Short-Wavelength InfraRed Image Sensor Technology SenSWIR™

Overview

SenSWIR is a wide-band and high-sensitivity SWIR image sensor technology implemented by the combination of compound semiconductor InGaAs photodiodes and Silicon readout circuits through Cu-Cu bonding.

SWIR (Short Wavelength Infra-Red) light penetrates and is absorbed by different substances than visible light, so its attributes can be applied in a variety of different situations.

Logo of SenSWIR

*) SenSWIR and its logo are registered trademarks or trademarks of Sony Group Corporation or its affiliates.

Technical Features

Higher pixel count, smaller systems

Creating SWIR sensors with smaller pixels than in current industrial CMOS image sensors has been challenging with conventional bump bonding, because a certain bump pitch must be maintained to bond the indium-gallium-arsenide (InGaAs) photodiode layer to the silicon readout circuit layer. With SenSWIR technology, Cu-Cu connection^*1 enables a finer pixel pitch and smaller pixels. As a result, smaller high-resolution cameras can be developed, which can support higher inspection precision.

*1)A technique that provides electrical conduction by bonding copper pads, as the pixel chip (top) and logic chip (bottom) are stacked. Advantages over the previous through-silicon via approach (which electrically connects top and bottom chips at the edge of the pixel area) include smaller systems and improved performance, which affords greater freedom in design and promises higher productivity.

Broad imaging that extends to the visible spectrum

The top indium-phosphorus (InP^*2) layer inevitably absorbs some visible light, but applying Sony's SWIR sensor technology makes this layer thinner, so that more light reaches the underlying InGaAs layer. The sensors have high quantum efficiency even in visible wavelengths. This enables broad imaging of wavelengths from 0.4 μm to 1.7 μm. A single camera equipped with the sensor can now cover both visible light and the SWIR spectrum, which previously required separate cameras. This results in lower system costs. Image processing is also less intensive, which accelerates inspection. These advances promise to expand the scope of inspection significantly.

*2) Substrate that forms the base of the InGaAs layer.

Comparison between bump connection and Cu-Cu connection

What's SWIR？

Generally, light with a wavelength of 400 nm to 780 nm is referred to as visible light, and light with a wavelength of 780 nm to 10⁶ nm as infrared light. The wavelength band of SWIR is from 900 nm to 2,500 nm, which is the region of infrared light closest to visible light. Image sensors equipped with SenSWIR technology are capable of broad imaging over the range of 400 nm - 1,700 nm, including visible light as well as SWIR light.

Applications

Sorting fruits and vegetables
Container content inspections
Contaminant detection
Detection of hair contamination in processed foods
Sorting materials
Resin products inspection
Positioning in Semiconductor Manufacturing
Manufacturing lithium-ion batteries
Solar panel inspection
Temperature monitoring
Firefighting
Remote monitoring
Observation of agricultural lands

Sorting fruits and vegetables

Water becomes black in an image taken with a SWIR image sensor set at the wavelength of 1,450 nm because water absorbs the light at this wavelength. Since this makes it possible to detect moisture contained in objects, it can be used for applications such as fruit and vegetable sorting.

Example of sorting fruits by detecting dents or scratches

Image captured under visible light in sorting fruits and vegetables

Image captured under SWIR in sorting fruits and vegetables

SWIR imaging makes the moisture concentrating in dents on the apples visible.

Related sectors

Agriculture and Farming

Container content inspections

In food manufacturing processes, performing final content inspections is difficult if the food packages are opaque. In some cases contents are pinched in the sealing part when the package is closed, which are also hard to detect.

Even if packages appear opaque in the visible range, it may be possible for light in SWIR wavelengths to penetrate them and allow their contents to be observed. By using this property, the inner areas of packages can be checked non-destructively and pinching errors can also be detected.

Example of detecting content conditions by observation that penetrates through a plastic container

Image captured under visible light in Container content inspections

Image captured under SWIR in Container content inspections

Explanation of photo: SWIR imaging makes it possible to check the content of opaque containers.

Related sectors

Food/Medicine/Cosmetic Manufacturing

Contaminant detection

In food product manufacturing, inspections to check for the introduction of contaminants are essential. However, if contaminants mixed in with a product are similar in color to the product itself, it may be difficult to identify them using only visible light.

Utilizing the absorption and reflection properties of SWIR-band light makes it possible to perceive differences between substances that are difficult to see with visible light alone. With these attributes, SWIR image sensors can be used for contaminant detection and similar applications.

