Chemical Mechanical Polishing Polishing Research Articles

Corrosion inhibitors in H2O2 system slurry for Ru based barrier layer Cu interconnect chemical mechanical polishing and optimization

Ruthenium (Ru) has the advantages of low resistivity, high thermal stability, and good adhesion and wettability, making it an important solution for breaking through the 14 nm process as a new barrier layer material for multilayer copper interconnect. The purpose of copper (Cu) interconnect Ru-based barrier layer chemical mechanical polishing (CMP) is to obtain a highly flat surface, which not only includes the ideal Ru removal rate and selectivity of the Cu/Ru removal rate, but also inhibition of Cu corrosion. In this article, the corrosion inhibition mechanism of environmentally friendly and highly water-soluble inhibitor 5-methylthio-1 H-tetrazole (MTT) on Cu was studied. The inhibitory effect of MTT on Cu was analyzed through static etching rate testing, electrochemical experiments, and surface characterization. A systematic test was conducted on the planarization performance of the optimized polishing slurry. The corrosion inhibition mechanism of MTT on Cu was revealed through quantum chemical calculations and XPS testing. The experimental results show that the corrosion inhibitor MTT can effectively improve the surface quality of Cu. Quantum chemical calculations and experimental results indicate that MTT can be adsorbed on the surface of Cu through both physical and chemical methods, forming a dense passivation film, thereby inhibiting the corrosion of Cu. The optimized ratio of Ru-based barrier layer CMP polishing slurry was 5 wt% SiO2, 0.15 wt% H2O2, 20 mM ethylenediamine (EDA), 5 mM K4Fe(CN)6, and 2000 ppm MTT, resulted in the Cu/Ru removal rate selectivity of 1:1.18 and the Cu/Ru corrosion potential difference of 3 mV. The average correction values for dishing and erosion on the test pattern wafer of the Ru barrier layer were 355 Å and 280 Å, respectively. The results of this study indicate that MTT is an effective copper corrosion inhibitor in alkaline H2O2-based polishing slurry, providing meaningful references for Ru-based barrier layer copper interconnect CMP.

(Dielectric Science and Technology Division Poster Award, 1st Place) The Influence of Pad Micro-Features on Chemical Mechanical Polishing (CMP) Performance: Insights from Computational Fluid Dynamics (CFD)

With semiconductor chip dimensions shrinking below 7 nanometers, Chemical Mechanical Planarization (CMP) becomes more critical [1], [2]. To meet the increasing demand for precision and uniformity in the fabrication of wafers and chips, enhancements in CMP consumables and operating conditions are necessary. To address these requirements, novel pad designs have been developed, particularly those that do not require conditioning [3]. It is known that the CMP polishing rate depends on pressure, pad wafer relative velocity, and the distribution of pad asperities [4]. As conventional pads provide small and random contact areas, the external pressure applied by the head determines the material removal rate (MRR) [5] Advanced pads with micro-features provide more contact areas and help deliver consistent pressure and slurry distribution during Chemical Mechanical Polishing (CMP) [6], [7]. This study uses computational fluid dynamics (CFD) to simulate the flow between pad and wafer surfaces to determine how pad surface micro-features affect the CMP process. The abrasive residence time, total fluid pressure, and slurry flow velocity are evaluated for four distinct pad micro-feature geometries: square, square-circle, circle, and ellipse. Increasing the number of edges in the pad micro-feature patterns increases pressure drop, leading to more interaction between pad and slurry and enhancement of MRR. The findings of this study confirm the effectiveness of micro-featured pads in optimizing CMP processes and provide valuable insights into the influence of microfeature geometries on critical parameters for improving the semiconductor fabrication process.[1] J. Seo, Journal of Materials Research, 36, 235–257, Jan, 2021.[2] B. Suryadevara, Advances in Chemical Mechanical Planarization (CMP). Woodhead Publishing, 2021.[3] S. Lee and E. Hwang, “Smart pad fabrication for CMP,” LMA Res Rep, vol. 10, pp. 100–104, 2002.[4] G. Ahmadi and X. Xia, J Electrochem Soc, 148, G99, 2001.[5] X. Xia and G. Ahmadi, Particulate Science and Technology, 20, 187–196, 2002.[6] S. Lee, G. Kim, M. Mcgahay, H. Kim, and H. Cho, “Pad Designs-To navigate the fundamentals of CMP.”[7] M. Mcgahay, G. Kim, S. Lee, H. Cho, and H. Kim, “Critical Feature Size of Polishing Pad and Its Performance.”

