Background information to the GSTP activity G617-085MM on “Digital Mirror Array (DMA) for space optical instruments”
TEC-MMO/2015, 9 July 2015 by Benedikt Guldimann
The Smart Slit Assembly (SSA) is placed where in classical instruments the spectrometer slit would be. A basic schematic of an example of an SSA is shown in Figure 1, below.
Figure 1: Schematic of a linear SSA. The SSA is placed in the focal plane of the front optics (e.g. telescope, objective) where the classical slit would be placed. The image field is coupled into the input waveguide array (WAi). On the other side, the spectrometer is fed by the output waveguide array (WAo). Between both waveguide arrays a Micro Shutter Array (MSA) unit is placed which modulates the light intensity of each of the waveguides of the WAi. To illustrate the basic functioning of the SSA, the pixel on the left is “on”, or transmits the light coupled into the waveguide as soon as the micro-shutter is open. On the other hand, all other SSA pixels are “off”, or not transmitting the light through the output waveguide, since they are blocked by the MSA. Such an SSA can be arranged also in a 2D-array. In this schematic the MSA is directly placed between both waveguide arrays, which corresponds to solution 1 discussed further down.
The functions of such an SSA are outlined in short here:
- Scene/Object selection (e.g. de-clouding, removing too bright scene parts)
- Scene/Object intensity equalization (reducing dynamic range of optical signal)
- Inflight monitoring and calibration functions (e.g. co-registration, dark current)
- Pixel level scrambling of intensity, polarization and phase
- Mechanical decoupling of optical instrument parts
- Increased spectral stability / reduced noise on spectral detection
- Reduce polarisation sensitivity. Operates as polarisation scrambler.
Those functions are introduced essentially to improve significantly the performance of spectrometers, for instance by lowering the straylight levels, and enable science experiments performing of very high contrast scenes or in proximity of very bright objects. Other characteristics of the SSA increase the design freedom for accommodating the opto-mechanics, increase thermo-mechanical ruggedness, simplify the alignment of the instrument, calibration as well as AIT.
The SSA can for instance be based on a linear Micro Mirror Array (MMA) of 1x100 micro-mirrors, or more generally of a linear Spatial Light Modulator (SLM) which can operate in reflection (MMA) or in transmission (Micro Shutter Array, MSA). In this text such an SLM will be called DMA, standing for Digital Mirror Array, since the device name of the ESA development activity is called DMA.
The concept of the SSA is however also applicable to DMAs of 2D arrays, for instance a made of 100 x 100 micro-mirrors or micro-shutters (Figure 4). In Earth observation applications, where so far often a thin spectrometer slit is used, a linear array would be sufficient. In astronomy, as for instance on NIRSpec (JWST, NASA), a 2D array will be more desirable.
In simple terms, the SSA has the function of a spatial light modulator with additional functions, namely light de-coherence, polarisation scrambling, scene/object/image homogenization.
In short, light intensity modulation is achieved by the DMA whereas spatial amplitude variations, phase and polarisation entering the SSA are scrambled by both waveguide arrays. The particular location of the DMA in the SSA, namely between two waveguide arrays, offers particular design advantages to the DMA as well as to the optical instrument design. This is explained here:
In a spectrometer equipped with a traditional slit mask (for instance the MSA of NIRSpec or DMD of Texas Instruments) useful light modulation at the entrance of the spectrometer can only be achieved by switching between fully closed or fully open slit element (e.g. pixel) position. This because, if for example only half of the light intensity was needed to pass the MSA and positioned such as to let half of the light through (e.g. open half of the slit width) for the entire integration time, half of the image would be cut off for the entire integration time. The missing half of the image would remain undetected and the spectrometer response would depend of the instantaneous slit aperture. DMAs who have to operate continuously in on/off switching mode require quite high switching frequencies in order to be able to modulate the light sufficiently well in a spatially homogeneous way. High speed switching leads usually to designs with small shutters/mirrors (e.g. DMD of Texas Instruments). In Earth observation applications the core size of a waveguide can be as large as 100 microns by 100 microns. In low Earth orbits the integration time can be clearly below a second, which could require switching frequencies higher than 1 kHz. The MSA design for NIRSpec for instance could never achieve such frequencies.
As a side notice here, a DMD type of device would need to modulate a significantly large sub- array of micro-mirrors (In the order of 12m x 12m) per slit pixel/waveguide (100m x 100m) to operate in a SSA.
