Semiconductor metrology is becoming more demanding as device architectures, materials and integration methods grow more complex. Nanoscale features can limit both device performance and the measurements used to characterize them. At the same time, the industry is moving beyond planar structures and relying less exclusively on dimensional scaling as a driver of progress.
For engineers, process developers and equipment teams, the result is a measurement problem that cannot be solved by a single instrument. Characterizing next-generation devices requires methods suited to three-dimensional structures, small measurement volumes, sensitive materials and features distributed across multiple layers.
Why semiconductor metrology is changing
Traditional CMOS architectures are approaching physical limits, while newer devices introduce intricate geometries and materials with different properties. Candidate technologies include devices based on graphene and transition metal dichalcogenide films, as well as one- and two-dimensional field-effect transistors and lateral or vertical heterostructures.
These materials make familiar measurements, such as film thickness, more dependent on understanding the properties of individual layers. A technique that works well for a conventional material stack may not provide enough information when material properties vary by layer or when the structure itself is difficult to access.
The measurement challenge also extends beyond materials. Metrology must connect the observed geometry and material characteristics to device structure and function, while accounting for the limits of the measurement method itself.
Three-dimensional devices raise new measurement challenges
Non-planar architectures require information about the device in three dimensions. Gate-all-around (GAA) transistors, including lateral and vertical configurations, make this task more difficult because measurement targets can be very small and the desired information highly localized. These conditions can produce low signal-to-noise ratios and complicate reliable characterization.
Monolithic three-dimensional integration adds another layer of difficulty. Measurements may have different sensitivities at different depths, so a method that detects one level effectively may not characterize another with the same reliability. Metrology approaches may need a substantial depth of focus or the ability to measure through the structure.
For 3D stacked chips and 3D very-large-scale integration, destructive analysis is also a concern: the tiers are functional parts of the device, so consuming them for characterization can be prohibitively expensive. This increases the importance of techniques that can provide useful structural information without damaging the device.
Lithography and material sensitivity complicate inspection
Each patterning route brings its own inspection requirements. Extreme ultraviolet (EUV) lithography faces challenges including mask defects, line-edge roughness and stochastic variation. Nanoimprint lithography requires attention to defects, overlay and template inspection. Directed self-assembly (DSA) introduces its own overlay and defectivity concerns, including nonsystematic overlay shifts.
New materials can be sensitive to the measurement process itself. Some two-dimensional materials are susceptible to electron-beam damage, while small features may provide low image contrast or limited cross-scattering. Obtaining optical properties such as refractive index and extinction coefficient can also be difficult. These factors influence which tools are appropriate and how measurement results should be interpreted.
Hybrid metrology combines complementary instruments
No single measurement technique provides the resolution, measurement range and low uncertainty needed to characterize every aspect of a complex three-dimensional feature. Hybrid metrology addresses this limitation by combining results from multiple instruments, each selected for the information it can contribute.
Depending on the measurement task, complementary methods can include:
- Scanning electron microscopy for imaging and dimensional information.
- Atomic force microscopy for surface and three-dimensional profile measurements.
- Critical-dimension X-ray scattering and scatterometry for information about patterned structures.
- Transmission electron microscopy for high-resolution examination of internal structure.
The value of this approach is not simply collecting more measurements. Statistical and combinatory analysis can bring together the strengths of each technique to produce a more complete interpretation of the same feature. The instruments and analysis methods need to be considered as a coordinated measurement strategy rather than as isolated tools.
Emerging methods and advanced analytics
Ptychography-based approaches are being explored to extend electron, optical and X-ray measurement capabilities. In electron ptychography, imaging of two-dimensional materials has been demonstrated without causing beam damage, with a reported resolution of 0.04 nm. X-ray ptychography has also been used to image and reconstruct whole chips, demonstrating three-dimensional imaging at a larger length scale. These examples illustrate how new measurement methods may address particular gaps, though their suitability depends on the application.
Machine learning and other advanced analytics are also expanding the role of data processing in metrology. Their use goes beyond classifying images or analyzing measurement outputs. Advanced analytics can support instrument and measurement-process optimization, as well as hybrid and virtual metrology approaches that use information from multiple sources.
For teams implementing these methods, the practical consideration is to match the measurement question to the available evidence: identify which feature or material property must be characterized, understand where a technique is sensitive or limited, and determine how complementary measurements can be combined.
Planning for the next generation of devices
As semiconductor structures become more three-dimensional and material stacks more varied, measurement planning must account for geometry, depth, signal quality, material sensitivity and the cost of destructive analysis. Hybrid metrology offers a way to combine tools with different capabilities, while advanced analytics can help synthesize their results. Emerging imaging methods may further address specific measurement gaps.
The practical implication is clear: semiconductor metrology must evolve alongside device design and lithography. Engineers and decision-makers should plan for a portfolio of complementary techniques and analysis methods, rather than expecting one instrument to deliver a complete characterization. That approach is essential to understanding increasingly complex devices and supporting their development and manufacturing.