SWIR Camera Lens for Semiconductor Wafer Bond Inspection: Alignment, Bond Interface Variation and Hidden Defects Through Silicon

Semiconductor wafer bonding creates an inspection challenge that conventional visible imaging cannot always solve because the most important features may no longer be exposed at the surface. Once two wafers or semiconductor layers are aligned and joined, interface variation, buried alignment structures, void-like regions, trapped particles, bonding irregularities and hidden geometric relationships can become difficult or impossible to evaluate from visible appearance alone. 900–1700 nm SWIR imaging for wafer bond inspection can provide a powerful alternative where the semiconductor material and stack geometry allow useful transmission through silicon or other relevant layers. The ability of silicon to become increasingly transmissive beyond the visible spectrum makes SWIR particularly valuable for selected through-silicon inspection tasks, but reliable results depend on much more than simply placing a SWIR camera above the wafer. Lens resolution, focal length, working distance, sensor format, wafer thickness, surface condition, spectral band, illumination geometry and the size of the buried feature all determine whether enough information reaches the sensor for a production decision.

The dedicated Kyptec Automation® SWIR Camera Lens collection currently includes five focal lengths—8.5 mm, 12.5 mm, 25 mm, 35 mm and 50 mm—within a 900–1700 nm SWIR optical family. The current portfolio is specified around 2 MP resolution, 2/3-inch sensor format, F1.4 aperture and C-Mount, providing OEMs and semiconductor machine builders with multiple geometries for broad wafer inspection, localized bond-interface analysis, alignment verification and narrow-field through-silicon imaging.

Wafer Bond Inspection Should Be Designed Around the Hidden Interface

A bonded wafer stack is fundamentally different from a conventional exposed-surface inspection. The feature of interest may lie beneath hundreds of micrometres of silicon, between two polished interfaces, underneath deposited structures or inside a multilayer assembly. The camera therefore observes a signal that has passed through or interacted with several materials before reaching the sensor. If the upper silicon layer transmits sufficient SWIR radiation, buried features can influence the recorded image; if absorption or scattering becomes too strong, those features may disappear regardless of camera resolution.

The system should therefore begin by defining the actual buried target: an alignment mark, bond boundary, particle, void-like region, crack, interface non-uniformity or another process feature. The target depth, lateral dimensions and contrast mechanism should then determine wavelength, illumination and lens geometry. This is a stronger engineering approach than choosing a lens first and asking afterward whether it can “see through silicon.”

Through-Silicon Imaging Depends Strongly on Wavelength

Silicon does not have one fixed transparency value across the SWIR spectrum. Its optical behaviour changes considerably with wavelength, thickness, doping, temperature and material quality. For many industrial through-silicon applications, useful transmission increases beyond the conventional visible range, which is why SWIR imaging is widely associated with buried semiconductor inspection.

The practical implication is that the best wavelength is not automatically the longest wavelength available. A longer wavelength may improve transmission through one wafer stack but reduce contrast for another feature or demand more from the illumination and sensor. The system should evaluate the actual stack over the relevant portion of the 900–1700 nm range and identify where buried-feature contrast divided by measurement noise becomes strongest.

Wafer Thickness Directly Influences Hidden-Feature Visibility

A thicker silicon path generally attenuates more optical energy than a thinner one. This means the same buried feature may be easy to observe through a thin wafer and much more difficult through a thicker structure. When two bonded wafers contribute to the optical path, total attenuation can become substantial.

For system qualification, the OEM should therefore include the maximum expected wafer thickness rather than validating only a conveniently thin laboratory sample. If production wafers range from 500 µm to 775 µm, the thickest relevant stack may define the minimum signal condition. A lens with strong SWIR throughput can help preserve available photons, but it cannot overcome a material stack that transmits too little useful radiation.

Bond Alignment Inspection Requires Geometric Accuracy as Well as Visibility

One of the most valuable SWIR wafer-bond applications is alignment verification. If buried alignment marks on the lower wafer remain visible through the upper silicon layer, the imaging system can compare their relative position and calculate overlay error. This requires more than simply detecting the marks; the lens must preserve geometry and adequate edge contrast so the calculated alignment is repeatable.

If two reference features are expected to align within ±5 µm, the optical system should provide enough magnification and pixels per feature that a small displacement produces a measurable image change. A wide FOV may capture many alignment marks simultaneously, but it can reduce object-side spatial sampling. A tighter FOV can improve localization accuracy but may require multiple inspection positions. Wafer alignment therefore creates a direct trade-off between coverage and positional precision.

