North Carolina Headlines

A friend of mine in gas detection once told me a story. Their sensor module had no accuracy issues on its own—until they swapped in a different filter. The zero-point drift immediately went off by 15%. Two weeks of troubleshooting later, they traced it back to the root cause: the filter’s center wavelength was off by 3nm. Three nanometers. A difference invisible to the naked eye—and the entire production line’s detection data was rendered useless.

An infrared filter is exactly that kind of component. Unassuming, a small disc, thin as a coin. But it sits right at the throat of the infrared sensing chain, determining what your sensor can “see” and what it cannot.

This article walks through the principles, classifications, key parameters, substrate materials, application scenarios, and selection logic of infrared filters—from start to finish.

What Does an Infrared Filter Actually Do?

Put simply, it’s a spectral gatekeeper.

The infrared spectrum is broad, spanning from near-infrared at 0.75μm all the way to far-infrared at tens of microns. But most applications don’t need such a wide spectral range—in many cases, they only require a very narrow slice of it. The filter’s job is to block the unwanted bands and transmit only what’s needed.

There are two technical approaches. One relies on the material’s inherent absorption characteristics—this is called an absorptive filter. Silicon, for instance, transmits in the 1.1–7μm range and absorbs beyond it, determined by the material’s physical properties. The other approach is based on optical interference: depositing dozens or even hundreds of dielectric layers on a substrate, precisely controlling the thickness of each layer to create constructive interference at specific wavelengths (enhancing transmission) and destructive interference at others (blocking them). This is the interference filter—and the dominant technical route in industrial applications.

The accuracy gap between the two is significant. Absorptive filters are constrained by material properties, offering little flexibility. Interference filters are far more adaptable—by adjusting layer materials, layer count, and thickness, center wavelengths can be designed to within ±1nm precision, and bandwidths can be narrowed to tens of nanometers or even single digits. Most infrared narrowband filters on the market today follow the interference route.

Four Types, Each with Its Own Role

By function, infrared filters fall into four main categories.

Bandpass filters transmit only a specific wavelength band, blocking both below and above. They are most commonly used in gas detection: CO₂ at 4.26μm, CH₄ at 3.3μm—a narrow window opened exactly at the target gas’s characteristic absorption peak. Take MULTI IR’s 4260nm narrowband filter: center wavelength 4260±42nm, FWHM available in 180±20nm and 90±20nm, peak transmittance ≥80%, and out-of-band average transmittance from UV to 11μm controlled to <0.1%. What does this blocking depth mean? It means that virtually all radiation outside the 4.26μm band is blocked out. In multi-component gas detection, if the out-of-band blocking of the CO₂ channel and CH₄ channel filters isn’t deep enough, the two channels will cross-talk, and detection accuracy collapses.

Longpass filters transmit long wavelengths and block short ones. They are most common in thermal imaging systems. MULTI IR’s LP5500nm longpass filter has a 5% cutoff wavelength at 5.5±0.4μm, an effective transmission range of 7.5–13.5μm, average transmittance ≥70%, and blocking ≤0.5% over the 0.4–5μm band. This specification corresponds exactly to the 8–14μm atmospheric window—the core working band of infrared thermal imaging. It blocks visible and near-infrared light entirely, allowing only thermal radiation information to reach the detector.

Shortpass filters do the opposite: transmit short wavelengths, block long ones. They are used less frequently but are indispensable in certain near-infrared applications.

Narrowband filters are essentially extreme versions of bandpass filters—with exceptionally narrow bandwidths, FWHM as tight as 20–50nm or even narrower. Their purpose is to extract light with extreme precision. MULTI IR’s narrowband product line covers the full spectrum from 850nm to 16000nm, including a dozen center wavelength specifications: 850nm, 1653nm, 1940nm, 3400nm, 3900nm, 4260nm, 6300nm, 7300nm, 10600nm, and more. For the 850nm model, the passband is 837–862nm with average transmittance ≥95% (minimum ≥92%), and out-of-band average transmittance <0.2% over both 400–820nm and 878–1100nm. This is extremely clean spectral performance for facial recognition and near-infrared vision applications.

In practice, bandpass and narrowband filters are the most widely used. Most infrared sensing applications require precise locking onto a specific wavelength band—not a blanket “collect everything.”

Five Core Parameters—None to Be Taken Lightly

Filter selection inevitably comes down to a few core parameters, each tied to real-world pitfalls.

Center Wavelength (CWL) —This defines who gets through. Choose wrong, and nothing else matters. For CO₂ detection, you need 4.26μm; for CH₄, 3.3μm. These wavelengths correspond to molecular vibrational-rotational absorption peaks—miss by a bit, and the signal drops. We’ve encountered cases where a customer used a 4.20μm part instead of 4.26μm, and the signal strength dropped by a full 40%. Molecular absorption peaks are sharp—drift away, and they simply won’t recognize it.

