Optical Filter Performance Characteristics

Precision optical filters illustrating spectral transmission and filter performance

Two optical filters designed for the same general wavelength region can perform very differently within an optical system. A centre wavelength or filter type alone does not define how effectively a component will transmit the required signal, suppress unwanted radiation or maintain its spectral response under operating conditions.

Filter performance is therefore described using a combination of spectral, optical and physical characteristics. Centre wavelength and bandwidth define where a passband is positioned and how wide it is. Transmission describes how much useful light reaches the system, while blocking and optical density define the degree to which unwanted wavelengths are suppressed. Angle of incidence, polarisation and temperature can alter the spectral response further, and properties such as surface quality and transmitted wavefront become important where the filter forms part of an imaging or precision optical path.

These parameters are interconnected. Improving one aspect of a filter specification can place greater demands elsewhere in the design or manufacturing process, particularly where narrow passbands, steep transitions and deep blocking are required simultaneously.

Understanding filter performance therefore means looking beyond a single wavelength value and considering how the complete component behaves within the optical system.

Reading an Optical Filter Transmission Curve

A spectral transmission curve is one of the most useful ways of describing the behaviour of an optical filter. It plots wavelength against the proportion of incident light transmitted through the component, providing a direct representation of the filter’s spectral response.

For a bandpass filter, the curve normally contains a region of relatively high transmission surrounded by areas of much lower transmission. The transmitting region is the passband, while the wavelengths intentionally suppressed outside it form the blocking regions.

The shape of this curve contains considerably more information than the filter’s nominal wavelength alone. It reveals the position and width of the passband, the transmission achieved within it, the rate at which transmission changes at each spectral edge and, provided the measurement has sufficient dynamic range, the degree of rejection outside the transmitting region.

A longpass or shortpass filter produces a different characteristic shape. Instead of a transmission band bounded by two spectral edges, the curve moves between a blocking region and a transmitting region around a defined transition. A notch filter reverses the principle, introducing a region of rejection within an otherwise transmitting spectral range.

Transmission curves should always be interpreted alongside the conditions under which the measurements were made. Angle of incidence, polarisation and temperature can all influence the response of an interference filter, so a curve measured at normal incidence does not necessarily represent its behaviour under different operating conditions.

 

Bandpass filter transmission curve showing centre wavelength, FWHM, peak transmission and blocking regions

Centre Wavelength

Centre wavelength, commonly abbreviated to CWL, describes the spectral position of a bandpass filter.

For a symmetrical passband, it may appear sufficient to identify the wavelength at the visual centre of the transmission peak. Precision filter specifications require a more clearly defined method because real spectral profiles are not necessarily perfectly symmetrical.

One common approach is to determine the wavelengths on either side of the passband at 50% of the filter’s peak transmission and calculate the midpoint between them. The resulting value establishes the centre wavelength of the passband.

The tolerance applied to CWL is important. A nominal 650 nm bandpass filter, for example, does not imply that every manufactured filter will have an identical spectral centre at exactly 650 nm. The permitted variation must be specified and controlled according to the needs of the optical system.

This becomes increasingly significant as bandwidth decreases. A small displacement of the centre wavelength may represent only a minor proportion of a broad passband but a substantial proportion of a very narrow one.

Centre wavelength should therefore be considered alongside bandwidth and manufacturing tolerance rather than treated as an isolated specification.

Bandwidth and FWHM

Bandwidth describes the spectral width of a filter’s transmission region. For bandpass filters it is commonly specified as full width at half maximum, or FWHM.

FWHM is measured between the two wavelengths at which transmission reaches 50% of the maximum transmission within the passband. The difference between those wavelengths gives the bandwidth.

If the 50% transmission points occur at 640 nm and 660 nm, for example, the FWHM is 20 nm.

This provides a repeatable method for comparing filters even where their maximum transmission levels differ.

A narrower FWHM provides greater spectral selectivity, allowing the optical system to isolate a smaller wavelength region. That can be essential in spectroscopy, fluorescence detection and other applications where the useful signal lies close to unwanted spectral information.

Narrowing the passband, however, places greater demands on coating design, manufacturing control and spectral stability. The allowable shift caused by angle, temperature or production tolerance becomes increasingly significant as the width of the transmission region decreases.

Bandwidth therefore needs to be selected according to the spectral separation actually required by the application rather than assuming that the narrowest available filter will necessarily produce the best system performance.

Transmission Through the Passband

A filter can have the correct centre wavelength and bandwidth but still perform poorly if insufficient useful light is transmitted.

Transmission describes the proportion of incident optical power that passes through the filter at a given wavelength. It is normally expressed as a percentage, although transmission can also be represented as a decimal fraction.