Assuming now the DMA placed between two sufficiently long multimode waveguide arrays (square shaped, rectangular shape other suitable shapes), either directly between the two waveguide arrays (solution 1, Figure 1) or with a relay optics (solution 2, Figure 2), spatial image homogenisation is achieved with the input and output waveguides. The spectral content of the image pixels is therefore also uniformly distributed at the waveguide output surface. In order to modulate the light intensity to half of the intensity for instance, a shutter or mirror cutting off half of the beam provided by the waveguide would suffice. This because of the homogenisation function of the input and output waveguides. The output waveguide homogenises the half transmitted by the DMA again to an uniform intensity and spectral distribution at the output waveguide feeding the spectrometer. The particular arrangement of the SSA allows therefore DMAs with mirrors/shutters operating in an analogue manner, allowing slower DMAs and DMAs with various shutter designs (spatially in-homogeneously) operating at slower speed than the switching ones. In general terms, the fact that the DMA is between two waveguide arrays (with or without relay optics) givesthe opportunity to relax the DMA design on speed requirements and allows designs where the optical beam is attenuated in an in-homogeneous way. In other words, the shutter does not need necessarily to be able to fully open and close. The shutter could be designed as a variable grid or sliding door for instance.
Reflective SLMs like the DMD of Texas Instruments or the MIRA development of EPFL and LAM  which could in principle be used as DMAs have the disadvantage to introduce optical aberrations which limits the optical imaging quality of the instrument. This is shortly explained here with an example of a 100m x 100m micro-mirror array based SLM for a smart slit which can be tilted from the in plane position to a 30°angular position. Reflective slits have to operate such as to direct the incoming beam into a direction which is not overlapping with the incoming beam. Each micro-mirror in “open” mode has therefore to be in a tilted position to function. In other words, the focal plane is then sampled by tilted mirrors. The “closed” position in this example would be the in-plane mirror position. Let’s assume that the focal plane is really flat and located at the surface of the micro-mirror array (in-plane) and the telescope diffraction limited and that the rotation axis of each micro-mirror is also on the un-tilted array surface. As a micro-mirror is tilted, the only in-focus image part reflected by each micro-mirror would be the image reflected along the rotation axis of each tilted micro-mirror. The edges of each micro-mirror will reflect an image part which is out of focus. The “closed” and in plane micro-mirror will reflect the focused image back to the telescope. A reflective smart slit introduces therefore aberrations. It is maybe possible in some cases to design the telescope or the spectrometer to compensate for this optical imaging degradation, or maybe the instrument requirements could live with such degradations, but it is considered as preferable here not to introduce new constraints or disadvantages in the optical design.
As in point one, replacing a stand-alone reflective SLM by a SLM (reflection or transmissions) placed between to waveguide arrays (SSA) helps to remove the disadvantages of the reflective SLMs (e.g. aberrations). The imaging performance of the spectrometer, being fed by the output waveguide array, is decoupled from the imaging performance of the optics between the waveguide arrays inside the SSA. It is sufficient that the micro-mirror array modulates the homogenised beam intensity coming from the input waveguide array well enough without introducing too much cross-coupling,
To summarize, the second advantage of the SSA compared to a reflective smart slit without multimode waveguide arrays, is that no imaging quality degradation (e.g. astigmatism) is introduced, hence the SSA makes it possible to obtain feasible and simpler optical instrument designs.
One of the limitations in designing such an assembly is given by the optical beam divergence, or numerical aperture, soon after the light has left the waveguides. The diverging beams of the waveguides overlap after a short distance and cannot be selectively modulated in intensity anymore once this propagation distance is passed. Two possible design solutions for an SSA, incorporating a DMA, have been identified so far:
The first solution is to couple the light into the output waveguide at a very short distance from the emitting waveguide (Figure 1). Modulation has in this case to be achieved with a thin shutter/modulator between the waveguides. This shutter/modulator would have to operate in transmission. It is a waveguide array – shutter/modulator array – waveguide array arrangement (wsw). If a shutter/modulator operating in reflection was placed after the emitting waveguide, which has typically a cladding diameter of typically 125 microns, the free space propagation distance (and therefore the beam diameter) would become too large for selective coupling into the corresponding output waveguide. Micro-lenses at the fiber to DMA interfaces could be used to increase the coupling efficiency or to reduce the size of the transmission area of the DMA/shutter.
