Chemical Mechanical Polishing (CMP) is a critical planarization process in semiconductor manufacturing, enabling the production of highly uniform, defect-free wafer surfaces required for advanced electronic devices. The performance of a CMP process is strongly influenced by slurry properties, particularly the size, concentration, and stability of abrasive particles suspended within the formulation. Even subtle changes in particle behaviour can affect material removal rates, surface quality, process consistency, and overall manufacturing yield.
Colloidal silica is one of the most widely used abrasive materials in CMP slurries for silicon wafer polishing. These nanoscale silica particles are typically dispersed in water and stabilized through surface charge interactions. However, colloidal stability is highly sensitive to the chemical environment. Changes in pH, ionic strength, or processing conditions can alter particle interactions and promote aggregation.
Aggregation is a significant concern because larger particle clusters can introduce scratches and defects on wafer surfaces. Coarse particles may also contribute to variations in polishing performance and material removal rates. As semiconductor devices continue to scale and process tolerances become increasingly stringent, understanding and controlling slurry stability has become an essential aspect of CMP process development and quality assurance.
Effective particle characterization is fundamental to evaluating slurry performance and identifying early signs of instability. Conventional analytical techniques such as dynamic light scattering (DLS) are widely used for routine monitoring, while electron microscopy provides detailed structural information. However, these methods may have limitations when investigating complex dispersions or detecting low levels of aggregation in particles suspended in their native liquid environment.
Nanoparticle Tracking Analysis (NTA) offers an alternative particle-level approach by tracking the Brownian motion of individual particles in suspension. This methodology enables simultaneous measurement of particle size distribution and particle concentration while providing direct insight into heterogeneous particle populations. Advanced data processing approaches, such as Finite Track Length Analysis (FTLA), can further improve size distribution resolution and help identify subtle changes that may be difficult to detect using ensemble-based techniques alone.
The full application note explores how pH influences the stability of colloidal silica used in semiconductor CMP. Using nanoparticle tracking analysis, silica dispersions were evaluated under acidic, neutral, and alkaline conditions to assess changes in particle size distribution, particle concentration, and aggregation behaviour. Measurements were performed under carefully controlled conditions using a NanoSight Pro system equipped with a low-volume flow cell and FTLA data analysis.
The study demonstrates that particle behaviour varies substantially across different pH environments. While some conditions maintain a stable and relatively uniform particle population, others exhibit broader size distributions and characteristics consistent with aggregation. These observations provide valuable insight into the relationship between slurry chemistry and colloidal stability in semiconductor manufacturing applications.
The observed behaviour is closely linked to the surface chemistry of silica particles. The ionization state of silanol (Si–OH) groups changes with pH, affecting the density of charged sites on the particle surface. Under neutral and alkaline conditions, increased negative surface charge enhances electrostatic repulsion between particles, helping maintain dispersion stability. Under acidic conditions, reduced surface charge weakens these repulsive forces and increases the likelihood of agglomeration.
Understanding these mechanisms is important for slurry formulation, materials research, process optimization, and manufacturing quality control. Early detection of aggregation can help identify risks before they affect polishing performance or product quality.
Register to access the full scientific asset and explore the complete experimental methodology, detailed particle characterization results, comparative pH analysis, and discussion of how nanoparticle tracking analysis can support CMP slurry development, quality control, and semiconductor process optimisation.
Maintaining colloidal silica stability is critical in semiconductor CMP because particle aggregation can adversely affect polishing performance, increase defectivity, and reduce process consistency. This application note demonstrates how NanoSight Pro nanoparticle tracking analysis (NTA) can be used to characterize the pH-dependent behavior of colloidal silica suspensions relevant to semiconductor CMP. By evaluating particle size distribution, particle concentration, and aggregation behavior across different pH conditions, this study highlights the value of particle-level analysis for understanding and optimizing CMP slurry performance.
