Chemical Mechanical Polishing Planarization Research Articles

Measurement of number-weighted particle size distribution for CMP slurry using nanoparticle chip

Chemical mechanical polishing/planarization (CMP) slurry contains mainly abrasive grains (primary particles) with few aggregations (secondary particles). Measuring the particle size distribution (PSD) of CMP slurry is crucial for improving the productivity of the CMP process. For the quality management of CMP slurry, it is necessary to evaluate both the sizes and quantities of both the primary and secondary particles. Conventional PSD analysis methods, except for image analysis, face challenges in identifying primary and secondary particles for PSD measurements. For image analysis, the particles must be transferred from a suspension to a substrate using the conventional sampling method; however, this creates aggregates, resulting in a change in the PSD of the particles on the substrate compared to that in suspension under poly-dispersed conditions. Thus, this study proposed a novel particle sizing method using nanoparticle chip (NPC) to assist in the image analysis of the PSD. The NPC can pick up a single particle in a small volume droplet to avoid aggregation and maintain the poly-dispersed condition of the particles in suspension. The primary and secondary particles can be identified using scanning electron microscope (SEM) or atomic force microscope (AFM), and the PSD can be evaluated by measuring the area and height of the particles. Further, the quantities of both the primary and secondary particles can be counted from the substrate. This study presented a comparison of the diameter measurements using NPC and the conventional method. The results show that the NPC identified the primary and secondary particles and decreased the measurement error of the particle diameters. Consequently, the NPC proves to be superior to the conventional method for use in the PSD analysis of CMP slurry.

Analysis on Pad Surface Roughness of Diamond Conditioning Process for CMP

Chemical Mechanical Polishing/Planarization (CMP) is the key process of the wafer and thin film planarization process for semiconductor manufacturing. In the CMP process, the pad can restore its surface topography and efficiency by diamond conditioner or named as diamond conditioning process. The pad surface roughness is essential to the removal rate in the CMP process because it can represent the contact level of the polishing pad and wafer. This study has developed a simulation program based on precise diamond-grits conditioning routes and numbers of cross points. By studying the OP (overlap cutting points) distribution in a specific area on the pad which is generated by CL (cutting locus), the trend of roughness parameters is discovered. The simulation and experimental model can be effective to observe pores deformation and predict pad surface topography. The results show the pad roughness would become steady over a certain OP number, and the OP can be precisely calculated and compared to the actual pad surface. The final results of the break-in and pad conditioning process can be estimated or further optimized. The other high-density and customized diamond-grits arrangement of conditioner can be designed according to the study in future work.

Effective Particle Analysis on Wafer in the EKF-CMP System

This study aims to develop a three-dimensional electro-osmosis flow (3D-EOF) cell model for effective particle analysis on the wafer at steady-state under electro-kinetic force (EKF) assistance during chemical mechanical polishing/planarization (CMP). A simulation software is used to simulate the abrasive particle motion with three functional modules including the electric current, the laminar flow, and the particle trajectories. Parameter designs of various simulation conditions such as electrode gap spacing, direct current voltages, and polishing pad thickness have been investigated to analyze the motion of silica abrasive nanoparticles due to EOF. Simulation results of the EOF velocity of slurry flow circulation in different conditions have compared with theoretical calculation results. Results have shown that the total number of effective particles intensifies significantly with increasing electrode voltage, but decreases in both cases as raising electrode gap and larger pad thickness. Experimental results of EKF-CMP process can improve 25.03%, 2.52 nm, 1.39% for material removal rate (MRR), surface roughness, non-uniformity, respectively. It can explain that the wafer surface polishing qualification is significantly by motion of effectual abrasive particles. Results of this study can be extended to contribute to improvement and optimization of EKF-CMP process for Copper CMP process used in IC fabrication.

Automated Linescan Analysis for CMP Modeling

Many of the process steps used in semiconductor chip manufacturing require planar (smooth) surfaces on the wafer to ensure correct pattern printing and generation of multilevel interconnections in the chips during manufacturing. Chemical-mechanical polishing/planarization (CMP) is the primary process used to achieve these surface planarity requirements. Modeling of CMP processes allows users to detect and fix large surface planarity variations (hotspots) in the layout prior to manufacturing. Fixing hotspots before tape-out may significantly reduce turnaround time and the cost of manufacturing. Creating an accurate CMP model that takes into account complicated chemical and mechanical polishing mechanisms is challenging. Measured data analysis and extraction of erosion and dishing data from profile linescans from test chips are important steps in CMP model building. Measured linescans are often tilted and noisy, which makes the extraction of erosion and dishing data more difficult. The development and implementation of algorithms used to perform automated linescan analysis may significantly reduce CMP model building time and improve the accuracy of the models. In this work, an automated linescan analysis (ALSA) tool is presented that performs automated linescan delineation, test pattern separation, and automatic extraction of erosion and dishing values from linescan data.