Full Width at Half Maximum (FWHM) —This defines how wide the filter’s “aperture” is. For gas detection, CO₂’s absorption peak around 4.26μm has a certain width—180nm bandwidth is generally sufficient. But if you’re doing isotopic analysis or high-resolution spectroscopy, you might need 90nm or even narrower. Narrower bandwidth means better interference rejection, but also greater manufacturing difficulty, often at the cost of transmittance. MULTI IR offers the 4260nm filter in both 180nm and 90nm FWHM variants, giving customers a choice based on their precision requirements. It’s a trade-off—narrower isn’t always better; it depends on how fine a spectral selection your application actually demands.

Blocking Depth (OD Value) —This indicates how effectively the filter blocks out-of-band wavelengths. OD4 means transmittance below 0.01%, OD6 below 0.0001%. Many engineers focus on CWL and bandwidth while treating blocking depth as a casual “deep enough.” Then in real-world applications with strong background radiation, stray light floods in and SNR collapses. In industrial high-temperature thermometry, a one-order-of-magnitude difference in blocking depth can mean measurement deviations of dozens of degrees. MULTI IR’s 4260nm filter achieves out-of-band average transmittance <0.1%—roughly OD3 level—with even higher values near the peak. In multi-component gas detection systems, this blocking depth directly determines the level of crosstalk between channels.

Peak Transmittance —How much of the target band actually gets through. Ideally 100%, but impossible in practice. Multilayer coatings introduce some loss—typically, interference filters achieve 85%–95% peak transmittance. MULTI IR’s 850nm narrowband filter averages ≥95%, a very high level in the near-infrared range. Near-infrared facial recognition and structured light applications demand high signal strength—every percentage point of transmittance directly improves SNR at the detector end. The remaining loss comes from layer reflection and absorption, which should be accounted for in system-level signal budgeting.

Angle of Incidence —This is easy to overlook but frequently trips up real applications. When light hits the filter at an angle, the center wavelength shifts toward shorter wavelengths, approximated by λ(θ) = λ₀ × √(1 – sin²θ/n²), where n is the effective refractive index of the coating. What does this mean? If your system has a large field of view, the light at the edge of the field enters at an angle, and the actual effective center wavelength differs from the center of the field. Spectral response becomes non-uniform across the field. The coating design must factor in the actual angle of incidence used by the customer—pre-compensation is necessary. This is knowledge built from years of experience, not something you find in a textbook.

Substrate Materials: The Invisible Foundation

The performance ceiling of a filter depends heavily on the substrate material. Different substrates determine the working band, temperature tolerance, and environmental stability.

Silicon (Si) is the most common choice. It offers high transmittance in the 1.2–7μm band, low cost, decent hardness, and good chemical stability. For near-infrared to mid-infrared applications, silicon covers most needs: conventional thermometry, gas detection, near-infrared facial recognition—silicon substrates are sufficient. MULTI IR’s 850nm, 1940nm, 3400nm, 3900nm, and 4260nm filters all use silicon as the default substrate.

Germanium (Ge) has a broader transmission range, covering 2–14μm. Its high refractive index (~4.0) and stable performance after coating make it the primary substrate for the 8–14μm atmospheric window—the core band for thermal imaging. MULTI IR’s BP10600nm bandpass filter uses a germanium substrate for automotive electronics applications. That said, germanium is expensive, and it becomes opaque above ~15μm—a limitation worth noting in certain far-infrared applications.

Chalcogenide glass has gained increasing attention in the mid-to-far-infrared range in recent years. Its transmission range extends to 12μm and beyond, and it can be molded into complex shapes like aspheres—reducing the cost of traditional grinding and polishing. However, its hardness is low and it’s sensitive to humidity, requiring careful handling. MULTI IR’s early investment in chalcogenide glass-based filter products aligns with its lead role in drafting the infrared filter industry standard—the substrate materials and performance specs referenced in the standard are largely defined by the technical capabilities of the industry’s leading players.

Other materials include Calcium Fluoride (CaF₂), which transmits from 0.13–10μm with excellent chemical stability and resistance to acids and alkalis, suitable for harsh environments; Sapphire (Al₂O₃), with a transmission range of 0.15–5.5μm and extreme hardness (Mohs 9), ideal for window applications requiring wear protection; and Zinc Selenide (ZnSe), covering 0.5–20μm, a classic material for CO₂ laser systems (10.6μm)—MULTI IR’s 10600nm narrowband filter (CWL 10600±100nm, FWHM 190±20nm, Tpeak≥65%) targets industrial thermometry and laser applications at this wavelength. There is no universal substrate—the choice must be driven by your actual working band, environmental conditions, and cost constraints.

