Infra-red Interferometry

Interferometry is used as a routine test for optical components and systems at wavelengths ranging from the visible to above 10 microns. In the infra-red the main wavebands used are from 3 to 5 microns and from 8 to 12 microns. Typical laser sources used are HeNe (3.39microns) and HeXe (3.49 microns) in the shorter waveband and CO2 (10.6 microns or tuneable) in the longer waveband. Infra-red interferometry presents some problems which require that special care is taken in the setting up of the measurement, and that the results are analysed in as rigorous a manner as possible.

Infra-red optical systems are being produced with smaller optical apertures than previously for reasons of weight, cost and configurability. This trend has been aided by improvements in the detector technology. The result is that the IR lens for a camera system may have an aperture comparable in size to a visible camera system. However the wavelength in the IR system may be as much as twenty times greater than in the visible band. This makes the effects of diffraction much more dominant in IR interferometry, requiring that optical pupils are well focused onto the interferometerâs camera sensor surface. Because of the double pass nature of most commercially available interferometers the aperture is Îseenâ twice by the interferometer - if the two images of the optical aperture are not conjugate with one another then some diffraction is inevitable. The effects of this must be minimised to enable accurate analysis.

 Figure 1 shows two ways of analysing a lens with an interferometer. One test uses a spherical reference mirror while the other uses a flat reference mirror but requires an additional high quality focusing lens (or reference sphere) in order to make the measurement. From the point of view of diffraction the measurement using the reference sphere is preferred, since the return mirror can be very close to the pupil of the lens under test, therefore the pupil and its image via the mirror can both be very close to focus on the interferometers camera faceplate. This is not possible using the other arrangement, diffraction will be present especially if the spherical mirror radius is small so that it is physically far from the lens under test. Many tests in the infra-red use high grade ball bearings as return mirror ö it is important that any analysis of the interferograms takes into account the limitations of the method.

Figure 1. Reflection Configurations

Figure 2 shows the effect of diffraction on the pupil imagery in the interferometer.

In the unfocused case, diffraction rings can clearly be seen inside the pupil. This has the effect of breaking up the interference pattern such that the fringes are discontinuous or vary in width due to contrast changes. Any automatic fringe analysis program must be used with caution when analysing such fringes. In general the peak to valley wavefront aberration reported will be worse than the actual for the lens under test.

While figure 2 shows the effect of diffraction within the optical pupil, the other common case is when the diffracted radiation appears outside the pupil. This has the effect of blurring the pupil edge so its size is uncertain and, because the diffracted radiation has a different curvature to the main beam, any interference fringes will curve within this zone. The effect is very similar to that seen with Spherical Aberration especially that of higher order and if not corrected can lead to a large departure in measured vs. actual performance. This situation is shown schematically in Figure 3.

In this case, if information is known about the size of the pupil within the interferogram,  it is possible to mask out much of the effects of diffraction simply by ignoring those parts  outside the pupil. 

There will be cases where it is impossible to achieve good pupil imagery without going to great expense. In such circumstances, double pass interferometry can still provide a confidence test on system performance, but ultimate proof may require a single pass test, either by interferometry or by non coherent broad band MTF testing or similar.

Infra-red interferometry relies on sensing using an Infra-red camera, either staring array CMT or a Pyroelectric vidicon tube. The latter is the most used, mainly because of the cost of the camera. The cameras exhibit random noise on the video signal. this is most obvious in the 3 to 5 micron waveband, where available, affordable lasers tend to be of low power (typically 5-10mWatts). High signal gain is required to provide sufficient contrast for recording fringe patterns or analysing fringe patterns using either static or phase shifting software. Also of relevance when laser power is marginal is the profile of the laser beam. As much of the beam as possible must be used to maximise contrast but this does mean that, due to the profile, the contrast varies across the interferogram. This is particularly noticeable when using HeNe lasers at 3.39microns where perhaps only 5 to 10mWatts of power is available. This can have the effect of making fringes appear both fainter and thinner towards the edge of the pupil. When analysing with static fringe analysis packages care needs to be taken to check that the digitised fringes accurately represent the real shape of the fringe pattern. 

The problems raised by laser power profiles is greater when using Phase Shifting systems. Phase shifting system involve a threshold contrast below which phase will not be measured. This normally shows as a blank space in the resulting phase map. This can be remedied by reducing the threshold. However, using Pyroelectric vidicons with marginal laser power represents a very noisy environment. Reducing the threshold too far mean that the system can pick up on random noise in the picture and allocate  phase values which  significantly larger than any other area of the pupil. This has the effect of  a gross distortion of the results. A fine balance needs to be trodden between these two situations such that phase information is available over the whole pupil without random noise causing a problem. Generally with 3.39 micron systems, the threshold needs to be set at around 2.5% to 3.5%.

At 10,6 microns, these problems donât occur. The main problem in the upper IR waveband is finding ways of dumping excess energy ö normally with CaF2 filters though with tuneable systems a wire grid polariser is necessary. This means that the central part of the beam can be used ensuring a much flatter profile, and sufficient power can be allowed through in order to enable the pyroelectric vidicon camera to be set to low gain, thereby reducing noise to an insignificant level. Further development in compact high quality 3-5 micron lasers is eagerly anticipated.