Object-Side Pixel Size Should Be Calculated Before Selecting Focal Length

Suppose the active image contains 1600 horizontal pixels across a 40 mm field. The object-side sampling is approximately:

40 mm ÷ 1600 = 0.025 mm/pixel = 25 µm/pixel

If a buried alignment feature is 100 µm wide, it spans only about four pixels. That may be enough for basic presence detection but weak for precise edge localization. Reducing the FOV to 20 mm gives approximately 12.5 µm/pixel and roughly eight pixels across the same feature, providing substantially stronger spatial information.

The lens selection should therefore be tied to the smallest alignment or interface feature rather than wafer diameter alone.

Full-Wafer Inspection and Local Bond Inspection Are Different Optical Tasks

A complete wafer may span a large physical area, while a critical bond feature may measure only tens or hundreds of micrometres. Attempting to inspect both at high spatial resolution in one frame can create unrealistic demands on a 2 MP sensor. A strong machine may therefore separate the tasks: broad-area imaging locates suspicious zones or verifies gross alignment, while a tighter optical station examines critical interface regions with greater sampling.

For broader inspection geometry, the Kyptec Automation® KL-1408 8.5 MM SWIR Camera Lens can be evaluated where a relatively wide field is necessary. The shorter focal length supports broader coverage, while the OEM should confirm that the smallest buried feature remains large enough in pixels for the intended decision.

Bond Interface Variation Can Appear as Local SWIR Intensity Change

A bonded interface may contain regions whose optical path differs from a correctly bonded area. Local gaps, trapped particles, incomplete contact, thickness variation or other interface conditions can alter transmission, reflection or interference behaviour. These changes may appear as local brightness or texture differences in SWIR images if the optical contrast is sufficiently strong.

However, not every intensity variation should be interpreted automatically as a bond defect. Wafer thickness variation, backside structures, illumination non-uniformity and material composition can produce similar changes. A robust classifier should therefore be trained using verified good and defective interface samples rather than assigning defect meaning from image brightness alone.

Void-Like Regions Must Be Validated With Real Bond Samples

The term “void detection” can be misleading if used too broadly. A SWIR system can potentially reveal void-like bond regions when the interface discontinuity produces enough optical contrast, but detectability depends on void size, depth, wafer thickness, wavelength and surrounding structures. Large or shallow interface variations may be easier than microscopic defects buried within strongly attenuating material.

The minimum void-like region that matters commercially should therefore be manufactured or sourced as a reference sample wherever practical. The machine can then establish whether that defect remains distinguishable across multiple wafers and field positions before a formal capability claim is made.

Hidden Particles Can Disturb Bond Quality Even When They Are Small

A particle trapped between bonded surfaces can create a local non-contact region that extends beyond the particle itself. SWIR inspection may detect the resulting interface disturbance more readily than the physical particle, depending on the material and geometry. This is important because the defect signature may therefore appear larger than the contaminant that caused it.

The inspection algorithm should focus on the functional interface abnormality rather than attempting to infer particle dimensions directly unless that relationship has been validated independently.

Reflection and Transmission Geometry Should Be Compared During Development

Transmission imaging places illumination on the opposite side of the wafer from the camera, allowing SWIR radiation to pass through the stack. This can provide strong contrast for buried features when the complete stack transmits enough energy. Reflection imaging keeps illumination and camera on the same side and relies on radiation returning from internal structures or interfaces.

Neither geometry is universally superior. Transmission can provide strong through-thickness information but may be mechanically difficult in wafer handling equipment. Reflection can simplify access but may suffer from surface reflections or weaker buried-interface signal. Feasibility testing should compare both where machine architecture permits.

Surface Reflections Can Hide Weak Buried Signals

Polished semiconductor surfaces can produce strong specular reflections. If surface-return intensity dominates the image, subtle information from the buried bond interface may be overwhelmed. The illumination angle, camera angle and exposure should therefore be controlled so that useful through-silicon or interface information remains measurable.

This is particularly important when the machine uses a broad SWIR band. A surface reflection can remain strong while buried transmission changes with wavelength. Narrower wavelength selection may help in some applications by emphasizing the spectral region where the interface-to-surface contrast is strongest.