Trivalent lanthanum and ytterbium doped meso-silica/ceria abrasive systems toward chemical mechanical polishing (CMP) and ultraviolet irradiation-assisted photochemical mechanical polishing (PCMP)

Ceria (CeO2)-based abrasives are widely utilized in ultra-precision grinding and chemical mechanical polishing (CMP) applications over silica materials due to their unique physicochemical properties. Both mechanical and chemical contributions to polishing processes are highly affected by the size, shape, structure, component, defect of CeO2 abrasives. Herein, the composites involved mesoporous silica (mSiO2) cores and La- or Yb-doped CeO2 shells were synthesized and characterized in terms of X-ray diffractometry, scanning electron microscopy, transmission electron microscopy, Raman spectroscopy, scanning transmission electron microscopy–energy dispersive X-ray mapping methods. The polishing effectiveness of the proposed composites toward quartz glass was experimentally evaluated under both CMP and ultraviolet irradiation-assisted photochemical mechanical polishing (PCMP) conditions. The polishing results indicated that the low-modulus mSiO2 cores strongly exerted the cushion effects for the friction and abrasion to substrates. Consequently, the heterostructured mSiO2/La-CeO2 and mSiO2/Yb-CeO2 abrasive systems offered nearly non-damage and ultra-smooth surfaces with angstrom-level roughness compared to conventional rigid abrasives. The enriched Ce3+ and oxygen vacancy defects at La- and Yb- doped CeO2 surfaces were responsible for the improvements of tribochemical and photochemical activities, thus allowing evidently enhanced removal efficiency with the assistance of ultraviolet irradiation. A possible polishing mechanism on the multi-component abrasive systems was also proposed.

Environment-friendly chemical mechanical polishing for copper with atomic surface confirmed by transmission electron microscopy

Atomic and close-to-atomic scale manufacturing is a new technology that can achieve atomic-level manufacturing precision and functional critical dimension. Here we propose an efficient approach for atomic-scale chemical mechanical polishing (CMP) surface of fine copper utilizing optimized CMP slurry and polishing parameters. The optimal polishing slurry is composed of silica, hydrogen peroxide, aspartic acid, sodium phosphate and deionized water, where abrasive particles of silica nanospheres synthesized are uniformly dispersed. Prior to CMP, copper was precisely lapped by ceria slurry. After CMP, the surface roughness is 0.099 nm with an area of 50 × 50 µm2. Atomic-level flatness surface is acquired. To the best of our knowledge, it is the lowest surface roughness with such a relatively large scanning area, compared with those published previously. High resolution transmission electron microscopy (HRTEM) images of polishing surface further prove that atomic-scale surface roughness is achieved. X-ray photoelectron spectra clarify CMP mechanism that complex oxidation, ionization and chelation are sparked on copper surface and the soft layer is formed. The presented efficient approach for atomic-scale polishing will widen the application fields of copper materials and copper-based components.

Contrast Experiments in Dielectrophoresis Polishing (DEPP)/Chemical Mechanical Polishing (CMP) of Sapphire Substrate

The broad applications of sapphire substrates in many fields warrants an urgent demand for a highly efficient and high precision polishing method for the sapphire substrates. The authors proposed a novel sapphire substrate polishing method that is based on the dielectrophoresis (DEP) effect. The principle of dielectrophoresis polishing (DEPP) is described. A non-uniform electric field was added in the polishing area to drive abrasives moving in the direction towards the plate by the DEP force. The amount of abrasives that participates in the polishing action increases as the distribution of polishing slurry on sapphire surface changes, leading towards the improvement of sapphire polishing both in quality and efficiency. Comparative experiments between DEPP and traditional chemical mechanical polishing (CMP) were carried out. It was found that the maximum increase of sapphire MRR for DEPP is 71%, reaching 13 mg/h, and the minimum increase was 9.5%, reaching 4.6 mg/h. The surface roughness of the sapphire substrate decreases faster and more uniform with DEPP. The final surface roughness of the sapphire substrate after DEPP was Ra 0.87 nm and the flatness was 0.3078 waves (RMS value), which is better than 0.6863 waves (RMS value) of sapphire substrate with traditional CMP polishing.

