All About Digital UV and IR. Page 5: Filters for IR

You cannot record infrared images unless there are subjects which reflect at least some infrared light, there are source(s) of IR light illuminating these objects, and visible light is prevented from reaching the CCD sensors. The last point, not entirely intuitive, results from the fact that IR will record in CCD channels already sensitive to visible rays. So, if visible light is allowed to reach the CCD, it will literally wash out the IR image.

Because the filters are opaque, you have to focus the lens before the filter is attached. A hinged attachment that allows the filter to be swung out of the way for focusing is highly practical.

To capture invisible IR rays onto your CCD, an opaque filter capable of blocking visible light is needed. Not surprisingly, such filters appear black to the naked eye.

From left to right, the filter types are as follows:

The 'quick-and-dirty' IR solution: A cut piece of non-exposed, but processed, E-6 film. It yields quite convincingly IR images, albeit these come out slightly on the soft side. I prefer using 120-format film because it is wider than 35 mm material. Sheetfilm 4x5" is too thick unless you love really soft images. A gelatin-filter holder comes in handy for mounting the filter onto the lens.
Wratten 89B is a very dark black-red filter and probably the one best suited to most digital cameras. It produces a mild "IR look" and yet gives useful speed response to the camera. An 89B can be had mounted in glass (above) or as gelatine sheets (below).
Wratten 87 is a typical black IR filter which scarcely lets any visible light through. Even the bright sun disc will only show as a weak dullish red glow. It is perfect for masking your flash so as not to be seen when it fires, for IR candid photography. It is readily available in gelatine sheet form and can be had as resin or glass-mounted filters as well.

While a 87 is the filter to user for film-based IR, it often cuts too much of the feeble IR sensitivity of today's better dSLRs, in which the designers have done their best to remove IR response. I have read statements to the effect that the 87 filter is optimal for the Nikon D100 camera, but haven't tried to verify this myself. However, with D70, the 87 works nicely.

Even denser filters, such as the Wratten 87C or Hoya RM1000, are less suited for digital SLR systems, because of the likely very low sensitivity of the CCD sensor to longer IR wavelengths. However, some DSLRs do take on even these dense filters and I have obtained excellent IR imagery with D2H and D70 using these filters. Not surprisingly, the exposures do tend to get long, so activating a noise-reduction feature is paramount in this situation.

For a simulated "false-colour IR" effect, you can shoot through stacked polarizers. This technique, which may produce quite pleasant images, is still poorly understood as to how the colours arise. However, this shouldn't prevent anybody from trying this approach.

If you want to go into filter details, browsing the Kodak Filter Handbook may provide the insight you need. Further pointers may be found at W.J. Markerink's website.