A. Why new filters?

A set of broad-band and narrow-band infrared filters was required for use with the 8.2-m Subaru Telescope and the 8.0-m Gemini North Telescope. The required filters were large (up to 60 mm in diameter) and had to suitable for use with adaptive optics. Recognizing the opportunity to define an improved filter set as well as to reduce the cost per filter by organizing a group purchase of filters, it was decided to form a filter purchase consortium.

One of the main objectives in defining the filter bandpasses was to reduce the effects of the atmosphere as much as possible while retaining as high a throughput as possible. Another objective was to minimize the color transformations between a high altitude site such as Mauna Kea to a lower altitude site. An analysis of the optimum filter bandpasses was performed by Doug Simons (Gemini Telescopes) and it can be viewed at: Gemini Technical Note.

Alan Tokunaga (IFA, Univ. of Hawaii) undertook the responsibility of organizing a consortium of instrument groups who were interested in participating in a group purchase of filters as well as to identify vendors who were willing to build filters to our specifications. For reasons listed below this is the best way to reduce the cost per filter.

Another very important reason for considering consortium purchases of filters is that the more users there are for a specific filter set, the more likely that a common infrared filter set can be established. This will lead to a greater degree of commonality in infrared photometric systems and to lower systematic errors in transforming from one system to another.

B. Why are filter fabrications costs so high?

It is surprisingly difficult to find filter companies who are willing to make filters to our specifications. The reasons are:

(1) Filter companies are typically not interested in small volume production runs since the profit margin is very small.

(2) Astronomical filters are very demanding, requiring sharp edges on the transmission curve and very low out of band transmission, typically <0.0001. For the latter the control of surface defects (such as pinholes) is critical. (3) We require the filters to be suitable for use with adaptive optics systems. The wave-front error tolerance is very small for filters to be used with adaptive optics, and this requires the use of very flat single substrates.

(3) Infrared filters have to be cooled, and this requires precise prediction of the filter wavelength as function of temperature. In addition, it is possible for the filters to delaminate if proper precautions are not taken during the coating and cutting to size.

(4) Instruments for 8-m class telescopes require large filters (up to 60 mm in diameter or larger).

Typical production runs for astronomical filters with our specifications can run as high as $20,000 to $30,000. Since many filters can be made at once with little additional cost, there is a significant cost savings if many groups could share in a single production run.

C. The Mauna Kea Observatory Near-Infrared (MKO-NIR) Filter Set.

A new set of J, H, K, Ks, K', L', M' broad-band photometric filters was fabricated in 1998 by OCLI (Santa Rosa, California), following the specifications defined by D. Simons in his Gemini technical note. A consortium consisting of about 30 members ordered these filters. These included the following observatories and institutions: Anglo-Australian Observatory, California Institute of Technology, Cambridge Univ., Canada-France-Hawaii Telescope, Carnegie Institution, Center for Astrophysics, Cornell Univ., European Southern Observatory, Gemini Telescopes, Korean Astronomy Observatory, MPI-Heidelberg, MPE-Garching, NASA Infrared Telescope Facility, NASA Goddard Space Flight Center, Kiso Observatory, National Optical Astronomy Observatories, National Astronomical Observatory of Japan, Nordic Optical Telescope, Ohio State Univ., Osservatorio Astrofisico di Arcetri, Subaru Telescope, United Kingdom Infrared Telescope, Univ. of Grenoble, Univ. of Hawaii 2.2-m Telescope, Univ. of California Berkeley, Univ. of California Los Angeles, Univ. of Kyoto, Univ. of Wyoming, and the William Herschel Telescope.

We denote these filters as the Mauna Kea Observatory Near-Infrared (MKO-NIR) filter set. These filters have been adopted by the NASA IRTF, UKIRT, Gemini Telescopes, Subaru Telescope, and the UH 88-inch Telescope. As noted above and in the Gemini technical note, these filters have been designed to be useful at lower altitudes as well.

Standard star magnitudes and color transformations obtained at the UKIRT with these filters are given at: UKIRT faint standards. Color transformations obtained at the IRTF are given at: Characterization of New Mauna Kea IR Filter Set .

The specified center, cut-on, and cut-off filter wavelengths (micrometers) are given below. The cut-on and cut-off are the wavelengths where the transmission is 50% of the peak.

The filter profiles may be obtained at: New Mauna Kea Filters The manufacturing specifications of the filters were designed to allow the use of these filters with adaptive optics. Therefore careful attention was given to the substrate flatness. A detailed list of the manufacturing specifications are given below:

1. Out of band transmission <0.0001 out to 5.6 micrometers.

2. All parameters for 65 K; cold filter scans of witness samples to be provided with prediction of wavelength shift with temperature.(*)

3. >80% average transmission (goal >90%).