XPS, UV–vis spectroscopy and AFM studies on removal mechanisms of Si-face SiC wafer chemical mechanical polishing (CMP)

Chemical mechanical polishing (CMP) removal mechanisms of on-axis Si-face SiC wafer have been investigated through X-ray photoelectron spectroscopy (XPS), UV–visible (UV–vis) spectroscopy and atomic force microscopy (AFM). XPS results indicate that silicon oxide is formed on Si-face surface polished by the slurry including oxidant H2O2, but not that after immersing in H2O2 solution. UV–vis spectroscopy curves prove that •OH hydroxyl radical could be generated only under CMP polishing by the slurry including H2O2 and abrasive, so as to promote oxidation of Si-face to realize the effective removal; meanwhile, alkali KOH during CMP could induce the production of more radicals to improve the removal. On the other side, ultra-smooth polished surface with atomic step structure morphology and extremely low Ra of about 0.06nm (through AFM) is obtained using the developed slurry with silica nanoparticle abrasive. Through investigating the variations of the atomic step morphology on the surface polished by different slurries, it's reveals that CMP removal mechanism involves a simultaneous process of surface chemical reaction and nanoparticle atomic scale abrasion.

Tribochemical interaction between nanoparticles and surfaces of selective layer during chemical mechanical polishing

Nanoparticles have been widely used in polish slurries such as those in the chemical mechanical polishing (CMP) process. For understanding the mechanisms of CMP, an atomic force microscope (AFM) is used to characterize polished surfaces of selective layers, after a set of polishing experiments. To optimize the CMP polishing process, one needs to get information on the interaction between the nano-abrasive slurry nanoparticles and the surface of selective layer being polished. The slurry used in CMP process of the solid surfaces is slurry with large nanoparticle size colloidal silica sol nano-abrasives. Silica sol nano-abrasives with large nanoparticle are prepared and characterized by transmission electron microscopy, particles colloidal size, and Zeta potential in this paper. The movement of nanoparticles in liquid and the interaction between nanoparticles and solid surfaces coating with selective layer are very important to obtain an atomic alloy smooth surface in the CMP process. We investigate the nanoparticle adhesion and removal processes during CMP and post-CMP cleaning. The mechanical interaction between nanoparticles and the wafer surface was studied using a microcontact wear model. This model considers the nanoparticle effects between the polishing interfaces during load balancing. Experimental results on polishing and cleaning are compared with numerical analysis. This paper suggests that during post-CMP cleaning, a combined effort in chemical and mechanical interaction (tribochemical interactions) would be effective in removal of small nanoparticles during cleaning. For large nanoparticles, more mechanical forces would be more effective. CMP results show that the removal rate has been improved to 367 nm/min and root mean square (RMS) of roughness has been reduced from 4.4 to 0.80 nm. Also, the results show that the silica sol nano-abrasives about 100 nm are of higher stability (Zeta potential is −65 mV) and narrow distribution of nanoparticle size. Consequently, one kind of alkali slurry containing 100-nm silica sol for the solid surfaces CMP coating through selective transfer is studied in this paper.

Wafer back pressure control and optimization in the CMP process

Chemical mechanical polishing (CMP) is the most effective wafer global planarization technology. The CMP polishing head is one of the most important components, and zone back pressure control technology is used to design a new generation of polishing head. The quality of polishing not only depends on slurry, but also depends on the precise control of polishing pressures. During the CMP polishing process, the set pressure of each chamber is usually not the same and the presence of a flexible elastic diaphragm causes coupling effects. Because of the coupling effects, the identification of multi-chambers and pressure controls becomes complicated. To solve the coupling problem, this paper presents a new method of multi-chamber decoupled control, and then system identification and control parameter tuning are carried out based on the method. Finally, experiments of multi-chambers inflated at the same time are performed. The experimental results show that the presented decoupling control method is feasible and correct.

