Here are several ways to do this.
1) Edmund Optics in Barrington, NJ, USA sells a variety of filters.
Some of these are longpass filters made of Schott glass. These are high quality filters that are better than necessary for most purposes. It is advisable to choose a very long cutoff wavelength, preferably near or over 800 nanometers, since wavelengths in the 700s of nanometers are slightly visible. Wavelengths in the low-mid 800s are very slightly visible.
Among the Schott glass filters, RG-850 has the longest cutoff wavelength that is short enough to pass the wavelengths that silicon photosensors are most sensitive to.
UPDATE 7/22/2012: I just tried an experiment to see what wavelengths from nominally 940 nm high efficiency infrared LEDs are being seen when these LEDs are visible. My results: Generally 780-900 nm, with the peak product of spectral power distribution and spectral sensitivity of my eyes being around 830-840 nm. As a result, I expect a Schott RG850 filter to reduce visibility of these LEDs significantly, but not completely make them invisible.
Link to Edmund Optics for Schott glass longpass filters.
Edmund also sells more general purpose visible-blocking IR filters.
So do some surplus dealers that probably advertise in magazines catering to
electronics hobbyists and photographers.
2) Use a combination of one red, one blue, and one or two green acrylic sheets such as "plexiglas". These should be the usual fairly deep red, green, and blue colors. Epoxy can be used between sheets to reduce reflection losses. Wavelengths shorter than approx. 760 nm are largely blocked.
3) Use a combination of deep red, deep green, and deep blue stage lighting filter gels. Adding deep blue-green to this and/or using two layers of each color makes the filter darker, but less efficient. Stage lighting filter gels of nearly all brands, colors, etc. pass IR.
Approx. Wavelength nm
Combination for 50 % for 90 %
blocking blocking
---------------------------------------------------------------------------
Edmund Scientific:
#823 (med.red) + #874 (med. grn) + #877 (med. blu. grn.) 765 740
Roscolux:
Two #19 ("fire") + #83 (med. blue) + #90 (dk. yel. grn.) 740 720
GamColor:
Two #250 (med. red) + #690 (bluegrass) + #850 (pri. blue) 730 710
----------------------------------------------------------------------------
3. Wratten filters of numbers 87, 87C, and 88A block visible light and pass IR. The 87C is the most visible-blocking of these, with almost complete blocking of wavelengths below 800 nM, and some significant absorption even into the mid 800s. I have seen a mention of less common but even "darker" (more extreme) 87B and 87A.
Wratten filters are available at some photographic supply stores and some scientific supply stores.
4. Someone out there reminded me of the color film trick, and gave me some specifics. Developing unexposed color slide film produces a filter that passes infrared and blocks visible. He mentions Ectachrome E6 film and claims optical density figures of 3 for visible (99.9 percent absorption) and .3 for 780 nM (50 percent absorption).
I have seen a site saying that two layers of such film rather than one approximates an 87C filter:
Making an Improvised Infrared Transmitting Filter
5. Hoya filters, such as RM90 and RM100.
More IR and IR filter info at other sites:
IR and red filter info at Willem-Jan Markerink's photo site (via the Wayback Machine)
The Infra-Red FAQ v1.10 (96/10/19)
Bill Beaty's Page for low cost IR-pass, visible-block filters.
Do not look at the sun through a dark IR-pass and/or UV-pass filter. Doing so can burn a spot on the retina, and that can cause a permanent blind spot. IR doing such damage may give no warning signs before a spot on your retina gets permanently toasted.
As for what to use with xenon flashlamps in order to minimize obtrusiveness of flashing while there is visible ambient light: The shortest prominent wavelength of xenon's strong cluster of near-infrared wavelengths is 823 nm. About 30-40% of the photographically useful content in xenon's strong near-IR lines is in the 820-845 nm range.
Hoya IR-80 and Schott RG-780 are the most restrictive high quality glass filters that easily pass 823 nm. A xenon flash filtered by these is generally visible to persons looking at it - merely unobtrusive. Schott RG-830 will greatly reduce the visibility but not make the flash completely invisible. Combinations of stage lighting filter gels or acrylic sheets cost less than glass filters and are sufficient to make a xenon flash unobtrusive.
The glass used for the outer bulbs of low pressure sodium lamps is also
IR-reflecting. It is coated with a special grade and thickness of either
tin oxide or indium oxide in order to do this.
If you decide to break or disassemble a low pressure sodium lamp, please beware
that the bulb contains a vacuum and could violently implode if broken. The
inner arc tube contains a near-vacuum and can also implode if broken. The inner
arc tube may also contain substantial amounts of sodium, which is hazardous.
Sodium quickly forms sodium hydroxide upon exposure to moisture or humidity. If
exposed to very dry air, sodium oxide forms instead, which becomes sodium
hydroxide when exposed to moisture or humidity. Sodium hydroxide is corrosive
and hazardous to most living tissue.
Sodium reacts very vigorously with water, forming both sodium hydroxide and
hydrogen gas. Violent spontaneous combustion has been known to occur from
this.
Break sodium lamps with appropriate precautions and only at your own risk.
If some absorbtion at visible wavelengths is tolerable, then there may be a solution.
This is often done with filters made of Schott KG series glass, or Hoya HA-30 glass. The four usual Schott ones are KG-1, KG2, KG3, and KG-5.
KG-2 is the least absorbing of IR and visible red light. Otherwise, higher numbers in this series indicate increased absorption of IR and visible red.
Edmund Scientific sells Schott KG-1, KG-3 and KG-5 filters in 3 mm thickness.
As for other options that *may* cost less than glass shortpass filters: One option is solutions of copper compounds in water - these largely absorb most infrared. Copper compounds must be of the more common "cupric" or "copper II" valence, not "cuprous" or "copper I". These have a very gradual cutoff, and absorb visible red light greatly, yellow and orange significantly, and yellow-green slightly to moderately.
The filters used in acetylene welding goggles absorb IR more than visible, especially if they are glass rather than plastic. Green glass sunglasses are a milder version of this, and offer some infrared absorbtion.
To pass a specific band of wavelengths and greatly block (mostly reflect)
just about everything else, use dielectric interference bandpass filters.
They are very effective. However, they are usually quite expensive, and the
passband shifts ton shorter wavelengths and often gets narrower and often gets
less transparent for light rays hitting the filter in a non-perpendicular manner.
These filters are available from Edmund Scientific.
A few of these are sometimes available from a few surplus dealers. Check the
ads in magazines related to electronics and photography.
Copyright (C) 2012, 2013 Donald L. Klipstein.
Please read my Copyright and authorship info.
Please read my Disclaimer.