Peak transmission identifies the highest transmission reached within a spectral region, but this figure alone can be misleading. A filter may achieve a high value at one wavelength while providing substantially lower transmission across the remainder of the required passband.

For many systems, minimum or average transmission across a defined wavelength range provides more useful information than a single peak value.

This distinction becomes important where the detector must collect as much signal as possible. Reducing transmission within the required spectral region can lower the available signal and may require longer integration times, greater illumination or increased detector sensitivity elsewhere in the system.

High transmission is therefore desirable, but only within the context of the complete filter specification. It must be achieved while maintaining the required blocking, bandwidth and spectral stability.

Blocking and Optical Density

Transmitting the required wavelengths is only half of a wavelength-selective filter’s function. Light outside the required spectral region must also be suppressed sufficiently to prevent it interfering with the measurement.

Blocking performance describes this rejection.

For applications requiring relatively modest attenuation, percentage transmission may be adequate. At much lower transmission levels, however, percentages become inconvenient. Optical density provides a more useful logarithmic representation.

Optical density is defined as:

OD = −log10(T)

where T is transmission expressed as a decimal fraction.

Optical Density Transmission
OD 1 10%
OD 2 1%
OD 3 0.1%
OD 4 0.01%
OD 5 0.001%
OD 6 0.0001%

An OD4 blocking requirement therefore permits no more than 0.01% transmission within the specified blocking region.

The spectral range attached to an optical-density requirement is just as important as the OD value itself. Stating OD4 blocking without defining the wavelengths over which that performance must be maintained does not provide a complete specification.

This is particularly relevant when a detector responds over a much wider spectral range than the desired signal. Radiation far outside the passband may still influence the measurement if the detector remains sensitive to it and the filter does not provide sufficient rejection.

Blocking should consequently be specified according to the source, detector response and spectral environment of the complete optical system.

Cut-On and Cut-Off Wavelengths

Longpass and shortpass filters are generally characterised by a spectral transition rather than a centre wavelength and bandwidth.

For a longpass filter, the cut-on wavelength identifies the transition from the blocking region into the transmitting region. A shortpass filter uses the corresponding cut-off wavelength to describe the transition from transmission into blocking as wavelength increases.

These values need a defined reference level. A cut-on or cut-off wavelength may, for example, be specified at 50% transmission, allowing the transition point to be measured consistently.

The transition itself is not infinitely sharp. Real filters move from low transmission to high transmission across a finite wavelength interval, making the width and steepness of that transition important where neighbouring spectral regions must be separated.

A nominal edge wavelength on its own therefore provides only part of the information required to characterise an edge filter.

Edge Steepness and Transition Width

Some optical systems require the filter to move from strong blocking to high transmission within a very small wavelength interval.

This characteristic is commonly described in terms of edge steepness or transition width.

A steep edge is particularly valuable where the desired and unwanted signals lie close together spectrally. The shorter the transition between blocking and transmission, the more effectively the filter can separate those regions without sacrificing useful signal or admitting unwanted radiation.

Achieving a steep transition can significantly increase the complexity of an interference coating. The spectral requirement must also remain achievable after manufacturing tolerances, angle-of-incidence effects and environmental conditions have been considered.

For this reason, transition width should be specified according to what the optical system actually needs. Requiring an unnecessarily steep edge can add complexity without improving the final measurement.

Angle of Incidence

The spectral response of an interference filter depends on the angle at which light encounters the coating.

At normal incidence, light travels perpendicular to the filter surface. As the angle of incidence increases, the effective optical path through the multilayer structure changes. Spectral features consequently shift, generally towards shorter wavelengths.

This shift can be relatively unimportant for a broad filter but becomes significant for narrow bandpass filters or systems requiring accurate positioning of a spectral edge.

The issue is not limited to the nominal mounting angle. Converging and diverging beams contain a range of ray angles, meaning different portions of the beam can experience slightly different spectral responses even when the filter itself is mounted perpendicular to the optical axis.

For this reason, filter specification should consider the angular distribution of light within the actual optical system rather than only the mechanical orientation of the component.

 

Interference bandpass filter transmission curves showing spectral shift at different angles of incidence

Polarisation

At normal incidence, the spectral behaviour of an interference filter is generally independent of the orientation of linear polarisation. At oblique incidence, the situation becomes more complex.

Light can be resolved into s-polarised and p-polarised components relative to the plane of incidence. These components interact differently with the interfaces within a multilayer coating, causing their spectral responses to diverge as angle increases.

The result can be a difference in transmission, edge position or passband shape between the two polarisation states.

This behaviour is particularly important in optical systems operating at substantial angles of incidence, in polarisation-sensitive instruments or where a tightly controlled spectral response must be maintained.