Where Are These Filters Used?

Gas detection is one of the most demanding application scenarios for filter precision. The core of an NDIR (Non-Dispersive Infrared) system is the narrowband filter—its bandwidth and blocking depth directly determine detection selectivity and accuracy. CO₂ detection uses 4260nm, CH₄ uses 3400nm, CO uses around 4600nm—each gas has its own “fingerprint wavelength,” and the filter’s job is to extract that fingerprint precisely. Demand in environmental monitoring, industrial safety, indoor air quality sensing, ventilation systems, and agricultural greenhouses has been rising steadily over the years.

Thermal imaging is another major application—power line inspection for hot spots, building diagnostics for insulation defects, medical screening for temperature anomalies. Behind all these scenarios are longpass or bandpass filters operating in the 8–14μm band, blocking visible and near-infrared light to let only thermal radiation information reach the detector. MULTI IR’s LP5500nm longpass filter is specifically designed for this band, with a germanium substrate and a 7.5–13.5μm transmission window—directly compatible with mainstream uncooled infrared detectors.

In facial recognition and near-infrared vision, while precision requirements aren’t as extreme as gas detection, volumes are massive—access control, time clocks, smart locks, phone unlocking—all use it. 850nm and 940nm are the mainstream bands. 940nm offers better rejection of solar interference, making it increasingly preferred for outdoor applications. This evolution itself reflects the interplay between filter characteristics and application scenarios. MULTI IR’s 850nm narrowband filter is a volume leader in this space—passband 837–862nm, average transmittance ≥95%, out-of-band blocking Tavg<0.2%, placing it in the top tier of spectral cleanliness.

Industrial thermometry has its own set of requirements. High-temperature applications (steelmaking, petrochemicals) often operate in the short-wave infrared—1μm, 1.6μm, 3.43μm, 3.9μm—all common bands. MULTI IR distributes EXERGEN non-contact infrared temperature sensors with accuracy as high as ±0.01°C, covering temperatures from below zero to hundreds of degrees, with paired filters deployed across medical equipment, hot-bending machines, printing presses, and more.

Emerging applications continue to appear—automotive infrared night vision for autonomous driving, hot-spot detection in photovoltaic panels, temperature monitoring in lithium battery production lines, infrared obstacle avoidance for drones. Every new use case brings with it a new demand for specific-band filters.

How to Actually Select the Right Filter

A few practical guidelines:

First, define your working band clearly. Are you using the 3–5μm or 8–14μm atmospheric window for thermometry? Which absorption peak are you locking for gas detection? 850nm or 940nm for facial recognition? Once the target wavelength is set, the filter’s core parameters become anchored.

Then look at the environment. At high temperatures, thermal expansion of coating layers can shift the center wavelength—thermal stability varies widely between different coating materials. High humidity affects certain substrate materials—chalcogenide glass, for instance, is far more sensitive than silicon or germanium. These aren’t lab-only concerns; real-world environments are often far harsher than controlled conditions.

Batch-to-batch consistency becomes especially critical at volume production. Your lab samples may perform well, but that doesn’t guarantee each piece will meet spec at production volumes. In multi-channel gas detection systems, each channel has its own filter—if CWL variations are too wide, each channel requires different calibration parameters, and production line calibration costs double. MULTI IR maintains a CWL tolerance of ±2nm in volume production—a level that stands up well in the industry. As the lead drafter of the industry standard, if the standard includes specs you can’t meet yourself, that standard would be a joke.

And one more thing: don’t just look at the unit price. The filter’s cost should be evaluated at the system level. A cheap filter that drops your system yield by 5% or doubles calibration time is far more expensive in the long run.

An infrared filter may account for less than 5% of the total cost of an infrared system. But it determines what the system can “see”—and what it cannot.

Choosing between CWL and bandwidth, transmittance and blocking depth, silicon and germanium—every choice involves trade-offs. There’s no “best filter”—only the one that best fits your scenario.

The one rule of filter selection: “good enough” is a trap. Off by 3nm, you lose 40% signal. Once you’ve been through that, you never cut corners again.

MULTI IR – National-level “Little Giant” Specialized and Sophisticated Enterprise. Over 10,000 types of infrared sensitive components in stock. Top 3 globally in comprehensive strength. Product portfolio covers infrared filters, optical coatings, infrared sensors, and other core categories. Lead drafter of the infrared filter industry standard.

Website: www.mirhz.com | Global Site: www.miroptech.com

Media Contact
Company Name: HANGZHOU MULTI IR TECHNOLOGY CO., LTD.
Email: Send Email
Country: China
Website: https://www.miroptech.com/