Blade Diamond Disk for Conditioning CMP Polishing Pad

A diamond conditioner or dresser is needed to regenerate the asperity structure of the pad and recover its designed ability in chemical mechanical polishing (CMP) process. In this paper a new design of diamond conditioner is made by shaping a sintered matrix of polycrystalline diamond (PCD) to form teethed blades. These blades are arranged and embedded in epoxy resin to make a designed penetration angle, called the blade diamond disk. The dressing characteristics of pad surface textures are studied by comparison with conventional diamond conditioner. It is found that the height variation of the diamond tip of blade diamond disk is significantly smaller than the conventional diamond disk. The dressing rate of blade diamond disk is lower than that of the conventional diamond disk, and hence the pad life is prolonged. As a result, reduction of the cost CMP is expected. In addition the pad surface roughness Ra of about 3.79μm is less than Ra of about 4.15μm obtained after dressing using a conventional diamond disk.

Development of a pad conditioning simulation module with a diamond dresser for CMP applications

Chemical mechanical polishing (CMP) is a key fabrication route for the modern semiconductor process. The degradation of the polishing pad during the polishing process can significantly influence the polishing performance. Therefore, pads need to be conditioned frequently and an efficient emulator for the dressing process optimization is required to guide the dressing process. In this work, a pad condition simulation module is designed and demonstrated. In addition, a new performance index, called as recover–area ratio (RAR), is defined. Subsequently, a series of parametric studies are performed to investigate the influence of key geometric or processing parameters for a pad conditioning process in terms of optimizing the conditioning performance. Finally, results reveal that it is possible to correlate the pad RAR and Preston’s constant variation and to integrate the pad conditioning emulator with the pre-developed CMP polishing simulation module as a whole utility to further optimize the existing CMP processes.

Measuring Surface Uniformity in Chemical Mechanical Polishing

Chemical mechanical polishing (CMP) is a technique used in semiconductor fabrication for planarizing the top surface of an in-process semiconductor wafer. Especially, Post-CMP thickness variations are known to have a severe impact on the stability of downstream processes and ultimately on device yield. Hence understanding how to quantify and characterize this non-uniformity is significant step towards statistical process control to achieve higher quality and enhanced productivity. The main reason is that the non-uniformed interface between the wafer and the machine-pad adversely affects the polishing performance and ultimate surface uniformity. The purpose of this paper is to suggest a new measure that estimates the uniformity of wafer surface considering the difference of the amount of abrasion between the center and the edge. This new measure which is called the Coefficient of Uniformity is defined as the following ratio: Geometric Mean (GM) / Arithmetic Mean (AM). This metric can be evaluated regionally to quantify the non-uniformity on the wafer surface from the center to the edge. Further simulations show that this new measure is insensitive to shift of the wafer center and sensitive to shift of the wafer edge. This trend indicates that this new measure is a very useful to test the non-uniformity of wafer after CMP polishing.

Dressing Behaviors of PCD Conditioners on CMP Polishing Pads

Diamond pad conditioner or dresser can determine the efficiency of chemical mechanical polishing (CMP) processes and the quality of polished wafers. Conventional diamond pad conditioners are made by adhering discrete diamond grits on a flat substrate. The size distribution of diamond grits coupled with the deformation of the substrate often make the tips of diamond grits lying at different heights. Instead of attaching individual diamond grits to a metal substrate, a revolutionary design of pad conditioners is based on carving a structure out of sintered polycrystalline diamond (PCD) matrix. The PCD dresser is manufactured by wire electro discharge machining to form cutting pyramids of a specific size with a designed shape. The dressing characteristics of pad surface textures are studied by comparison with conventional diamond pad conditioner. Experimental results indicate that the PCD dresser can dress asperities of the pad more uniformly than the conventional diamond dresser due to PCD dresser having identically shaped tip and the same height diamond. In addition the cutting rate of PCD dresser for IC1000 pad not only is reduced by about 30% but also it can dress pad more effectively than conventional diamond dresser.

