with desired stoichiometry, as well as rare-earth and transition metal based oxides.
Hydrothermal synthesis includes techniques of crystallizing substances from high-temperature aqueous solutions at high vapor pressures. One example is the synthesis of single crystals that depends on the solubility of minerals in hot water under high pressure. Such methods can be particularly suitable for the growth of good-quality crystals while maintaining good control over their composition. The crystal growth can be performed in an autoclave, a steel pressure vessel.
The nanoparticles resulting from the process 100 are substantially pure ceria oxide.
However, various nanoparticles having a shell of ceria and a core of a different material can also be produced using a modified synthesis based on the process 100. In general, nanoparticles of another material can be added to the initial mixture of step 102, e.g., added to the water before the cerium nitrite. Then steps 102-110 performed to grow a ceria shell around a core of the other material.
For example, a hydrothermal synthesis process 130 can be used to produce nanoparticles having a silica core and a ceria shell. Silica nanoparticles can be ultra-sonicated in DI water for 20-30 minutes in step 134 before the steps 102-110 are carried out to yield nanoparticles having a silica core and a ceria shell. The silica nanoparticles can be produced in step 132 using the hydrothermal synthesis process 150 illustrated in
In general, core-shell nanoparticles can be selected to offer selectivity tuning in polishing multiple films, e.g., high selectivity of silicon oxide versus silicon nitride.
The hydrothermal synthesis process 150 illustrated in
In addition, various nanoparticles having a shell of formed from silica and a core of a different material can also be produced using a modified synthesis based on the process 150. In general, nanoparticles of another material can be added to the initial mixture of step 152, e.g., added to the water before the tetraethyl orthosilicate. Then steps 152-160 performed to grow a silica shell around a core of the other material. For example, nanaoparticles having an alumina core and a silica shell can be synthesized.
In general, the nanoparticles fabricated by these processes can have a core that is about 30-100 nm diameter, and a shell that is 2-20 nm thick. Table 1 shows the results of various nanoparticles produced in the hydrothermal synthesis of abrasive particles.
Polydispersity, or polydispersity index can be measured by Dynamic Light Scattering (DLS). The polydispersity index is dimensionless and scaled such that values smaller than 0.05 are rarely seen other than with highly monodisperse standards. Values greater than 0.7 indicate that the sample has a very broad size distribution. The morphology and monodispersity of the nanoparticles can be controlled by various parameters such as the temperature and the pressure of the reaction, the reaction time, the pH and concentration of the precursor (e.g., cerium nitrate, and TEOS).
These nanoparticles can be used as the abrasive particles in a slurry of a CMP process. In particular, a slurry having these nanoparticles may be of particular use in an STI process, e.g., for polishing of the oxide layer during STI, due to the resulting low defect rate and good selectivity of oxide versus nitride. The presence of the thin layer of ceria shell in the nanoparticles can reduce the slurry induced defects caused by the abrasive particles in the slurry that participate in the polishing.
The CMP performance of the nanoparticles obtained from hydrothermal synthesis have been characterized. For example, polishing data was obtained from a polished substrate having an outer layer of silicon oxide. For the polishing process, slurry was dispensed at a flow rate of 200 ml/min, while a polishing pressure of 2 psi is applied using an IC1010 pad. The platen and the polishing head were turned at 87 and 79 rpms, respectively.
In one example, a first original internal slurry included 1.25 wt % of polyacrylic acid, and 1 wt % of ceria in 100 ml of the slurry. The polyacrylic acid functions as a surfactant in the slurry to enhance the ability of the ceria nanoparticles to remain in suspension, and to stabilize the slurry. A second original internal slurry included 2.5 wt % of polyacrylic acid, and 2 wt % ceria. These original internal slurries are very stable up to six to seven months.
For actual CMP characterization, the slurry is diluted to have ceria loading of 0.25 wt % or 0.13 wt %, respectively, by the appropriate addition of DI water. For example, by using one part of the first original internal slurry to three parts of DI water, a 0.25 wt % ceria diluted slurry mixture is obtained. In general, diluted slurries can be used to reduce the amount of slurry consumption as ceria is an expensive slurry. The dilutions generally do not affect the material removal rates too much. Without being limited to specific theories, ceria can have agglomeration issues that may lead to larger defects in the polished substrate. The number of ceria particles are reduced in diluted slurries for a particular unit volume of the slurry.