Application of Machine Learning-Based Electrochemical Deposition Models to CMP Modeling

Chemical mechanical polishing/planarization (CMP) is the primary process used for modern integrated circuits (IC) manufacturing. Modeling of the post-CMP surface profile is critical for detecting planarity hotspots prior to manufacturing and avoiding fatal failures of chips. Electrochemical deposition (ECD) is a key process for the void-free filling of interconnection wires and vias in modern chips. Large surface topography variations generated after ECD affect the post-CMP surface profile. In this paper, several machine learning approaches are used to model surface profiles after ECD that are used as input to CMP models. Different combinations of deep neural networks, long-shortterm-memory (LSTM) recurrent networks, convolutional neural networks and XGBoost algorithms are investigated and compared. The model based on the XGBoost library showed superior performance and accuracy

Pressure calculation using obstacle problem for CMP modeling

Chemical mechanical polishing/planarization (CMP) is one of the most important fabrication technologies in the semiconductor industry. CMP is used to achieve planar surfaces and remove excess deposited material from the wafer. During CMP, a chemical “slurry” containing abrasive particles and chemical reagents is deposited on the polishing pad. The polishing pad is then pressed against the rotating wafer. The combined action of the polishing pad and chemical slurry results in material removal and planarization of the wafer surface. Modeling of the CMP process allows detection of potential planarization defects in chips before manufacturing. Accurate computation of pressure distribution across the wafer surface is crucial for accurate modeling of the CMP process. The pressure distribution calculation is usually done using the Hertz contact theory and Chekina model, which involves computations of direct and inverse fast Fourier transforms (FFT and IFFT) and pad displacements updates. In this paper, we adopted an obstacle problem approach for calculating the pressure distribution across the wafer/die surface for CMP modeling. In an obstacle problem, the pad is assumed to be an elastic membrane with fixed boundaries that is displaced by the wafer surface. The goal is to find the equilibrium position for the pad and calculate the pressure distribution over the surface in contact. The main advantage of this approach is that computations of FFT and IFFT and recalculations of the pad displacements can be avoided. This can lead to a more accurate model that is independent of pad displacement update method. The approach is tested on several examples. The results show correct physical behaviour and are in agreement with expectations.