Example of detecting contaminants in food

Image captured under visible light in contaminant detection

Image captured under SWIR in contaminant detection

SWIR imaging makes it possible to distinguish between a food product (black beans) and black-colored contaminants.

Example of detecting walnut shells by processing images taken at multiple SWIR wavelengths

Image taken with a standard color camera in contaminant detection

Image taken and processed with SWIR (1,050/1,200/1,450 nm) in contaminant detection

SWIR imaging enables easy identification of walnut shells that are difficult to see with the naked eye.
The image on the right is a pseudo color image taken at three SWIR wavelengths. Such image processing makes it easier to distinguish between walnut shells and the walnuts themselves.

Related sectors
Food/Medicine/Cosmetic Manufacturing

Related Links
Proposal for Optimal Multi-Spectral Cameras Using Multi-Band Filters

Detection of hair contamination in processed foods

Measures to prevent hair contamination have always been important in the production lines of processed foods. It is not easy to quickly identify fine hair visually but, by using an SWIR image sensor, even hair mixed in with food can be detected.

Example of hair detection on a croquette

This is an example of three strands of hairs on a croquette (deep-fried roll) captured by an SWIR image sensor. Capturing an image at SWIR wavelengths enables multiple hairs on the surface of the croquet to be identified. This is because hair strongly reflects light in the SWIR wavelength range and the moisture in the processed food absorbs light around 1450 nm, so is rendered as black in the captured image.
These characteristics can be utilized to apply SWIR image sensors for hair detection in the manufacturing processes of various processed foods.

Sorting materials

The use of SWIR wavelengths makes it possible to identify and sort materials even when it is difficult to sort materials in the visible light range.
Using an SWIR image sensor to capture images of materials enables the sorting of plastic containers, which is important in the recycling process, as well as leather, cotton, and other fabrics.

Example of inspection to distinguish between cotton and polyester

Images of cotton (natural fiber) and polyester (synthetic fiber) are captured by an SWIR image sensor (using multiple wavelengths) to determine the types of fiber material.

Multispectral image composed from shots at three SWIR imaging wavelengths (1150/1250/1500 nm)

The image on the right was captured at three different SWIR wavelengths and pseudo-colored for easy identification. SWIR imaging produces images that show uniform reflectance for the same material, regardless of the color of the fiber. By taking images at multiple SWIR wavelengths and processing these images, polyester and cotton can be easily distinguished.
This technique of identifying the types of fiber material can be used in clothes sorting and recycling applications.

Example of inspection to distinguish genuine leather from artificial leather

Images of genuine and artificial leathers are captured by an SWIR image sensor to distinguish the material.

The images above show genuine leather and artificial leather side by side, captured with a color camera and an SWIR image sensor, respectively.
The SWIR image sensor makes it possible to distinguish between the materials, something that is difficult to do with color camera images. The genuine leather strongly reflects light at SWIR wavelengths and appears brighter in the image. However, polyurethane, which is widely used in artificial leather, absorbs SWIR light and thus appears darker. As shown above, utilizing these material-specific SWIR wavelength characteristics makes it possible to distinguish genuine leather from artificial leather, facilitating the authentication of genuine leather products.

Related sectors

Recycling

Resin products inspection

In product inspection of resin molded products, it is necessary to identify not only scratches and stains that can be seen from the exterior but also foreign materials and defects that are hidden inside. Internal defects that cannot be confirmed visually affect the quality and life of the product and can significantly impact the maintenance of product quality.

Bubble detection in resin artificial teeth

Captured with SWIR (1450 nm) (illuminated from the back side)

The images above show a resin artificial tooth. The color camera image shows the exterior but does not provide visibility of the interior condition. The image taken with the SWIR image sensor was able to capture two air bubbles inside the artificial tooth. This is because resin, the main component of the artificial tooth, has a property that transmits light in the SWIR wavelength range. Furthermore, utilizing SWIR wavelengths enables not only air bubbles but also hair, film, and other foreign materials to be detected.
This technique can be applied not only to artificial teeth but also to the quality inspection of other resin products.

Positioning in Semiconductor Manufacturing

Due to the miniaturization of semiconductor devices in recent years, high accuracy has become required even in layering processes for silicon wafers. In order to raise their accuracy, it is important to accurately align marks used for wafer positioning.

Light in SWIR wavelength bands has the property of penetrating the silicon layers of wafers, so applying SWIR image sensors can make it possible to clearly confirm positioning marks. High definition SWIR image sensors from Sony are also anticipated to be used to improve edge detection accuracy.

