Evanescent Wave Optical Filter

[Enthalpy]

Hello dear friends !

I conceived the following optical filter for improved stopband attenuation. I haven't found it described after a short search; an optics specialist (I'm not) could better tell if this filter is already known.

 

Like an interferential or dichroic filter, it comprises thin layers of varied index over a substrate, so the layers aren't to scale on the sketch, nor are all reflections shown. As an original feature, some layers are coupled by fading waves resulting from the varied refraction indices and from light's angle. The fading layer attenuates light but is thin enough to leave some through, and within the passband only, the following resonating layer(s) restores the amplitude.

Fibre optics uses similar filters on chips where fadings waves couple open or race tracks to make bandpass filters, used for instance for wavelength multiplexing in datacomms - but here light is not in a fibre. The frequency behaviour resembles also microwave filters made of coupled lines with their excellent stopband attenuation, and whose theory can be picked, thanks.

The fading wave can attenuate strongly, provided resonator(s) near enough restore the amplitude. This improves over a dichroic filter where attenuation results from destructive interferences, whose amplitude must match precisely, and are difficult to achieve over a broad frequency band. The fading wave filter has naturally a strong attenuation everywhere outside the passband.

This filter is naturally a bandpass, though it can be made wide enough to act as a high or lowpass over the frequencies of interest. The "reject" output would be a bandstop, but without the naturally high attenuation. Longer waves also pass more easily the fading thickness, enabling a lowpass of limited selectivity. Only the available materials and feasible thickness limit the frequencies, which include radiowaves, THz, IR, visible, UV.

The resonating layer(s) can be thick enough to show several resonances, to increase the selectivity when only one resonance falls within the frequencies of interest, or because several resonances are wanted; in an extreme case, this comb filter stabilizes short repetitive pulses, and its strong outband attenuation can help build the macroscopic equivalent of a ring laser, recover a clock signal...

A big external Q-factor at the resonating layer is often undesired: it needs precise thickness, angle, frequency, broadens the transmitted ray, can induce loss and a reaction time... But as flexibly as in electronics, several resonator layers (the sketch has only one) in a multipole filter answer that, to combine a big attenuation with a broader passband, as the resonators are more strongly coupled; the superior optical materials ease that.

The angle influences the attenuation, especially near the limiting angle, and the passband frequencies. Depending on the use, it's a drawback or an advantage, to tune the filter a bit. Optics can often be split, as in Lyot's coronograph, to have parallel rays for filters at some portion of the path. The filter can also split light in several bands using a uniform layer thickness if light bounces at varied angles, for instance because the first propagating layer is a wedge; for losing very little power, it looks nice at wavelength demultiplexing, and can losslessly multiplex as well, even broad strong rays.

With an easily tailored passband, and a strong outband attenuation to reduce the detector noise, this filter has uses even where light doesn't travel in fibres. Impossible to imagine them all: free-space Internet among buildings, communications with and between spacecraft, chemical analysis (methane on Mars!), search for and images of extrasolar planets...

Marc Schaefer, aka Enthalpy

[John Cuthber]

You need to say what "fading waves" are.

[Enthalpy]

My bad! Mistaken translation, I meant "evanescent wave", this one:

http://en.wikipedia.org/wiki/Evanescent_wave

Mea maxima culpa!

 

And I can't even improve the thread's title...

Edited April 20, 2013 by Enthalpy

[Enthalpy]

Known methods produce thin films on a substrate. Putting a thicker part over thin layers is also known, even with optical quality: glueing, ceramic-to-ceramic seals... The thicker part could have steps like a Fresnel lens to be less thick, but then phase isn't consistent among the faces. And if light must hit semiconductors, these can be put directly at the thin layers.

If the filter must be short but cover a wide light path, it may comprise several segments stacked, in parallel or opposing directions, or possibly conical like a Fresnel lens - if the stopband light is extracted or absorbed properly.

Two or more filters of opposite angles can make the center frequency less dependent of light's angle, but the frequency response will still be distorted.

Depending on the frequency and thickness, some layers can consist of vacuum, a gas like air, foam... Some materials change their thickness or index with the electric or magnetic field, which is a way to tune the filter, while most materials respond to strain and temperature, providing other ways.

A strong stopband attenuation needs to minimize stray light, which includes internal reflections. The substrate's shape can help by demanding many reflections, each one made faint, before stray light risks to exit in a direction not desired; or, possibly in addition to that, the substrate's surface itself can be made to absorb light - even before jumping to a different medium, which always leaves some reflection. Radioactivity is known to make glass brown; implantation, possibly by plasma, can damage glass or bring extrinsic colour centers faster and better than radioactivity but keep a smooth transition with the transparent part of the substrate, thus preventing reflections. Diffusion of impurities from the surface would do the same, as well as deep diffusion bonding with a coloured version of the substrate's material. I'd first make all faces opaque at the substrate, and later cut and polish the faces at light input and at the thin layers. Such non-reflective faces can have uses beyond this filter.

 

-----

 

Many thanks for the corrected title!

Marc Schaefer, aka Enthalpy