Vapor-phase Surface Cleaning of Electroplated Cu Films Using Anhydrous N2H4

Copper is widely used in the semiconductor industry as interconnects due to its low resistivity, high resistance to electromigration, low temperature coefficient of resistance, and good thermal stability (1). However, the exposed Cu interconnects during via-opening and post chemical mechanical polishing/planarization (CMP) process, are vulnerable to oxidation with water rinse and exposure to air, resulting in reliability degradation (2). Therefore, an additional process to reduce the copper oxide would be required. Copper cleaning can be achieved by either physical Ar sputtering or chemical reduction process (3). Recent demonstration of chemical-based cleaning of Cu interconnects is expected to overcome disadvantages of physical Ar sputtering process, such as chamfering and re-deposition on vias and trenches. A number of studies on vapor-based reduction of copper oxide under ambient pressure conditions and at temperatures below 400 °C using hydrogen, ammonia, carbon monoxide, forming gas, acetic acid, formic acid, and ethanol as reducing agents have been reported (4,5). On the other hand, Hydrazine (N2H4) can be used in the reduction of copper oxide due to its higher reduction capability (6). Inspired by hydrazine’s unique characteristics, we explore the feasibility of vapor-phase reduction of copper oxide using anhydrous N2H4 to achieve an ideal metallic Cu film in an ALD environment. Additionally, a detailed surface analysis and reaction pathway of reduction with N2H4 has not been reported yet due to lack of in-situ experiment. In this work, reduction of Cu samples with a native oxide were evaluated using N2H4 in a rapid thermal ALD system, as shown in Figure 1 (a). Before introducing the samples, the Cu surface was swept with compressed N2 gas, without any prior solvent cleaning, followed by loading into the ALD chamber. The representative time sequence of one cycle of the N2H4 treatment is illustrated in Figure 1 (b). The chamber was pumped down to 0.2 Torr without Ar carrier gas flow. N2H4 was exposed for 5 s with trapping for 120 s, followed by a purging time of 120 s. From ex-situ XPS analysis, the initial sample surface showed contamination with adventitious carbon species, resulting in relatively low intensity in Cu 2p narrow scan (Fig. 2(a)). In addition, the surface contained Cu(OH)x and a CuxO film approximately 2 nm thick, indicating the metallic copper surface had formed CuxO and Cu(OH)x from exposure to air. With N2H4 treatment at 200 oC, a significant amount of copper oxide and hydroxide were reduced to metallic Cu, as observed in a decrease in the O 1s peak. It implies that N2H4 can clean the surface by reducing the oxide to metallic Cu as well as removing the surface contaminants. In addition, in-situ reflection absorption infrared spectroscopy (RAIRS) was employed to elucidate the individual surface chemistry of copper films during the N2H4 exposure. By monitoring the interaction of N2H4 with the surface species, we found both removal of surface contaminants and reduction of CuxO to metallic Cu (Figure 3). The detailed experimental results will be presented. This work is supported by Rasirc Inc. by providing the anhydrous N2H4. R. P. Chaukulkar, N. F. W. Thissen, V. R. Rai, and S. Agarwal, J. Vac. Sci. Technol. A, 32, 01A108 (2014).Y.-L. Cheng, C.-Y. Lee, and Y.-L. Huang, in Noble and Precious Metals-Properties, Nanoscale Effects and Applications, M. Seehar and A. Bristow, Editors, p. 216–250, Intechopen (2018).C. K. Hu et al., Microelectron. Eng., 70, 406–411 (2003).L. F. Pena, J. F. Veyan, M. A. Todd, A. Derecskei-Kovacs, and Y. J. Chabal, ACS Appl. Mater. Interfaces, 10, 38610–38620 (2018).Y. Chang, J. Leu, B.-H. Lin, Y.-L. Wang, and Y.-L. Cheng, Adv. Mater. Sci. Eng., 2013, 1–7 (2013).D. M. Littrell, D. H. Bowers, and B. J. Tatarchuk, J. Chem. Soc. Faraday Trans. 1 Phys. Chem. Condens. Phases, 83, 3271–3282 (1987). Figure 1

Role of the Adhesion Force during Copper Chemical Mechanical Planarization

The adhesion force between slurry particles and the wafer surface has a significant effect on the material removal and ultra-smooth surface formation during the chemical mechanical polishing/planarization (CMP) process. Based on zeta potentials measurements, we calculate the value of adhesion force, which cannot be ignored compared with the Hertz contact force for small sized abrasive particles at low polishing pressures. More over, the impacts of slurry (particle size and pH value) and wafer surface roughness on the adhesion force are evaluated. Theoretical analyses manifest that the adhesion force mainly originates from the van der Waals force and electrical double-layer force, which increases with the increase of the particle size and slurry pH, while decreases with the increase of the wafer surface roughness. The relationships between the adhesion force and scratch depth indicate that scratching would be avoided while enhancing the adhesion force during CMP. These findings would provide some inspirations to understand the CMP technology in depth, and they are also beneficial to engineering super-smooth and super-lubricant surfaces with nano-scaled roughness.

Polystyrene core–silica shell composite particles: effect of mesoporous shell structures on oxide CMP and mechanical stability

Organic/inorganic composite particles with a core–shell structure exhibit potential applications in chemical mechanical polishing/planarization (CMP) for mechanically challenging materials (copper and low-k dielectrics etc.).

Review on formation mechanism of chemical reaction layer during chemical mechanical polishing of monocrystalline SiC and sapphire substrates