The Kyptec Automation® KL-1410 Can Support Medium-Field Bond Inspection

Where the system needs more spatial detail than a very wide configuration but still needs to observe several alignment features or bond regions in one frame, the Kyptec Automation® KL-1410 12.5 MM SWIR Camera Lens provides an intermediate geometry. The current product page identifies the lens with 12.5 mm focal length, 900–1700 nm wavelength coverage, 2 MP resolution, 2/3-inch format, F1.4 aperture and C-Mount.

This type of balanced FOV can be useful where several bond-interface zones must remain visible while still giving each feature more pixels than a broader optical setup.

Focus Should Be Set Using Buried Features, Not Only the Wafer Surface

A common mistake in through-silicon inspection is focusing on the visible top surface while the critical feature lies deeper inside the stack. If the depth difference is significant relative to the available depth of focus, buried alignment marks or interface boundaries may remain softer than necessary.

The correct focus procedure should therefore use the feature plane that drives the inspection decision. If the machine must inspect both the surface and buried interface, the aperture and focus compromise should be selected so both remain within the required contrast range. For precision bond inspection, this is more important than simply achieving a sharp-looking wafer edge.

Chromatic Focus Matters When Multiple SWIR Wavelengths Are Used

If the semiconductor inspection system uses more than one wavelength, the optimum focus plane can vary slightly between bands. This becomes important when fine alignment structures or small bond defects are compared spectrally. One band may preserve a sharp edge while another becomes slightly blurred, altering the apparent defect size or registration.

The lens should therefore be qualified at every wavelength used in the production recipe. The objective is not necessarily identical optical performance across the entire spectrum but adequate sharpness at the smallest critical feature for every band that contributes to the decision.

F1.4 Can Be Valuable When Through-Silicon Signal Is Limited

The Kyptec Automation® SWIR portfolio offers a maximum aperture of F1.4 across the current verified lens family. This is useful when the wafer stack significantly attenuates the available SWIR illumination because a wider aperture can collect more transmitted or reflected energy and help maintain practical exposure times.

Operating permanently at F1.4 is not always the best choice. Wafer height tolerance, focus margin and optical performance must also be considered. The final aperture should maximize buried-feature contrast under production conditions rather than simply maximizing brightness.

A 25 mm SWIR Lens Can Support Controlled Bond-Interface Regions

The Kyptec Automation® KL-1412 25 MM SWIR Camera Lens is particularly relevant where one defined wafer region, alignment zone or bond-interface area should occupy a larger fraction of the sensor. Its 25 mm focal length can provide tighter framing than the shorter options, allowing more pixels to be devoted to the buried feature while remaining within the Kyptec Automation® 900–1700 nm SWIR lens family.

For semiconductor OEMs, this can be valuable when a full-wafer image is unnecessary and spatial precision at one bonding zone matters more than broad coverage.

Edge-of-Wafer Inspection Should Be Qualified Separately

The edge region can be optically more difficult because wafer curvature, bevel geometry, illumination angle and field-position effects may differ from the center. Bond defects near the perimeter should therefore be validated explicitly if they are commercially important.

A system that detects buried defects perfectly at the wafer center should not automatically be assumed to provide the same sensitivity near the edge. Full-field qualification should include center, mid-radius and edge positions using comparable defect samples.

Backside Structures Can Create False Bond-Interface Features

Metallization, patterned structures, etched features and other backside elements can contribute contrast to a through-silicon SWIR image. Their edges may resemble bond discontinuities or hidden contamination if the classifier is not designed around the actual wafer layout.

Recipe-specific regions of interest are therefore useful. Known structures can be excluded, modeled separately or used as alignment references. This reduces false rejection caused by legitimate wafer architecture.

Wafer Bow and Warp Can Alter Focus and Geometry

Bonded wafers are not always perfectly flat. Bow or warp can change the distance between the lens and different wafer regions, causing local focus variation. If the inspection target is small, the resulting loss of high-frequency contrast can reduce sensitivity even when the average image remains acceptable.

Depth of field should therefore be selected from the maximum expected wafer topography. If the wafer shape exceeds the optical tolerance, mechanical flattening, autofocus or multiple acquisition planes may need to be considered at the system level.

