Optoelectronics Basics #02
What Can SWIR See? The Layer Visible Light Misses
Hold a silicon wafer in your hand and it is an opaque gray. Under a SWIR camera, the same wafer looks like a piece of frosted glass: you can see directly whether there are cracks inside it.
This is not a special treatment; it is just a different waveband. Shortwave infrared (SWIR, taken as 0.9–1.7 µm in this article) forms images in much the same way as visible light: by reflected light, with shadows and contrast, and with illumination needed. The difference is only that light in this band interacts differently with materials, so some things become transparent and some things become visible.
This article covers four things SWIR can actually do that visible light cannot. If you are new to optoelectronics, you may want to start with the first article in this series, What Is Optoelectronics?
1. The first thing: silicon becomes transparent
Silicon is a semiconductor with a band gap of 1.12 eV. A photon is absorbed only if its energy is above the band gap; converted to wavelength:
λ = 1240 / 1.12 = 1107 nm ≈ 1.11 µm
For light at wavelengths longer than 1.11 µm, the photon energy is not enough to push an electron across the band gap, so absorption in silicon drops sharply and it gradually becomes semi-transparent. The transition is quite fast, but it is not a vertical line: for a common 775 µm thick wafer, internal transmission is below 1 % at 1.0 µm, about 30 % at 1.05 µm, about 75 % at 1.1 µm, and above 1.15 µm there is hardly any absorption.
This has several direct uses in the semiconductor industry:
- Internal cracks and microcracks in wafers: internal damage from dicing, grinding or handling is invisible on the surface but shows up as soon as SWIR is used
- Bonding alignment inspection: after two wafers are bonded, the alignment marks on the lower wafer can be seen through the upper one
- Backside inspection of flip-chip packages: looking in through the back of the silicon substrate to check bump bonding
- MEMS structure inspection: moving structures sealed in silicon cavities can be inspected without opening them
2. The second thing: water absorbs light
Water has two distinct absorption bands in the shortwave infrared: at about 1.45 µm and 1.94 µm. The first falls within the response range of standard InGaAs (0.9–1.7 µm); the second needs an extended-wavelength sensor. In the absorption band, wet areas turn dark; outside it, the same areas are bright. Subtract the two images, and you have a map of moisture distribution.
This is much faster than "weighing to measure moisture content", and it is contactless, full-field and real-time. Practical uses include:
- Real-time monitoring of drying processes: which areas are not dry yet
- Uniformity of wet film thickness in coating and painting
- Moisture grading of food and agricultural products
- Leak inspection of package seals: leaked water stands out clearly in SWIR
Incidentally, water vapor in the air absorbs in the same neighboring bands: the atmosphere absorbs strongly at about 1.35–1.45 µm and 1.8–1.95 µm, so long-range SWIR observation must avoid these two ranges and use the atmospheric windows on either side (see selection criterion 03 in the band reference table). For long-range observation this is a nuisance; for close-range measurement it is a tool.
3. The third thing: many plastics are transparent
Common packaging plastics such as polyethylene and polypropylene mostly have considerable transmission in SWIR, even when they are opaque white or colored plastic in visible light. So:
- Fill level inspection of opaque bottles and containers: no unsealing, no weighing, just see where the liquid surface is
- Foreign matter detection inside sealed packages
- Interlayer defects in multilayer films
Note that "mostly" does not mean "all". Plastics containing carbon black (as many black plastic parts do) are just as opaque in SWIR, because carbon black absorbs across a broad band. So for this kind of application we always recommend shooting real samples first, before discussing specifications: two suppliers' "black plastic housings" may have formulations that give completely different results.
4. The fourth thing: seeing invisible lasers
Two laser wavelengths common in industry and communications, 1064 nm (Nd:YAG) and 1550 nm (fiber-optic communication and eye-safe ranging), both fall within the SWIR range, and both are completely invisible to the naked eye.
With a SWIR camera you can directly do:
- Beam profile measurement: the shape of the beam, its energy distribution, and whether it is symmetrical
- Optical path alignment: no more trying section by section with a fluorescent card
- Fiber end face and coupling inspection: whether light exits at the right position, and whether there is leakage
This is a simple but very practical one: turning something invisible into an image, so alignment changes from "going by experience" to "adjusting while you watch".
5. A common misunderstanding: SWIR is not thermal imaging
SWIR images reflected light and needs illumination; LWIR images self-emitted radiation and needs no illumination.
This is the point most easily confused in selection. Side by side:
| SWIR (0.9–1.7 µm) | LWIR thermal (8–14 µm) | |
|---|---|---|
| What it sees | Light reflected by the object | Radiation emitted by the object itself |
| Illumination needed? | Yes | No |
| What the image looks like | Like a black-and-white photo, with shadows and material texture | Like a temperature map, without shadows |
| Objects at room temperature | Visible only with lighting | Visible as they are |
| Detector | InGaAs | Uncooled microbolometer |
| Typical tasks | Seeing through, seeing materials, seeing invisible lasers | Seeing temperature, seeing thermal anomalies |
So "I want to see whether the object is getting hot" calls for LWIR, and "I want to see what is inside the object" calls for SWIR. Both are called infrared, but their uses hardly overlap.
The middle ground is very hot objects: for glass, metal and melts above several hundred °C, the short-wavelength tail of their emission already extends into SWIR, so a SWIR camera can image them without lighting, and because the wavelength is shorter, with much better spatial resolution than LWIR. Process monitoring in the glass industry and metal smelting often uses this.
6. When to choose SWIR
Back to the most practical question. SWIR is worth considering when:
- What you need to see is inside silicon or plastic
- What you need to distinguish is moisture content or composition, not shape
- What you need to observe is a 1064 or 1550 nm laser
- The object is very hot and needs better spatial resolution than LWIR
The cases where it is not worth it are just as clear: pure dimensional measurement, character recognition, surface scratches; visible cameras do these better and more cheaply. A SWIR camera usually costs at least an order of magnitude more than a comparable visible camera (because of InGaAs sensor processing and yield), and should not be used for problems that visible light can solve.
Closing: change the waveband, change the question
At its core, optoelectronic system integration is about "deciding which part of the electromagnetic spectrum to ask the question with". Something that is an impossible task in visible light may be a matter of switching on one lamp in SWIR; conversely, forcing a visible-light problem into SWIR only multiplies the cost tenfold.
So the judgment at the selection stage often decides success or failure more than the hardware purchasing that follows, because it decides whether the job can be done at all.