Chemical mechanical polishing/planarization (CMP) of wafers is routinely used as the last mechanical process to produce super-smooth or critically planarized surfaces by removing machining-induced surface/subsurface damages and at the same time reduce the surface roughness to the level nanometers or even sub-nanometers without sacrificing the overall flatness. Successful CMP of monocrystalline silicon carbide (SiC) and sapphire substrates—two major substrates besides silicon for fabrication of high-brightness light emission diode (LED) devices, and also two important optical materials for space applications—demands a deeper mechanistic understanding into the CMP material removal process of these two materials. One of crucial challenges is to reveal the mechanism of formation of the chemical reaction layer affected by chemical-mechanical interaction during CMP. This paper critically reviews the current status of relevant progresses, controversies and challenges. First, we reviewed the debate about the thermodynamic behavior of formation of the chemical reaction layer by carefully comparing the original experimental results and arguments by different groups, including numerous studies on the equilibrium of the reactions involved and the stable reaction products. Second, the inconsistencies on the reaction kinetic behavior of formation and removal of the reaction layer, i.e., the temperature dependent materials removal rates in terms of activation energy, were critically compared and re-analyzed by deriving the numerical values of activation energy from the existing literature. Third, the challenge in establishing the model CMP substrate systems was highlightened in order to stress the urgent requirement for a firm foundation without which the results of different labs were difficult to be critically compared. Last, problems of incomplete characterization of the structure and compositions of the reaction layer and substrates as well as the poor cross-lab repeatability of characterization results were discussed with examples. In the perspective part, we propose a few directions for possible breakthroughs in the area of the mechanisms of formation of the chemical reaction layer: (1) establishment of reliable model CMP substrate systems with controlled surface structure by taking into account a few key parameters including surface/ subsurface defect density, misorientation angle (off-cut angle) as well as crystallography planes exposed; (2) determination and control of the reaction rate to facilitate decoupling reaction and diffusion processes with the help of the frictional force microscopy technique. A multidisciplinary strategy for addressing the issues discussed above is thus strongly advocated through combining materials/surface science/chemistry related approaches—careful and detailed characterization using a variety of state-of-the-art analytical techniques, precise control of surface structure and defects for developing model CMP substrate systems, in-depth analyses of surface structure and compositions, chemical thermodynamic and kinetic studies on chemical reaction mechanisms, as well as microscopic frictional/wear tests and surface force measurements. This review will also have important implications for CMP processing of many more different types of materials other than sapphire and SiC, including but not limited to other promising LED substrates such as GaN and ZnO, optical materials for laser inertial confinement fusion applications such as hard-brittle fused silica and soft-brittle KDP crystals, as well as conventional and novel ceramics and crystals for strong lasers such as YAG and sesquioxides (Lu2O3, Y2O3).

Effectiveness Evaluation of Novel Pad Dressing Method by Flexible Fiber Dresser - Tool Life Evaluation of Flexible Fiber Dresser

Chemical mechanical polishing/ planarization (CMP) is a key technology for fabricating high-efficient semiconductor devices, and the CMP characteristics (removal rate and accuracy, etc.) is depended on the various consumables represented by slurry, polishing pad and dresser used in the CMP process. Currently, in the pad, there are many studies that have pointed the evaluation methodology and the correlation between the pad surface asperity and the CMP characteristics. On the other hand, the dresser is one of the most important consumables because the dresser can control the pad surface asperity, and the diamond grains electrodeposited dresser (diamond dresser) is frequently used. One drawback point of the diamond dresser is that the dressing performance declines owing to the deterioration of the diamond grains. Previously, we have developed a novel flexible fiber dresser that would ensure high performance and longer life of tools, and we have reported the fundamental characteristics by the flexible fiber dresser compared with that of the diamond dresser. In this paper, we will show the results of tool life evaluation of the flexible fiber dresser using a contact image analysis method. As a result, the flexible fiber dresser can be continuously used over 35 hours. Furthermore, the result of having examined the stability of the removal rate of a silicon wafer is reported. Therefore, we found that the flexible fiber dresser is one of the most effective dressing methods for the polishing pad.

Barrier Metal Slurry for Low Defect Copper Damascene Chemical Mechanical Polishing

In the semiconductor process, chemical mechanical polishing/planarization (CMP) is inevitable in copper (Cu) interconnect forming. Cu damascene scheme is achieved by help of CMP to form highly ordered metal lines. As the CMP process uses a set of partial polishing steps to complete a damascene structure, slurries of each partial step is designed to have different performances to that of other steps. Among the steps, final partial step, so called, barrier metal (BM) CMP is a critical step because it determines the superior flat finish. In the BM CMP, a structure consists of heterogeneous materials such as Cu, dielectric oxide, and BM should be polished at once. Thus, abrasive suspended in the BM slurry should be optimized to achieve the superior flat finish. In the present study, 3 kinds of BM slurry were utilized. Abrasive of those slurries are made of colloidal silica in the size of 30 nm to 60 nm. The abrasive surfaces, however, are differently modified to have suitable performance for various polishing environments. The modification involves shape of abrasive, chemical additive to increase suspension stability, coating with polymers, and operation pH. Their effects on the CMP performance not only for blanket wafers but also for pattern wafers having 2x nm pitch were observed experimentally. The results indicated that not only polishing performance but also defect occurrence, which is important in mass fabrication, should be considered equally. Finally, strategy for reducing defect occurrence while maintaining Cu damascene performance will be discussed.