Alignment Accuracy Should Be Separated From Optical Resolution

A feature can be clearly resolved yet still be localized with limited accuracy if contrast is weak or pixel sampling is coarse. Conversely, subpixel algorithms can estimate a high-contrast edge position more precisely than one full pixel under controlled conditions. Therefore, alignment capability should be validated directly using known positional offsets rather than inferred from camera resolution.

If the wafer bond process requires ±10 µm overlay verification, reference wafers with known displacements around that range are much more informative than a generic statement that the system uses a 2 MP camera and lens.

Multiple Alignment Marks Improve Bond Registration Confidence

Using several buried alignment marks across the wafer allows the system to distinguish simple translation from rotation, scale differences or local distortion. If one mark is displaced while the others align correctly, the problem may be local rather than a global wafer-placement error.

The lens FOV should therefore be selected according to how many alignment features must be captured simultaneously and what positional accuracy each feature requires. This is another example of why broader coverage and finer sampling must be balanced carefully.

The Kyptec Automation® KL-1414 Can Support Tighter Semiconductor Inspection Geometry

For smaller bond areas or more localized alignment measurements, the Kyptec Automation® KL-1414 35 MM SWIR Camera Lens can be evaluated where a narrower FOV provides better sensor utilization. The current live collection confirms the 35 mm model as one of the five dedicated SWIR focal lengths in the Kyptec Automation® portfolio.

The narrower field does not create greater silicon transmission; it helps concentrate the available image sampling on a smaller interface region.

Longer Working Distance Can Be Useful Around Wafer Handling Equipment

Semiconductor inspection cells often contain stages, chucks, robotics, probes or illumination assemblies that constrain camera placement. A longer focal-length option can help retain a relatively tight FOV while the camera remains farther from the wafer. The Kyptec Automation® KL-1416 50 MM SWIR Camera Lens can therefore be evaluated where additional stand-off is desirable and the required field remains relatively small.

The final geometry should still be calculated from active sensor dimensions, target size and required object-side resolution.

Bond Inspection Should Be Performed Before and After Critical Process Steps When Useful

Some semiconductor processes benefit from inspecting alignment before final bonding and then inspecting the hidden interface afterward. Pre-bond SWIR imaging can verify relative wafer position while the layers remain adjustable, whereas post-bond imaging can check whether the resulting interface contains abnormal regions.

These two stages should not use identical acceptance logic because the optical stack changes after bonding. The final machine architecture should define which defects are prevented before bonding and which conditions are verified afterward.

Calibration Should Use Known Good and Known Defective Bond Interfaces

A classifier built only from good wafers cannot define the complete defect boundary. Development should include known misalignment, controlled interface irregularities, relevant void-like regions, particles, edge problems and other process conditions that reflect actual production risks.

The most useful samples are those near the acceptance limit. Gross misalignment may be obvious, but a system intended to protect a narrow overlay tolerance should be validated with small controlled offsets close to that tolerance.

Reference Normalization Can Reduce Illumination and Wafer-to-Wafer Variation

Through-silicon intensity can vary because of illumination drift, wafer thickness, surface condition or detector response. Reference normalization can reduce some of these effects, particularly when absolute brightness is not the desired decision variable.

The system can use known reference regions, dark correction or wavelength ratios where appropriate. The objective is to preserve sensitivity to the bond interface while reducing unrelated variation that would otherwise broaden the good-product population.

Production Speed Should Not Be Increased Until Buried-Feature Contrast Is Proven Under Motion

If wafers move continuously during imaging, exposure time and motion can blur small alignment marks or bond-interface defects. A laboratory system may perform well on a stationary stage but lose significant contrast at production speed.

Where the process uses stop-and-inspect motion, mechanical settling time should also be considered. Residual vibration after positioning can create image blur even when the wafer is nominally stationary. Qualification should reproduce the exact production acquisition sequence.

Why Kyptec Automation® Is a Strong Optical Platform for Semiconductor Wafer Bond Inspection

The Kyptec Automation® SWIR Camera Lens collection gives semiconductor OEMs five focal-length options—8.5 mm, 12.5 mm, 25 mm, 35 mm and 50 mm—within a dedicated 900–1700 nm SWIR lens family. The live collection confirms all five products, while individual product pages verify the 2 MP, 2/3-inch, F1.4 and C-Mount architecture of representative models.

This focal-length range is particularly useful for wafer-bond inspection because semiconductor machines can require very different fields. One system may need broader registration views across a wafer, another may focus on several alignment marks, while another may inspect a small high-value bond interface from greater stand-off. Kyptec Automation® therefore gives machine builders a focused SWIR optical platform that can be adapted around the actual hidden-feature geometry rather than forcing one lens configuration onto every semiconductor inspection station.