Novel diamond conditioner dressing characteristics of CMP polishing pad

A diamond pad conditioner or dresser is required to regenerate the asperity structure of a pad used in chemical mechanical polishing (CMP). A conventional diamond dresser (CDD) is made by attaching discrete diamond grits to the flat surface of a substrate. Although the CDD has regularly distributed diamond grits, the tips of diamond grits are not leveled to the same height. Additionally, diamond grits have different shapes and orientations. A newly designed diamond conditioner, called the advanced diamond dresser (ADD), is made by sculpting a sintered polycrystalline diamond substrate to form identically shaped tips with diamond pieces of equal height, rather than by attaching individual diamond grits to a metal substrate. The dressing characteristics of pad surface textures are compared with those of a CDD. Experimental results reveal that the ADD can create asperities of the pad more uniformly than the CDD. Not only the dressing rate of the ADD is reduced by about 30% but also the stable pad surface texture is formed more quickly than the dressing can be performed using the CDD, extending pad life. The polishing rate of the ADD exceeds that of the CDD. The polishing rate varies with the polishing time, which is less for ADD than for CDD.

Research on the molecular scale material removal mechanism in chemical mechanical polishing

This paper investigates a novel molecular scale material removal mechanism in chemical mechanical polishing (CMP) by incorporating the order-of-magnitude calculations, particle adhesion force, defect of wafer, thickness of newly formed oxidized layer, and large deformation of pad/particle not discussed by previous analysis. The theoretical analysis and experimental data show that the indentation depth, scratching depth and polishing surface roughness are on the order of molecular scale or less. Therefore, this novel mechanism has gained the support from wide order-of-magnitude calculations and experimental data. In addition, with the decrease in the particle size, the molecular scale removal mechanism is plausibly one of the most promising removal mechanisms to clarify the CMP polishing process. The results are useful to substantiating the molecular-scale mechanism of the CMP material removal in addition to its underlying theoretical foundation.

Selective Chemical Mechanical Polishing Using Surfactants

Device fabrication using high density, small pattern size shallow trench isolation (STI) processes requires material removal selectivity during chemical mechanical polishing (CMP) steps for optimum product processing and quality control. To improve the selectivity of STI CMP processes, surfactants were applied to selectively polish silica as opposed to silicon nitrides surfaces. A ten-fold increase in selectivity over conventional colloidal silica slurry was achieved by the addition of sodium dodecyl sulfate (SDS) and pH adjustment. Adsorption characteristics of SDS on silica and silicon nitride were measured as a function of slurry pH and concentration of SDS. As indicated by streaming potential measurements and solution depletion adsorption experiments under acidic pH conditions, SDS adsorption on silicon nitride was significantly higher than silica primarily due to the electrostatic interactions. It was concluded that the preferential adsorption of SDS on silicon nitride results in the formation of a material-selective self-assembled passivation (lubrication) layer leading to selective polishing. Effects of different alkyl chain length of surfactants were tested. Various mixed surfactant systems were tested and it is believed that the addition of second surfactants promotes the adsorption on silica diminishing selectivity. The material-targeted boundary layer lubrication concept may be used to develop selective CMP polishing slurries.

ZnO homoepitaxy on the O polar face of hydrothermal and melt-grown substrates by pulsed laser deposition

2 cm diameter hydrothermal ZnO crystals were grown and then made into substrates using both mechanical and chemical-mechanical polishing (CMP). CMP polishing showed superior results with an (0002) Ω scan full width half maximum (FWHM) of 67 arcsec and an root mean square (RMS) roughness of 2 A. In comparison, commercial melt-grown substrates exhibited broader X-ray diffraction (XRD) linewidths with evidence of sub-surface crystal damage due to polishing, including a downward shift of c-lattice parameter. Secondary ion mass spectroscopy revealed strong Li, Fe, Co, Al and Si contamination in the hydrothermal crystals as opposed to the melt-grown substrates, for which glow discharge mass spectroscopy studies had reported high levels of Pb, Fe, Cd and Si. Low temperature photoluminescence (PL) studies indicated that the hydrothermal crystal had high defect and/or impurity concentrations compared with the melt-grown substrate. The dominant bound exciton for the melt-grown substrate was indexed to Al. ZnO films were grown using pulsed laser deposition. The melt-grown substrates gave superior results with XRD (0002) Ω and 2θ/Ω WHM of 124 and 34 arcsec, respectively. Atomic force microscope measurements indicated a low RMS roughness (1.9 nm) as confirmed by fringes in the XRD 2θ/Ω scan. It was suggested that the improvement in XRD response relative to the substrate might be due to “healing” of sub-surface polishing damage due to the elevated Ts used for the growth. Indeed the c-lattice parameter for the homoepitaxial layer on the melt-grown substrate had become that which would be expected for strain-free ZnO. Furthermore, the stability of the PL peak positions relative to bulk ZnO, confirmed that the films appear practically strain free.