Table 2 summarizes the oxide removal rate (Ox RR) in Angstrom/minute, its non-uniformity within the wafer after polishing, the nitride removal rate (Nitride RR), and its non-uniformity within the wafer after polishing for ceria loading of 0.25 wt %, in both a baseline (commercial) slurry, and a slurry diluted from the first original internal slurry. The oxide removal rate is about 20% lower in the internal slurry, and the nitride removal rate is about 10% lower in the internal slurry.
Table 3 shows the defect count on a TEOS wafer at ceria loading of 0.25 wt %, for the baseline slurry and the slurry diluted from the first original internal slurry. The defect count for the internal slurry is much lower than that from the commercial slurry. More defects were observed in the center of the wafer.
The results from Tables 2 and 3 are expected, as the smaller (particles are in the nanometer range for the internal slurry, instead of the micrometer range, as in the commercial slurry) particle sizes, and the more well controlled size distribution result in a somewhat lower removal rate, but a much lower defect count.
For the diluted slurry having 0.25 wt % ceria, a removal rate of 860 A/min on Thermal Ox, 389 Å/min on TEOS, 72 Å/min on nitride are obtained. The diluted slurry shows 25% lower defect count compared to commercial slurry. For the diluted slurry having 0.13 wt % ceria, a removal rate of 437 Å/min on thermal oxide, 28 Å/min on nitride are obtained in a first sample. A removal rate of 329 Å/min on thermal Ox, 29 Å/min on nitride are obtained in a second sample. The diluted internal slurries show 30-40% lower defect count compared to commercial slurry.
Table 5 summarizes the removal rate (RR) of material at various pressures for different ceria loading in different slurries. The standard deviation (Sdv) of the removal rate, and the non-uniformity (NU) are also provided. The ratio provided in parenthesis after each type of slurry is the ratio of the original (undiluted) slurry to the ratio of deionized water that is used to produce the diluted slurry at each specific ceria loading.
The diluted internal slurry (1:7) shows non-Prestonian behavior at pressure above 2 psi. In other words, the polishing rate does not scale linearly with applied pressure, but is stable despite the pressure increase from 2 psi to 3 psi or 4 psi.
The above described slurries can be used in a variety of polishing systems. Either the polishing pad, or the carrier head, or both can move to provide relative motion between the polishing surface and the substrate. The polishing pad can be a circular (or some other shape) pad secured to the platen, or a continuous or roll-to-roll belt.
In addition, in some implementations, any of the nanoparticles described above can be incorporated into a fixed-abrasive polishing pad rather than a slurry. Such a fixed abrasive polishing pad can include the nanoparticles embedded in a binder material. The binder material can be derived from a precursor which includes an organic polymerizable resin which is cured to form the binder material. Examples of such resins include phenolic resins, urea-formaldehyde resins, melamine formaldehyde resins, acrylated urethanes, acrylated epoxies, ethylenically unsaturated compounds, aminoplast derivatives having at least one pendant acrylate group, isocyanurate derivatives having at least one pendant acrylate group, vinyl ethers, epoxy resins, and combinations thereof. The binder material can be disposed on a backing layer. The backing layer can be a polymeric film, paper, cloth, a metallic film or the like.
The substrate can be, for example, a product substrate (e.g. which includes multiple memory or processor dies), a test substrate, or a gating substrate. The substrate can be at various stages of integrated circuit fabrication. The term substrate can include circular disks and rectangular sheets.
This application claims priority to U.S. Provisional Application Ser. No. 62/072,908, filed on Oct. 30, 2014.
Filing Document | Filing Date | Country | Kind |
---|---|---|---|
PCT/US15/54952 | 10/9/2015 | WO | 00 |
Number | Date | Country | |
---|---|---|---|
62072908 | Oct 2014 | US |