Example of imaging by penetrating silicon wafers

The photograph on the right was taken in an SWIR environment, so the resolution chart behind the silicon wafer is visible. This was taken using an IMX990 sensor at a high resolution of approximately 1.34 megapixels, so even small marks can be detected with high precision. In addition, using an IMX992 sensor, which has a resolution of approximately 5.32 megapixels, enables even higher-definition inspection and measurement applications.

Related sectors

Semiconductor Manufacturing

Manufacturing lithium-ion batteries

Lithium-ion batteries consist of a stacked structure of alternating layers of multiple electrode sheets and separators. As the electrode sheets and the separators are physically stacked, it is no easy task to correctly discern their positional relationship.

SWIR-band light can penetrate the separator. Utilizing this capability, SWIR image sensors are capable of detecting misalignment even though this is difficult to achieve with visible light.

Examples of photography of electrode sheets and separators arranged in layers

Images of electrode sheets and separators arranged in layers, shot with visible light

Images of electrode sheets and separators arranged in layers, shot with SWIR

Although the electrode sheets under the separators are difficult to see with visible light, SWIR wavelength imaging can captures the misalignment of an electrode sheet vividly.

Solar panel inspection

SWIR image sensors can also be used in an inspection method to evaluate the quality of panels for photovoltaic power generation: electroluminescence (EL) inspection.
In EL inspection, a voltage is applied to the photovoltaic modules that make up a solar panel. This voltage enables light emission in a specific wavelength band, and panel abnormalities are recognized based on the emission status of each cell. Normal areas emit light in the near-infrared region and can be captured by the SWIR image sensor.

Captured with a standard color camera (fluorescent light environment)

Normal cells emit white light while abnormal cells appear black in images captured by an SWIR image sensor, even though they look the same when inspected visually. This makes it easy to identify defective panels.
Further, the high sensitivity and high speed of image capturing allow the sensors to be used to their fullest potential in EL inspection.

Temperature monitoring

Image sensors can perceive differences in the temperatures of substances as differences in luminance. Among them, substances at temperatures of roughly 250ºC or higher emit light in the SWIR band, so SWIR image sensors can be used in applications for monitoring high temperatures of 250ºC or higher. This characteristic is expected to be applied in the steel industry.

Example of monitoring the temperature at the tips of soldering irons

In the SWIR image, it is possible to confirm not only that the tips of the soldering irons have become hot, but also to identify their differences in temperature.

Related sectors

Heavy industry and plant manufacturing

Firefighting

In firefighting, smoke may compromise the firefighters’ field of vision. SWIR image sensors, which are less affected by light scattering, can be used to ascertain the situation at the sites of fires and assist firefighting efforts, as they can capture images while suppressing the impact of smoke.
SWIR light is also emitted by flames, which can be clearly captured by SWIR image sensors. This makes SWIR image sensors particularly useful for identifying the source location of forest fires and other wildfires.

Remote monitoring

In remote monitoring, airborne micro-particles can cause far-off targets to become blurred, making it difficult to accurately capture them with cameras. However, a characteristic of SWIR-band light, which has longer wavelengths than visible light, is that it is less affected by airborne micro-particles than visible light, making it easier to clearly detect distant objects.

Examples of observation in a bay environment

In general, observation in a bay area is particularly susceptible to fog. This can make it difficult to see clearly in the visible range. By using an SWIR image sensor, the camera is able to clearly capture the very end of a long bridge reaching land in the distance, things that are hidden by fog in visible light.

The image taken with the color camera shows only a vague silhouette of a vessel in the fog, but the image taken with the SWIR image sensor clearly shows not only this offshore vessel but also several offshore wind turbines.

Monitoring under poor visibility

Fine particles such as PM2.5 and smog can lead to poor visibility and affect remote monitoring.

Images taken with a color camera are whitish overall, with objects becoming hazier as they are located toward the background, whereas images taken with an SWIR image sensor capture the entire cityscape with depth and allow objects in the distance to be seen.

While images taken with a color camera show only a faint reflection of the tower behind the buildings, images taken with an SWIR image sensor clearly show details of the tower and, behind this, the mountain range with its undulating surface.

Related sectors

Remote and Wide-area Monitoring

Observation of agricultural lands

At agricultural sites, initiatives to observe agricultural lands from the sky with cameras are currently proceeding. Now that it has become possible to understand the growth conditions of crops, limited additional fertilizing and yield forecasts based on data can also be implemented.

However, it is not a simple task to make judgments on growth conditions using only color data. If SWIR image sensors are used, moisture content can be visualized so growth conditions and distribution according to moisture volume can also be expressed visually, and more accurate judgments can be made.

Related sectors

Agriculture and Farming