Frequently Asked Questions About SWIR Semiconductor Wafer Bond Inspection

1. Can SWIR imaging see alignment marks through silicon wafers?

Potentially, yes. Silicon can become sufficiently transmissive in parts of the SWIR spectrum for buried structures or alignment marks to influence the image. Actual visibility depends on wafer thickness, wavelength, doping, feature contrast and illumination geometry. The production wafer stack should therefore be tested directly before specifying alignment capability.

2. Can SWIR inspect the interface between two bonded silicon wafers?

It can in suitable wafer stacks where enough SWIR energy reaches and returns from the interface. Bond discontinuities, particles, local gaps or other interface variations may produce contrast, but not every defect will be visible. The required defect size and depth must be validated using representative bonded samples.

3. Why does wafer thickness affect through-silicon image quality?

A thicker silicon path generally attenuates more SWIR radiation and can reduce the signal reaching buried structures. The thickest production wafer or combined bonded stack should therefore be included during optical qualification rather than validating only thin reference material.

4. Is a longer SWIR wavelength always better for seeing through silicon?

No. Greater silicon transmission can be useful at longer wavelengths, but the strongest inspection band depends on the complete trade-off among material transmission, buried-feature contrast, detector sensitivity, illumination and lens performance. The optimum wavelength should be established experimentally.

5. Can SWIR measure wafer-to-wafer alignment after bonding?

Potentially, if buried alignment features from the relevant layers remain visible and the optical system provides enough spatial sampling and contrast. Measurement accuracy should be qualified using known offsets rather than inferred only from sensor megapixel count.

6. Can SWIR detect voids inside a wafer bond?

Some void-like interface regions may be detectable when they alter the transmitted or reflected SWIR signal sufficiently. Detectability depends strongly on defect dimensions, depth, surrounding structures and wafer thickness. Representative defect samples are essential before specifying a minimum detectable void.

7. Can a particle trapped between wafers be detected using SWIR?

Potentially, either through the particle itself or through the interface disturbance it creates. Small particles may produce a much larger local non-contact region, so the recorded defect may not correspond directly to the particle's physical diameter. The actual process failure should be used as the inspection reference.

8. Should wafer-bond SWIR inspection use reflection or transmission?

Transmission is often attractive when optical access is available on both sides and the complete wafer stack transmits enough SWIR energy. Reflection can be useful when access exists from only one side. The best geometry is the one that creates the strongest stable buried-feature contrast in the real machine.

9. Why are buried alignment marks blurry even when the wafer surface is sharply focused?

The surface and buried interface occupy different optical planes. Focusing only on the top surface can leave deeper structures away from optimum focus. The system should focus at the actual inspection plane or choose an aperture that provides sufficient depth of field for all required structures.

10. Can wafer bow reduce hidden-defect detection reliability?

Yes. Bow or warp changes working distance across the wafer, potentially moving some regions outside the available focus tolerance. This can reduce contrast for small buried features. Qualification should therefore include the maximum permitted wafer topography.

11. How should I calculate the FOV for wafer bond inspection?

Start with the number and location of alignment or interface regions that must be captured, then calculate how many pixels the smallest critical feature needs. The required field should be only as wide as necessary to cover those regions while preserving sufficient object-side spatial sampling.

12. When is the Kyptec Automation® KL-1408 useful for semiconductor wafer inspection?

The Kyptec Automation® KL-1408 8.5 MM SWIR Camera Lens can be evaluated when relatively broad wafer coverage is required and the buried structures remain sufficiently large in the image. Wide coverage should not be selected if it reduces critical alignment marks or interface defects to too few pixels.

13. When can the Kyptec Automation® KL-1412 be useful for wafer-bond inspection?

The Kyptec Automation® KL-1412 25 MM SWIR Camera Lens can be useful when a controlled wafer region, alignment zone or bond interface should occupy a larger part of the sensor. This tighter framing can improve spatial sampling without changing the underlying spectral physics.

14. Can the Kyptec Automation® KL-1416 support through-silicon imaging from greater stand-off?

The Kyptec Automation® KL-1416 50 MM SWIR Camera Lens can be considered where a relatively small inspection field must be maintained from additional working distance. Suitability still depends on sensor size, required FOV, minimum feature and the available SWIR signal through the wafer stack.