A filter intended for oblique incidence should therefore be specified with its polarisation conditions understood rather than assuming that performance measured at normal incidence will remain unchanged.

Temperature and Spectral Stability

Optical filter performance can also change with temperature.

The refractive indices of coating and substrate materials vary with temperature, while thermal expansion alters physical layer thicknesses. Both effects can influence the optical thickness of the coating and consequently shift its spectral response.

The magnitude of the change depends on the materials, coating structure and operating wavelength. In many conventional applications it may be small enough to have little practical effect. In narrowband systems or applications exposed to large temperature ranges, it can become an important part of the filter specification.

Temperature is particularly relevant where optical equipment must operate outside controlled laboratory conditions. Aerospace, remote sensing, industrial and other demanding systems may experience operating environments very different from those under which a filter is initially characterised.

Spectral stability therefore needs to be considered over the intended operating range where wavelength position is critical to system performance.

Surface Quality, Flatness and Transmitted Wavefront

Spectral performance is not the only characteristic of an optical filter.

Because the filter occupies a physical position within the optical path, the quality and geometry of its surfaces can also affect system performance.

Surface imperfections such as scratches, digs and local defects can scatter light. In imaging systems, excessive surface irregularity or substrate distortion can degrade image quality, while transmitted wavefront error can alter the phase of light passing through the component.

Flatness requirements become particularly important where the filter is used in a collimated beam or forms part of a precision imaging or measurement system.

The required level of optical quality depends strongly on the application. A filter positioned close to a detector in a relatively low-resolution system may tolerate characteristics that would be unacceptable in an interferometric or high-resolution imaging instrument.

Specifying tighter surface and wavefront tolerances than the system requires can increase manufacturing difficulty unnecessarily. As with spectral characteristics, the appropriate requirement is determined by the optical system rather than by pursuing the most demanding specification available.

Environmental Durability

A filter that meets its spectral specification when first manufactured must continue to do so throughout its service life.

Thin-film coatings and optical substrates may be exposed to humidity, temperature cycling, abrasion, contamination, vibration and cleaning processes depending on the application. These conditions can affect both the physical integrity of the component and its optical performance if the materials and coating process are not suitable for the intended environment.

Environmental durability therefore forms part of filter performance rather than being considered separately from it.

The relevant requirements vary considerably. An optical filter installed permanently inside a sealed laboratory instrument faces a very different environment from one incorporated into an exposed aerospace or industrial optical assembly.

Materials, coating technology and mechanical design should consequently be selected according to the conditions the filter will experience in service.

From Spectral Requirement to Filter Specification

Descriptions such as 650 nm bandpass filter or IR-cut filter are useful starting points, but they are not complete engineering specifications.

A practical filter requirement needs to define enough information for the intended optical behaviour to be understood and reproduced. Centre wavelength or spectral edge may establish where the filter operates, while bandwidth, transmission and blocking define how it behaves around that region. Angle of incidence, polarisation and temperature establish the conditions under which those characteristics are expected to remain valid.

The substrate, clear aperture, dimensional tolerances and optical surface requirements then determine how the component integrates into the wider system.

Not every application requires every parameter to be tightly controlled. The important point is that the specification reflects the performance the system genuinely needs.

An optical filter should therefore be evaluated as a combination of spectral, optical, physical and environmental characteristics rather than by a single headline wavelength. It is the interaction between those characteristics that determines how the filter performs once it becomes part of a working optical system.

Related Guides

Optical Filter Performance – FAQs

Is centre wavelength the same as peak wavelength?

Not necessarily. Centre wavelength defines the spectral position of a passband using a specified method, commonly the midpoint between its 50% transmission points. Peak wavelength is simply the wavelength at which maximum transmission occurs. The two values may differ where the passband is not perfectly symmetrical.

Does a higher optical density mean better blocking?

A higher optical density represents lower transmission and therefore greater attenuation. Whether it constitutes better blocking depends on the application, however. The required OD and the wavelength range over which it must be maintained should be defined according to the unwanted radiation reaching the detector.

Does a narrower FWHM make an optical filter better?

Not inherently. A narrower bandwidth provides greater spectral selectivity but can reduce tolerance to spectral shifts and increase manufacturing demands. The appropriate FWHM is determined by the wavelength separation required by the optical system.

Why does the wavelength of an interference filter change with angle?

Changing the angle of incidence alters the effective optical path through the multilayer coating. This changes the interference conditions within the filter and generally shifts spectral features towards shorter wavelengths as the incidence angle increases.

Can temperature change optical filter performance?

Yes. Temperature can alter both the refractive indices and physical dimensions of the materials within an optical filter. These changes can shift the spectral response, particularly where narrow bandwidths or large operating temperature ranges are involved.