As semiconductor device geometries shrink, controlling trace metal contamination on silicon wafers becomes increasingly demanding. VPD-ICP-MS analysis—vapor phase decomposition followed by inductively coupled plasma mass spectrometry—helps laboratories measure surface contaminants at ultratrace levels. A key challenge is separating low-concentration analyte signals from interference generated by the silicon-rich, acidic sample matrix.
A method demonstrated with a triple quadrupole ICP-MS addresses that challenge by combining cold plasma, kinetic energy discrimination, and reaction-cell approaches. The practical lesson for semiconductor laboratories is that measurement conditions can be matched to different elements within a single analytical method, rather than relying on one mode for every analyte.
Why silicon wafer analysis needs sensitive methods
Silicon used in semiconductor manufacturing is specified at very high purity, and routine monitoring of wafer-surface contamination is important to process control. Total reflection X-ray fluorescence (TR-XRF) is widely used because it is non-destructive. However, the technique’s detection limits may not meet the requirements of newer wafer production technologies, particularly for wafers with oxide or nitride films.
Vapor phase decomposition takes a different approach. The surface film is etched, and a drop of concentrated acid is scanned across the wafer to collect contaminants for analysis. Coupling this preparation with ICP-MS can provide detection limits more than two orders of magnitude lower than TR-XRF, while supporting routine multi-element measurement.
Managing matrix effects in VPD-ICP-MS analysis
VPD samples can contain substantial acid and silicon, while target metals are present at very low concentrations. In the plasma and mass spectrometer, silicon, fluorine, oxygen, and other matrix components can form polyatomic ions at the same nominal masses as analytes. For example, silicon-based species can interfere with measurements of titanium, iron, nickel, copper, zinc, and germanium.
These overlaps can raise background and compromise accuracy if they are not controlled. Triple quadrupole ICP-MS and high-resolution ICP-MS are among the approaches used to address such interferences. In the demonstrated method, the instrument used three operating modes: single-quadrupole kinetic energy discrimination (SQ-KED), single-quadrupole cold plasma with ammonia reaction gas (SQ-CP-NH3), and triple-quadrupole operation with oxygen (TQ-O2).
How the measurement modes reduce interference
Cold plasma and kinetic energy discrimination
Cold plasma uses lower plasma power than conventional hot-plasma operation. In the reported configuration, it was paired with ammonia reaction gas for selected elements. Other analytes were measured using helium-based kinetic energy discrimination, which helps reduce the contribution of interfering ions. The choice of mode varied by analyte; no single setting was used for all 26 elements.
Triple quadrupole oxygen modes
Triple quadrupole operation adds a first mass filter (Q1) before the collision/reaction cell (Q2) and a second mass filter (Q3) after it. This sequence provides additional control over which ions enter the cell and which product ions reach the detector.
For titanium, oxygen is used in a mass-shift approach. Q1 selects the target mass, Q2 converts titanium ions to titanium oxide product ions, and Q3 selects those product ions. Potential matrix ions that do not undergo the same reaction can therefore be excluded. For germanium, the method uses oxygen in an on-mass approach: the silicon oxide interference is preferentially converted to higher oxides, while germanium remains at its original mass for selection by Q3.
Results in a silicon matrix
The evaluation used a simulated VPD sample prepared from a silicon wafer and diluted to a 200 mg/L silicon matrix. Standards were prepared in that matrix at 25, 50, and 100 ng/L. Recovery testing used 50 ng/L spikes, and detection limits were calculated as three times the standard deviation of ten calibration-blank measurements.
For the 26 reported elements, detection limits ranged from 0.1 to 2.3 ng/L. Spike recoveries were 90% to 104%, indicating accurate recovery across the tested analytes and conditions. The results also reported repeatability for five measurements of a silicon matrix sample; overall reproducibility was described as ±2.2% relative standard deviation. Detection limits depend on the sample and method conditions, so these values should be viewed in the context of the specific evaluation rather than as universal instrument limits.
Calibration curves for selected elements—including lithium, potassium, calcium, titanium, vanadium, and arsenic—showed strong linearity and sensitivity in the ng/L range. For challenging analytes such as titanium, triple quadrupole interference removal combined with cold plasma operation helped reduce background.
Implications for semiconductor laboratories
For engineers and laboratory managers, the central consideration is the variety of interference mechanisms across a multi-element panel. A robust method may need to combine plasma conditions and mass-filtering strategies, selecting the approach that best addresses each analyte’s matrix effects. The ability to switch between hot and cold plasma, and between single- and triple-quadrupole modes during a measurement, supports this tailored approach.
Sample preparation remains equally important. The demonstrated workflow used pre-cleaned PFA containers, ultrapure water, and semiconductor-grade acid rinse solutions—measures consistent with the need to control contamination when measuring ng/L concentrations. Matrix-matched standards also provide a calibration environment comparable to the silicon content of the samples.
Conclusion
VPD-ICP-MS analysis offers semiconductor teams a way to assess wafer-surface metals at concentrations below the practical reach of some non-destructive screening methods. In the demonstrated approach, cold plasma, KED, and triple quadrupole oxygen modes addressed different analytical challenges in a silicon-rich matrix. The reported detection limits, recoveries, and repeatability show the value of combining analyte-specific measurement modes with careful contamination control and matrix-matched calibration.