2.1.

Design of the interference filter

The design principle of the filter is a double sided coating on a fused silica Corning 7980 grade 1F substrate. On the first side of the substrate the coating has been designed as a 5-cavity Fabry-Perot (FP) bandpass design with 82 alternating layers of Nb₂O₅ and SiO₂. The basic building blocks of the FP-design are classical third and fourth order dielectric mirrors and first and second order resonator layers with high refractive index layers in Nb₂O₅:

(HL)³ 4H (LH)³ L (HL)⁴ 2H (LH)⁴ L (HL)⁴ 4H (LH)⁴ L (HL)⁴ 2H (LH)⁴ L (HL)³ 4H (LH)³ L

With H: quarter wave optical thickness Nb₂O₅ and L: quarter wave optical thickness SiO₂

To tune the full width at half maximum (FWHM) of the filter to 11nm selected mirror layers were modified to 3 quarter wave thickness layers. The nominal central wavelength is 1254nm. For a good antireflection of the filter’s passband, the thickness of the last 2 layers was optimized for maximum transmission in the passband. Figure 4 shows the theoretical design of the 5-cavity bandpass with FWHM 10.35nm for collimated light at an angle of incidence perpendicular to the filter surface.

Fig. 4

a) and b) show the calculated design curve of the bandpass coating on the first side of the substrate. The full width at half maximum of the theoretical design is 10.35nm and the slope is 1.69nm from T=5% to T=80%.

A blocking filter was coated on the second side of the substrate to achieve a blocking of optical density 5 (T=10^-5) in the range 1050-1650nm. This coating was also designed by the Fabry-Perot principle with 13 coupled resonators and 52 layers:

(H 2L HL)³ (H 4L HL) (H 2L HL)⁵ (H 4L HL) (H 2L HL)³ @ nominal design wavelength 1254nm

With H: quarter wave optical thickness Nb₂O₅ and L: quarter wave optical thickness SiO₂

The thicknesses of the last two layers of the blocking filter were optimized for maximum transmission of the filter in the passband of the bandpass. Figure 5 shows the theoretical transmission curve of the blocking filter (black solid curve) and the theoretical transmission curve of the bandpass coating (brown discontinued curve).

Fig. 5

shows the calculated design curve of the blockfilter coating on the second side of the substrate (black curve) and the theoretical design of the narrow bandpass on the first side (brown discontinued curve).

2.2.

Production of the bandpass filter

The deposition technique of the two filter layersystems have been chosen to be plasma assisted reactive magnetron sputtering (PARMS). The PARMS 0 process results in very dense and humidity resistant coatings with a very low temperature shift. Similar filters were measured to have a temperature shift from room temperature to 77°K of -0.125% of the central wavelength, which accounts for 1254nm to a shift of -1.57 nm. This temperature shift was taken into account at the centering of the filter during its production. The mechanical and environmental resistance of the coatings are excellent due to the high density of the metal oxide films. The coatings resist cleaning with alcohol or acetone.

The coating process stability and particularly the thickness monitoring is of highest importance for the production of high quality bandpass filters during the coating process. In production practice a combination of direct monochromatic [3] and rotation monitoring was proven to be very accurate and reproducible. The direct monochromatic transmission monitoring provides a self-compensation of layer thickness errors, when the quarterwave optical thickness layers are terminated always in the maximum or minimum points of the transmission measured at the centerwavelength of the Fabry-Perot filter. Thereby, should a previous layer be to thick or too thin, the next layer which is stopped again at the next extremum of the transmission will partly compensate for the thickness error and only a small fraction of the transmission of the filter in in central wavelength, that corresponds to the optical thickness error is lost. Usually this error is only in the order of magnitude of tenths of a percent or even lower. For the so-called rotation-monitoring, the number of rotations of the substrate holder was monitored, which was calibrated by the optical thickness increase per rotation in previous layers, that were monitored optically by measuring the transmission change during the layer growth. By this method, we insure a very accurate insitu-measurement of the actual thickness. A substrate rotation of 180 rpm and a resulting high rate of data sampling result in a very accurate thickness monitoring and tangential uniformity in the sense of rotation. The radial uniformity of the center wavelength of the filters was tuned by distribution shields and magnetic field tuning of the magnetron sputter sources.

It is also possible to deposit an AR-coating on the entrance side of a substrate and a bandpass coating with deep and wide blocking on the backside to achieve a similar spectral behaviour. In practice however this solution that requires more than 130 layers in a single coating has the tendency to show more oscillations of transmission in the passband of a filter. Furthermore if the bandpass coating and a blocking coating can be deposited independently on two sides of a component the spacial uniformity of the bandpass is generally much better in control. The bandpass coating on one side can be optimized for uniformity of centralwavelength and FWHM and the blocking coating can be optimized for high transmission, low ripple and high rejection. These were the main considerations for the choice of the current filter layer designs.