Material polishing, encompassing both polishing and lapping, has long been a fundamental manufacturing process and remains an essential advanced processing technique across a wide range of industries [1]. In semiconductor manufacturing, the invention of the transistor in 1947 created a need for methods capable of producing distortion-free, mirror-finish surfaces on semiconductor wafers, particularly those made from germanium and silicon. As device performance and complexity increased, polishing technologies evolved to meet increasingly stringent surface quality requirements.
To fulfill these rigorous demands, Chemical Mechanical Polishing (CMP) was developed. CMP combines mechanical and chemical processes to remove material in a controlled manner, enabling highly planar and defect-free wafer surfaces. Today, CMP is a critical planarization process in semiconductor fabrication, where slurry properties such as particle size, concentration, and stability directly influence polishing performance and wafer quality (Fig. 1).
![[AN260714-figure1.png] AN260714-figure1.png](https://dam.malvernpanalytical.com/8a67fad7-f9a2-45fe-8b43-b48700324ecc/AN260714-figure1_Original%20file.png)
The most critical element in CMP is the slurry, an aqueous dispersion of fine abrasive particles such as silica or metal oxides. Specifically, colloidal silica produced via ion exchange is widely used for silicon wafer polishing [2].
This colloidal silica consists of negatively charged, siloxane-structured silica particles dispersed in water. Through careful control of the manufacturing process, suspensions with narrow particle size distributions, typically ranging from 20 to 200 nm, can be produced. However, the stability of colloidal silica is highly dependent on its chemical environment. Colloidal silica remains susceptible to agglomeration induced by pH fluctuations, high ionic strength, or ultrasonic treatment. Particle aggregation is a significant concern in CMP processes.
Large agglomerates, especially those exceeding 200 nm, induce surface scratches, increase wafer defectivity, and cause erratic spikes in the material removal rate (RR). Since these secondary particles degrade the wafer surface, precise evaluation of the dispersion state and rigorous aggregation control are paramount for slurry development, quality control, and process optimization.
Accurate characterization of colloidal silica is essential for understanding slurry stability and identifying conditions that may lead to particle aggregation. While direct imaging techniques such as transmission electron microscopy (TEM) provide detailed particle morphology, they require sample drying and thus deviate from in-process conditions and do not represent particles in their native liquid environment. Dynamic light scattering (DLS) is widely used in quality control and process monitoring as a rapid technique for evaluating CMP slurry conditions. However, DLS may lack the detailed resolution required for stringent applications where high-resolution data is essential for troubleshooting and research and development (R&D) and may have limited sensitivity in detecting low levels of larger agglomerates within complex or polydisperse samples.
Nanoparticle Tracking Analysis (NTA) offers a complementary approach by tracking the Brownian motion of individual particles in suspension. This particle-by-particle measurement enables simultaneous determination of particle size distribution and particle concentration, while also providing direct insight into the presence of aggregated populations. Furthermore, the Finite Track Length Analysis (FTLA) algorithm in the NS Xplorer software improves the resolution of particle size distributions, helping to reveal subtle changes in particle populations that may not be readily detected using conventional ensemble techniques.
This application note demonstrates the efficacy of NTA by analyzing colloidal silica across varying pH levels using the NanoSight Pro, detailing pH-dependent aggregation behaviors and high-resolution structural distributions.
A colloidal silica (40 wt%) dispersion was used in this study. The dispersion was adjusted to pH 2, 7, and 12 to investigate the effect of pH on particle stability and aggregation behavior.
Each sample was diluted 200,000-fold with particle-free water to achieve a particle concentration suitable for Nanoparticle Tracking Analysis (NTA).
Following pH verification, the diluted silica dispersions were loaded into 1 mL disposable syringes and introduced into the NanoSight Pro Low Volume Flow Cell (LVFC) using a syringe pump. Measurements were performed using a 405 nm laser. For each sample, ten standard videos (750 frames) were captured under precisely controlled flow conditions. These data were combined and analyzed using the Finite Track Length Adjustment (FTLA) algorithm to obtain high-resolution particle size distributions and particle concentration data.