Effects of wet treatment conditions and pattern densities on interfacial bonding characteristics of Cu–Cu direct bonds

Effects of both wet chemical treatments and line pattern densities on the interfacial bonding characteristics of Cu–Cu pattern direct bonds were systematically investigated. The silicon oxide (SiO2) on a parallel Cu lines patterned wafer could be removed effectively by using a solution of buffered oxide etch and sulfuric acid (BOE/H2SO4) to improve the bonding quality of Cu–Cu pattern direct bonds. The Cu surface after BOE/H2SO4 wet pretreatment revealed the complete removal of both the residue particles and the surface oxide layer. After BOE/H2SO4 wet pretreatment, the interfacial adhesion energies of 0.06, 0.09, and 0.23 pattern densities were 7.9, 4.8, and 4.1 J/m2, respectively, where the dielectric erosion due to chemical–mechanical polishing (CMP) process with increasing pattern density significantly affected the Cu–Cu pattern direct bonding quality. Therefore, CMP planarization performance is critical for reliable Cu pattern direct bonding.

Chemical roles on Cu-slurry interface during copper chemical mechanical planarization

In order to optimize the existing slurry for low down-pressure chemical mechanical polishing/planarization (CMP), copper CMP was conducted in H2O2 based slurries with benzotriazole (BTA) and glycine at different pH values. The film composition was investigated by the Nano Hardness Tester and XPS tests. Furthermore, the film structure forming on the copper surface at different pH values was investigated by adopting electrochemical impedance spectroscopy (EIS) technology. In the acidic slurry, discontinuous and porous BTA film covering the Cu/Cu2O surface enhanced the mechanical effect during Cu CMP process, resulted in highest CMP removal rate. In neutral slurry, the lowest CMP removal rate and static corrosion rate were resulted from compacted passivation film on the copper surface. In the alkaline slurry, the mechanical effect was limited by the rapid chemical dissolution. The results will benefit optimization of the slurry and operate conditions during low down-pressure CMP process.

Planarization properties of an alkaline slurry without an inhibitor on copper patterned wafer CMP

The chemical mechanical polishing/planarization (CMP) performance of an inhibitor-free alkaline copper slurry is investigated. The results of the Cu dissolution rate (DR) and the polish rate (PR) show that the alkaline slurry without inhibitors has a relatively high copper removal rate and considerable dissolution rate. Although the slurry with inhibitors has a somewhat low DR, the copper removal rate was significantly reduced due to the addition of inhibitors (Benzotriazole, BTA). The results obtained from pattern wafers show that the alkaline slurry without inhibitors has a better planarization efficacy; it can planarize the uneven patterned surface during the excess copper removal. These results indicate that the proposed inhibitor-free copper slurry has a considerable planarization capability for CMP of Cu pattern wafers, it can be applied in the first step of Cu CMP for copper bulk removal.

Planarization optimization of RF-MEMS switches with a gold membrane

In this paper, a new process is analysed to improve the contact area between a metallic bridge and a signal line of a capacitive switch thanks to a better membrane planarization. Two planarization methods are investigated: thermal process with hard-bake steps and chemical mechanical polishing planarization. Moreover, two different sacrificial layers are used: polyimide (PMGI) and photoresist (AZ1529). For each sacrificial layer and planarization method, the influence of planarity and roughness on down-state capacitance is evaluated. Results show that down-state capacitance is more sensitive to flatness than roughness. Moreover PMGI is the best sacrificial layer to improve down-state capacitance mainly for high resonance frequency switches due to capacitance surface reduction.

Advanced ultraviolet cross-link process and materials for global planarization

The use of conventional thermal cross-link materials such as negative resists, antireflective coating, and planarizing layers does not lead to excellent planarization for multilevel interconnects and specially via arrays prior to trench patterning for an advance lithography. The large thicknesses bias between the blanket areas and interconnect areas, and between the blanket areas and via arrays are usually observed. Large thickness bias creates problems during next lithography by narrowing the process latitude. Recently, chemical mechanical polishing (CMP) technology has been proposed to achieve the planarization. However, CMP planarization technique is very sensitive to pattern density, and there is a strong possibility that chemical etching reaction will increase the dielectric constant. The current CMP technique still requires a new investment in the equipment. We report another novel approach for global planarization using UV cross-link material (XUVTM) and the dielectric UV exposure unit in coater equipment (Clean Track). This planar technique provides benefits for reducing the thickness bias observed in the 22- to 65-nm generation lithography and imprint processes. Using this technique, XUVTM TNG076 has achieved global planarization of 10-nm thickness bias in 85-nm diameter via topography when the blanket film thickness was only 110 nm.