Stabilization of gamma alumina slurry for chemical–mechanical polishing of copper

Stabilization of γ-alumina suspension for chemical–mechanical polishing (CMP) of copper was investigated. Citric acid and poly(acrylic acid) (PAA) ( M w = 5000 ) were used as dispersant. The stability of suspension was evaluated from the changes in viscosity, particle size and zeta potential. It appears that metastable γ-alumina mainly due to its high specific surface area and to the presence of aluminol groups on its surface is progressively transformed to bayerite ( β-Al(OH) 3) by hydration procedure. Citric acid molecules were adsorbed onto γ-alumina surface effectively and exhibited the excellent hydration inhibition effect. Although citrate–alumina surface complexes give barrier to the flocculation, the repulsion potential is based mainly on the electrostatic repulsion, thereby steric hindrance caused by the adsorption of these small molecules is very weak. The electrosteric repulsion, which provides more effective dispersion stability than electrostatic repulsion force, can be expected by using polyelectrolyte such as PAA; however, adsorbed layers of PAA onto solid/liquid interface are loosely formed. Therefore, a large amount of PAA was required to inhibit the surface hydration of γ-alumina suspension, thereby the excess addition of PAA decreased the electrosteric repulsion and re-bridging of the dispersant between particles caused an increase in suspension viscosity. Therefore, synergistic effect can be expected in mixed dispersant system of citric acid and PAA, since small citric acid molecules are adsorbed faster than PAA, inhibiting the progress of surface hydration, and then adsorbed PAA layers exhibit the effective electrosteric repulsion interaction between particles with a small amount compared with PAA alone. It was revealed that the γ-alumina slurry dispersed by mixed dispersant exhibited the improved removal rate of Cu layer by CMP polishing test.

Influence of Slurry Chemistry on Frictional Force in Copper Chemical Mechanical Polishing

In situ frictional force measurement technique was developed for chemical mechanical polishing (CMP) process, and the influence of slurry chemistry of glycine/quinaldic on the frictional force was investigated. The results indicate that the CMP polishing rate increases nonlinearly with frictional force, and the coefficient of friction changes at the transitional friction force which is controlled by the concentration of the chelating agents.

Analysis of Frictional Heating of Grooved and Flat CMP Polishing Pads

A detailed model is described for the generation, transport, and exchange of thermal energy in rotary chemical mechanical polishing (CMP) tools. Frictional energy generated due to abrasion of the wafer by the pad and slurry particles is partitioned between the pad and the rotating wafer, with the majority going to the latter. The slurry at the same time provides a major cooling mechanism, drawing heat away while it is entrained under the wafer and then redistributing it over the pad by radial convection. The slurry flow component of the theory includes thermal effects due the presence of concentric grooves. The model accounts for the time and radial dependence of pad temperature measurements performed on flat and concentrically grooved pads at the leading and trailing edges of the wafer, at the pad center and pad margins, and on the wafer carrier. It suggests that the temperature increase on the wafer may be approximately twice the increase measured on the pad. © 2004 The Electrochemical Society. All rights reserved.

Estimation of Fractional Surface Coverage of Particles for Oxide CMP

The surface coverage of particles between pad and wafer has been studied as a function of down load, particle size, and concentration in order to investigate oxide chemical mechanical polishing (CMP) mechanism. In situ friction force measurements have been conducted to investigate the interfacial contact behavior during polishing. In the absence of particles, friction force increased with increasing down load. In the presence of particles, friction force increased with an increase in down load and solids loading. An equation derived from Amonton’s second law was used to determine the fraction of abrasive particles that actually come into contact with the wafer. The fractional surface coverage of abrasive particles was independent of down load and increased with increasing solids loading and decreasing particle size. Determination of the amount of abrasive particles effectively involved in polishing will make clearer the oxide CMP polishing mechanisms improving slurry design leading to further optimization of the CMP process. © 2004 The Electrochemical Society. All rights reserved.