Many years ago, while working in industry, I developed an interest and fascination with periodic surfaces.  The surfaces that I studied were those made of metal and are known as metal mesh filters.  The term METAL MESH FILTER has been replaced by the term FREQUENCY SELECTIVE SURFACE or STRUCTURE in recent times.  These surfaces are most easily described as two dimensional amplitude diffraction gratings whose spatial period is less than the wavelength of the incident radiation.  With this condition, there are no diffracted orders other than the "zeroth" order, also sometimes referred to as the "specular" order.  Configured in this way, the periodic surface acts as a polarization dependent spectral filter.  Hence, the name 'metal mesh filter.'   Probably the most common mesh filter is the one in your microwave oven, that allows you to see the item inside the oven (due to the illumination provided by a light bulb inside the oven), but protects you from being bombarded with the microwaves that are heating your food.  This is a rather dramatic example of the mesh acting as a filter, to reflect radiation with microwave frequencies, while transmitting radiation at optical frequencies.  We are currently working to develop narrowband spectral filters, some of which are spectrally tunable.  We are also working on metal grids that act in a manner similar to "optically active" materials, in that they respond to right and left circularly polarized waves differently!

Our modeling of these structures has followed two approaches: (1) the first being a modal approach, and (2) the second an FDTD approach.  Both methods have their advantages and limitations.   The primary advantage of modal method is that it provides results in a reasonably short timeframe.  In addition, we have the ability to model multiple apertures within a unit cell of the structure.  Its disadvantage (to one way of thinking) is that it is based on the use of perfectly conducting metals.  Prediction of the optical properties of FSS's that are designed for middle and near infrared use are therefore suspect, since metals in this spectral region are definitely not perfect conductors. !  Also, in our modal method we have limited the types of apertures to either rectangular or circular.  In our FDTD software, we can model materials that are both perfect conductors and non-perfect conductors.  Along with this capability, we model the dispersive nature of both metals and dielectrics!  Hence we obtain a more accurate determination of both spectral and polarizing properties, as well as loss predictions.

Our fabrication capability is also increasing.  We have designed and built a new mount that will more accurately position our smaller substrates so that we can produce 'direct-write' e-beam fabricated devices.  We are currently able to produce 0.5 mm wide lines.  We are also improving the registration capability of our e-beam system and should be able to produce a wide variety of FSS geometries soon.

My most recent explorations concern the losses present as one makes the conducting elements smaller and smaller in size.  As would be expected the loss increases with decreasing cross section, but not at the rate that one might expect from a basic argument based on resistance as determine as calculated from (rho)((l)/A.   The reason for this is that not all of the cross sectional area is used at the frequencies typically encountered in the near infrared and visible spectral regions, resulting in an effective area that is not changing linearly with the width of the conductor.   These statements assume that classical conduction is still appropriate, and that no quantum mechanical processes are yet involved.  Hopefully, my retirement will permit me to explore this area to a greater extent than I've been able to at present!!