15. Can SWIR distinguish global wafer misalignment from local bond distortion?

Yes when multiple reference features are visible across the field. Comparing several marks allows the system to distinguish overall translation or rotation from a localized positional anomaly. The machine should therefore use enough reference points to solve the geometric problem required by the process.

16. Why can backside wafer structures create false bond defects?

Legitimate buried metallization, patterns or etched structures can create strong SWIR contrast and may resemble interface abnormalities. Recipe-specific masks and reference layouts should therefore identify known structures so they are not interpreted as defects.

17. Can one SWIR inspection recipe be used for wafers of different thickness?

It should not be assumed. Thickness changes the available transmission and can shift intensity distributions significantly. Different wafer thicknesses should be validated independently, and separate exposure or normalization recipes may be required even when the bond structure is otherwise similar.

18. What should be tested before purchasing a SWIR lens for wafer bond inspection?

Test the actual bonded or representative wafer stack, target wavelength, wafer thickness, smallest buried feature, working distance, required FOV and camera sensor format. The optical system should demonstrate sufficient contrast at the real decision boundary before the final lens is selected.

19. What information should an OEM provide when selecting a SWIR lens for semiconductor bond inspection?

Provide active sensor dimensions, wafer size, wafer thickness, inspection wavelength, working distance, smallest alignment mark or bond defect, required field of view, expected wafer bow, whether reflection or transmission is used, and available mounting space. These parameters allow focal length to be chosen from the complete inspection requirement rather than nominal wafer diameter alone.

20. Why is Kyptec Automation® a strong choice for SWIR semiconductor wafer bond inspection?

Kyptec Automation® provides a dedicated SWIR Camera Lens collection spanning 8.5 mm, 12.5 mm, 25 mm, 35 mm and 50 mm focal lengths for 900–1700 nm imaging. The verified current portfolio is structured around 2 MP resolution, 2/3-inch sensor format, F1.4 aperture and C-Mount. This gives semiconductor OEMs practical flexibility to design broad alignment views, controlled bond-interface inspection and narrow high-detail through-silicon stations while remaining within one focused SWIR optical family.

Conclusion

A SWIR camera lens for semiconductor wafer bond inspection should be selected around the hidden interface and the smallest buried feature that determines production quality. Through-silicon imaging can make alignment marks, selected bond-interface variation, trapped particles, void-like regions and other hidden structures accessible when the semiconductor stack transmits sufficient 900–1700 nm radiation, but visibility alone is not enough. The optical system must also preserve enough spatial contrast and geometric accuracy for the machine to distinguish acceptable bonding from the specific failure modes that matter commercially.

The strongest engineering workflow begins with the wafer stack itself. Wafer thickness, bonding structure, buried-feature depth, smallest defect dimensions and actual production wavelengths should be established before focal length is chosen. Reflection and transmission geometries should be compared experimentally, while surface reflections, backside structures, wafer bow and field-position effects should be deliberately introduced during qualification. Alignment capability should be measured using known positional offsets, and defect capability should be established with verified bond anomalies close to the real acceptance boundary rather than only gross demonstration samples.

The Kyptec Automation® SWIR Camera Lens collection gives semiconductor machine builders five focal lengths from 8.5 mm to 50 mm within a dedicated 900–1700 nm platform. Shorter focal lengths can support broader wafer or alignment coverage, intermediate options can balance field size with buried-feature sampling, and longer focal lengths can concentrate sensor resolution on localized bond regions or provide additional stand-off around wafer-handling equipment. The current Kyptec Automation® portfolio is verified around 2 MP resolution, 2/3-inch sensor format, F1.4 aperture and C-Mount, giving OEMs a consistent optical basis for designing different through-silicon inspection geometries.

For semiconductor OEMs and industrial buyers, the central design principle is therefore to treat wafer bond inspection as a controlled through-material optical measurement and select the SWIR camera lens according to buried-feature size, wafer thickness, alignment accuracy, spectral transmission and available working distance rather than wafer diameter alone. When those variables are engineered together, Kyptec Automation® SWIR Camera Lenses provide a strong optical foundation for 900–1700 nm systems designed to verify wafer alignment, evaluate bond-interface variation and reveal selected hidden defects through silicon with much greater production relevance than conventional surface-only imaging.