Analysis on contact and flow features in CMP process

Contact pressure and flow features of chemical mechanical polishing/planarization (CMP) process were analyzed, taking advantage of the one-dimensional contact model of two layers and considering slurry flows. In this model, deformations of the bulk pad substrate and the asperities were considered. The deformations of the bulk pad substrate and the asperity layer, as well as the contact pressure and fluid pressure, were revealed with numerical methods. Numerical simulation results show a counterintuitive phenomenon: a diverging clearance is formed in the leading region of the wafer and thereby it gives rise to a suction pressure (subambient pressure). A high stress concentration is presented at the wafer edge and thereby over polishing can be introduced. The research provides some theoretical explanations for these two fundamental features of usual CMP processes.

Microelectronically fabricated LiCoO2∕SiO2/polycrystalline-silicon power cells planarized by chemical mechanical polishing

Monolithic integration of different electronic circuit elements onto a single chip is beneficial for the purpose of reducing overall system cost as well as improving performance and reliability. Integrating a power unit onto a silicon chip requires the implementation of a thin-film solid-state battery that is compatible with existing silicon integrated circuit technology in terms of manufacturing methods, materials, and performance. One of the materials in the battery industry, studied by us, is silicon dioxide (SiO2). The high level of process control for thin SiO2 layers which is a result of microelectronic advances, its insulating properties, and its past reported permeability to light ions have motivated us to exploit SiO2 as a solid-state electrolyte in our integrated thin-film battery. This SiO2 electrolyte layer is thermally grown from an amorphous silicon (a-Si) thin layer deposited on doped polycrystalline silicon (polysilicon) which serves as the anode against a thin-film-sputtered LiCoO2 cathode. We have fabricated and characterized ultrathin solid-state thin-film power cells consisting of LiCoO2, SiO2, and polysilicon. Cells containing an ultrathin SiO2 lithium-free electrolyte with a thickness range of 7–40nm and active-area sizes of 5×5, 2×2, 1×1, and 0.5×0.5mm2 were created using established microelectronics processing and expertise. Ultrathin cells require smooth surfaces (nanometer-scale roughness) as opposed to higher roughness encountered in bulk batteries and microbatteries. To ensure a very thin and flat electrolyte, this work demonstrates the implementation of a planarization step by using chemical mechanical polishing (CMP) in the fabrication of the integrated solid-state thin-film lithium-ion battery. Polishing the polysilicon layer reduced its 1×1μm2 root-mean-square roughness from 8.06to0.53nm and led to smoother interfaces and to higher quality of the SiO2 grown on top of it. The cells were charged and discharged using conventional microelectronic electrical testing equipment and exhibited improved performance when prepared with the additional CMP planarization step. Up to 40% of the charge was retrieved from the planarized cells compared to a maximum of 14.5% retrieved from the nonplanarized cells. The open circuit voltage (VOC) of the LiCoO2∕SiO2/polysilicon cell was estimated by comparing the initial charge voltage values obtained with different electrolyte thicknesses and was found to be 2.19±0.02V. The results presented in this work show the importance of interfacial quality in the process of moving to integrated power on a chip.

Chemical mechanical polishing defect reduction via a plasma etch in the 0.15 μm shallow trench isolation process

The resultant surface microscratches due to chemical mechanical polishing (CMP) can cause device failure and reduce production yield in the shallow trench isolation (STI) process. By using a plasma etch process to replace the overpolish step of CMP planarization (CMP+plasma etch), we have reduced 10% of the total defect counts in the 0.15 μm STI process. The CMP microscratches have been effectively smoothened by the additional plasma etch process. The electrical properties of the devices made via the CMP and the CMP+plasma etch processes have been examined and compared. The off-state currents are similar. A higher breakdown voltage is found for the CMP+plasma etch process and can be interpreted by the local electrical-field concentration. Although the n+/p and p+/n junctions fabricated by the CMP+plasma etch processes exhibit slightly larger leakage currents for certain test patterns, due to plasma damage, the products still meet the inline production requirements, and show promising results for yield imp...