4. Peak transmission level of broad band filters to have a ripple of less than ±5% of average transmission between 80% points.

5. Cut-on and cut-off: ±0.5%.

6. Roll-off spec: %slope less than or equal to 2.5%.

7. Substrate flatness <0.0138*lambda/(n-1), where n is the index of refraction of the substrate (for compatibility with AO systems). For example, for n=1.5, lambda=2200 nm, the substrate flatness should be <61 nm. For n=3.4, lambda=2200 nm, the substrate flatness should be less than 13 nm.(**)

8. Scratch/Dig: 40/20.

9. Substrate surfaces parallel to 5 arcsec or better to suppress ghost images.

10. Filter to be designed for a tilt of 5 degrees to suppress ghost images.

11. Diameter: 60 mm, or cut to requested size. Maximum useable size is 54 mm.

12. Maximum thickness: 5 mm.

13. Single substrate filters (cemented filters are not acceptable due to deformation when cooled).

14. No radioactive materials such as thorium to be used (may cause spurious noise spikes).

1. (*) Filters are expected to be used in the 50-77 K temperature range.

2. (**) This is a flatness spec, not a roughness spec.

3. Slope is defined to be: [ lambda(80%)-lambda(5%) ] / lambda(5%).

D. Narrow-band, broad-band, and special filters.

A set of narrow-band, broad-band, and special filters were also required for adaptive optics use on large telescopes. These filters are being produced by NDC Infrared Engineering, Essex, UK. The specified center, cut-on, and cut-off filter wavelengths (micrometers) are given below.

Note:

1. "HK Notch" is a 1.47-2.4 micrometer filter with notch at 1.8 micrometer that blocks out the atmospheric water band. This filter is designed for deep imaging. The half-power points are at 1.48, 1.80, 1.98, and 2.30 micrometers.

1. For narrow-band filters: 65% peak transmission (goal >70%). Bandwidth less than or equal to 1.5%.

2. For broad-band filters and special filters: >80% average transmission (goal >90%) .

3. Half-power points: ±0.2% for narrow-bands; ±0.5% for broad-bands; ±1.0% for special filters.

4. Roll-off spec: <0.5% slope for narrow-bands; <2.5% for broad-bands; <3.0% for special filters.

5. Substrate flatness prior to coating is <= lambda/10 peak-to-valley at 0.63 micrometers on a best effort basis.

A special long-wavelength blocker was also specified to suppress long-wavelength leaks. It was anticipated that achieving the out-of band blocking would prove to be very difficult, and this was the case. After delivery of the initial set of filters, it was noticed that some of the filters had long wavelength leaks near 4 micrometers. These filters should be used with a long wavelength blocker to ensure that the filter is completed blocked.

E. Order-sorting filters.

The manufacturing specifications are given below:

1. Area used: 92% clear aperture (For example, for the 25.4 mm filter the area used will be 23.4 mm.)

2. Thickness: all filters for a given project to be parfocal (same optical thickness). Therefore if the fused silica filter is 5 mm, then the Si substrate is 2.1 mm etc.

3. Thickness tolerance: fused silica ±40 micrometers; Ge, Si ±15 micrometers; other materials according to index.

4. Surface curvature <1.5 fringe for fused silica, <0.25 fringe for Ge, <0.35 fringe for Si, etc.

5. Irregularity peak-to-valley <1/2 wave (6328 Angstrom) for fused silica, <1/4 wave for Ge, Si, etc.

6. Filter properties specified at 77K.

7. Peak transmission: >80% .

8. 50% transmission tolerance: ±1%.

9. Out-of-band blocking: <0.01% (0.3-5.6 micrometers).

10. Roll-off tolerance: [lambda(90%)-lambda(10%)]/lambda(50%) <2%.

11. Free of pinhole defects.

12. Single substrate is greatly preferred. If in two pieces, no cement should be used and combined filter must meet tolerances.

Note:

1. The adopted optical thickness can be adjusted if it helps in manufacture or in meeting specs, provided all filters are not thicker than specified and all filters can be handled with reasonable precautions.

F. Vendor information.

1. For information on any remaining narrow-band, special filters, or order-sorting filters, contact:

NDC IR Engineering
Essex, UK
Contact:  Henry Orr
email: hjborr@ireng.com
phone: 44-1621-852244
fax:   44-1621-856180

2. Other vendors who also produce astronomical quality infrared filters:

Barr Associates
Waltham, Massachussetts
Contact: Dale Taylor
Work: 1-978-692-7513
Fax:  1-978-692-7443
email: barr@barrassociates.com

REOSC
St. Pierre du Perray, France
Contact: Amaury Le Jemtel
Work: 33 1 69 89 76 87
Fax:  33 1 69 89 76 69
email: reosc-ve@sfim.fr