The important criteria for the choice of substrate types were high transmission, low climatic sensitivity, low autofluorescence and good behaviour in the polishing process.

2.3.

Bandpass filter specification

The fused silica substrates (7980 grade 1F) with a diameter 68.0mm +/- 0.1mm and thickness 6.0mm +/- 0.1mm were polished to P3 quality the transmitted wavefront should be smaller than lambda/4 @ 633nm over the clear aperture 65mm and smaller than lambda/8 @ 633nm in every subaperture 25x25mm. The surface defects were specified by scratch-dig 60/40 per MIL 13830B and the parallelity should be smaller than 30 arcsec. To accomplish that, the double sided polishing was applied at SCHOTT Suisse S.A..

At an angle of incidence of 2.2° the bandpass should be centered at 1254 +/-2nm taking the temperature shift and angular shift into account. The maximum transmission should be higher than 90% and the out of band blocking should be 10-5 average (OD5) from 1050-1650nm. The uniformity of the central wavelength, measured by a 5-point measurement (center point and 4 points about 5-10mm from the edge should be +/-2nm. The peak transmittance uniformity commonly named “passband ripple” should be smaller than 5% peak-to-valley and the peak transmittance flatness defined by the difference between minimum and maximum value within the passband should be smaller than 5%. The Full-width-half-maximum (FWHM) was specified by 11.0+/- 0.5 nm. The slope of the rising and falling edges should be better than 0.135% evaluated as |λ@80%-λ@5%|/CWL.

3.1.

Transmission measurement results

Spectral transmission measurements were performed on a Agilent CARY 5000. The measurement was performed with a data interval 0.1nm, scan rate 60nm/min, averaging time 0.1sec, slit bandwidth 0.2nm, aperture half-angle 7,5°, angle of incidence 2.2° with background and zero-correction at room temperature (21°C). The spectral uniformity was measured by a 5-point measurement (center + 4 corner points). Due to the lateral extension of the probe beam the lateral measurement points of the filter were chosen at approximately 5-10 mm distance from the edge. As shown in Fig. 6, the specified maximum transmission was exceeded.

Fig. 6

transmission measurement performed on 5 points of the filter (see insert right upper corner). The discontinued curve shows the theoretical filter curve calculated for the aperture half-angle of 7.5° in the photospectrometer used and an incident angle of 2.2° specified for the application.

A wavelength offset from the measurement conditions to a collimated beam was taken into account with +0.4nm. So the target centering was derived from the measurement offset of +0.4nm and the thermal wavelength shift of +1.57 nm. This leads to a target centering of the bandpass of 1256.97 nm.

The uniformity of the filter was evaluated from the measurements. In Table 1 the characteristic values of the measurement in Fig. 6 are displayed. The central wavelength (cwl) uniformity of the filter is 0.14% and the variation of the maximum transmission over the 5 points is only 0.33%. The FWHM shows a variation that is close to the specified resolution of 0.2 nm of the Agilent CARY 5000 instrument in the NIR.

Table 1

evaluation of the characteristic values of the 5 point measurement shown in Fig. 6

The out of band blocking from 1050-1650nm is better than the specification OD5 (T < 10^-5) on average as demonstrated by measurements shown in Fig. 7. The average value of blocking between 1050 nm and the shortwave 1%-footpoint of the bandpass is 6.34 OD-units. The longwave blocking value between the longwave 1%-footpoint of the bandpass region and 1650nm is 6.62 OD-units. The measurements were performed with state-of-the-art photometers Agilent CARY5000 that have a very good dynamic range of their detectors to resolve a blocking of OD7 in certain wavelength ranges. The blocking measurements were carried out with reference beam attenuation. The measurement parameters for the blocking measurement were interval 3nm, scan rate 360nm/min, averaging time 0.5 seconds, slit bandwidth 3 nm, aperture half-angle 7,5° and angle of incidence 0° with background and zero-correction at room temperature (21°C).

Fig. 7

blocking measurement performed with an Agilent Cary 5000 with reference beam attenuation showing an average blocking from 1050-1650nm of better than OD 6.

3.2.

Measurement result of the transmitted wavefront error

The transmitted wavefront was measured with a Zygo XPZ interferometer on the clear aperture of 65 mm diameter and shows a distortion of 0.01 wave evaluated with the reference wavelength 546nm. See Figure 8.

Fig. 8:

measurement result of transmitted wavefront evaluated according to DIN ISO 10110 (ref. wavelength 546nm) to be 13/0.01(0.29/0.09).

The local transmitted wavefront was evaluated in all subapertures of 25x25mm on the clear aperture and the worst distorted spot was evaluated to have a sagital error of 0.02 waves @ 546nm (see Fig. 9). As conclusion the filters exceed the global and local specification of transmitted wavefront distortion. Finally the parallelity of the filter was measured to be 5 arcseconds.

Fig. 9:

measurement result of transmitted wavefront in all local subapertures 25x25mm.