The NTA FTLA results are summarized in Table 1 and Fig. 2. Examination of the particle size distributions revealed clear differences between the pH conditions. At pH 7 and pH 12, the distributions exhibited sharp, unimodal peaks, indicating a relatively uniform particle size distribution indicating a stable dispersion state. In contrast, the size distribution at pH 2 displayed a broader profile with a pronounced tail toward larger particle sizes, suggesting increased particle size heterogeneity and the likely presence of aggregates. This confirms that while a stable dispersion state is maintained under neutral to alkaline conditions, significant aggregation occurs under acidic conditions.
| # | Sample | Mode [nm] | Mean [nm] | Standard deviation [nm] | D10 [nm] | D50 [nm] | D90 [nm] | Concentration [particles/mL] |
|---|---|---|---|---|---|---|---|---|
| 1 | pH 7 | 124.5 | 125 | 18 | 108 | 124 | 139 | 3.90×108 |
| 2 | pH 12 | 117.5 | 118 | 17 | 101 | 117 | 133 | 4.04×108 |
| 3 | pH 2 | 116.5 | 145 | 46 | 103 | 129 | 206 | 1.58×108 |
Table 1. Particle size distribution metrics
The particle size metrics further support these observations. At pH 7 and pH 12, the mode and mean particle sizes were similar, consistent with narrow size distributions. In contrast, the pH 2 sample exhibited a larger mean size and substantially increased standard deviation, indicating greater variability in particle size. Furthermore, only at pH 2 did the D90 value increase substantially (to 206 nm), clearly indicating the formation of larger, coarse particles or agglomerates under acidic conditions.
![[AN260714-figure2.png] AN260714-figure2.png](https://dam.malvernpanalytical.com/30b229ff-701c-4ae4-8965-b4870032b8ed/AN260714-figure2_Original%20file.png)
Differences were also observed in particle concentration. The estimated particle concentrations at pH 7 and pH 12 were comparable, at 3.90 × 108 particles/mL and 4.04 × 108 particles/mL, respectively. However, the concentration decreased to 1.58 × 108 particles/mL at pH 2, representing a reduction of approximately 60% relative to the neutral and alkaline samples. This decrease is consistent with particle aggregation, whereby multiple primary particles combine to form larger agglomerates, reducing the total number of detectable particles. Representative particle images (Fig. 3) further support the direct observation and presence of aggregation under acidic conditions.
![[AN260714-figure3.png] AN260714-figure3.png](https://dam.malvernpanalytical.com/2d3ca521-0f52-42dd-a7a9-b4870032b7ea/AN260714-figure3_Original%20file.png)
The dispersion stability of colloidal silica is highly dependent on the ionization state of the surface silanol groups (Si–OH), which varies with pH. Under neutral to alkaline conditions, deprotonation of these groups increases the density of negatively charged Si–O− sites on the particle surface. The resulting electrostatic repulsion between particles helps maintain a stable dispersion and suppress aggregation. In contrast, under acidic conditions, protonation of silanol groups reduces the surface charge, weakening electrostatic repulsion and increasing the likelihood of particle agglomeration.
The NTA results are consistent with the expected pH-dependent behavior of colloidal silica. The observed changes under acidic conditions indicate reduced colloidal stability, leading to particle aggregation and the formation of larger particle populations. From the perspective of the CMP process, the presence of coarse particles exceeding 200 nm is of particular concern because agglomerates can increase the likelihood of wafer scratches, defect formation, and variations in material removal rate (RR). The increase in D90 to 206 nm at pH 2 suggests the presence of coarse particles that may negatively impact polishing performance and wafer surface quality.
Moreover, this study demonstrates that NTA accurately detects the initial stages of micro-aggregation, evaluates the broadening of the large-particle size distribution (as evidenced by the increased D90), and monitors changes in dispersion states via decreases in absolute particle concentration. NTA enables the detection of aggregation-related changes that may not be fully captured by ensemble-averaging techniques alone. This capability makes NTA a valuable tool for slurry development, quality control, and process optimization in CMP applications.