In practice, the DMA and the waveguide arrays will probably have to be manufactured separately and assembled at a late stage. The waveguide to waveguide distance requirement, being in the range of 20 μm or a bit larger, is asking for new MEMS shutter designs since mechanically stable silicon chip or wafer thickness is usually about several hundreds of microns. Possible shutter MEMS designs could be based on a hollow waveguide structure next to the shutter which would guide the light from the proximity of the shutter to the multimode waveguide array (e.g. optical fiber array). On the light path shortly after the first waveguide array the thin shutter mechanism has to modulate the light without leaking too much to the other modulation channels or waveguides. Since the output waveguide array can in practice probably not be assembled right after the shutter, due to chip mechanical stability issues, an integrated waveguide, for instance hollow, could be manufactured on the modulator/shutter chip in order to interface between the light shutter and the waveguide array.
The second solution is to place relay optics, imaging the emitting waveguide array on to the shutter/modulator and another pass through a relay optics (Figure 2) to couple/image the light passing the shutter/modulator into the output waveguide array. It is a waveguide array – optics- shutter/modulator array – optics – waveguide array arrangement (wosow). This assembly could also be designed in reflection so that only one optics has to be assembled and is used in double pass, once to the modulator/shutter array and back to the output waveguide array on the same side as the input waveguide array. It requires bulky optics to be placed on the optical path between both waveguide arrays. This hardware configuration has been mentioned in the literature [1, 2] for multiplexing and field-slicing applications for ground based astronomy. The use of such hardware for intelligent slits, slit masks or object/image intensity homogenisation has to our knowledge however never been reported.
A potential third solution based on microlenses in front of each waveguide array to collimate the beams propagating to and away from a DMA, usually a micro-mirror array. This solution is commonly used in telecom optical fiber switch arrays, where microlenses are in front of single mode fibers (or lens shaped fiber ends) behaving like point sources. However in the present case, obtaining a good collimation is not straight forward since square or rectangular multimode waveguides with thin claddings are used. Therefore this third solution is not considered seriously for spectrometer applications unless the microlenses were used as an optional feature in solution 1 where micro-shutters are used in close proximity to the fibers:
- Microlenses could be used at the beginning and the end of the SSA in order to couple light from the slit aperture into the waveguide array and to ensure a better slit filling at the output slit aperture (end of output fiber array).
- Microlenses could be used between fibers and DMA in order to improve the throughput.
The complete assembly between the entrance slit plane and the exit slit plane is called Smart Slit Assembly (SSA) and a possible concept, corresponding to the first solution discussed above, is presented in Figure 1. The waveguide array (WAin) in front of the DMA brings the light from the telescope focal plane (entrance slit plane) to the DMA. The waveguide array after the DMA (WAout) brings the light to the spectrometer entrance (exit slit plane, slit viewed by the spectrometer). The second SSA design solution is shown in Figure 2, below.
Figure 3 shows a basic linear SSA input and output geometry based on an on/off modulating DMA. Figure 4 shows a generalisation of the concept of a Smart Slit Assembly (solution 1).
Figure 2: Example of Smart Slit Assembly (SSA) design solution(2) viewed from the side with a DMA operating in transmission or in reflection. Light from telescope (for instance F/3) is focused on the slit entrance (along track view) and guided through the first waveguide array (1D or 2D), imaged onto the shutter/mirror array (1D/2D DMA), re-imaged into the second waveguide array (1D/2D) and then transmitted to the spectrometer slit.
Figure 3: Basic DMA geometry: Example with a1D array. Top slit corresponds to a state of the art slit (aperture typically in a metallic membrane). The slit in the middle is an example of a smart slit, where certain parts of the scene/object are blocked. The slit below is a closed slit, for instance during commissioning to avoid direct sun-illumination reaching the detector or during detector frame transfer to reduce read-out noise.
2. Functions of the core elements
As a background information, the waveguide arrays at the entrance (WAi) and/or exit (WAo) of the SSA have several functions:
At DMA level, the major role of the optical waveguides is to homogenise the intensity variations of the object/scene imaged to the entrance aperture of the slit in the focal plane of the front optics (e.g. telescope) allowing DMA to attenuate the light coupled into the output waveguide in an inhomogeneous way (e.g. cutting beam, variable grid, variable shuttering). The optical waveguide array after the shutter/micromirror array also has the role to homogenise the light again allowing so shutter the light in-homogeneously. This is important to relax the design and operational specifications of the DMA compared to a device which has to modulate the light homogeneously.
- At system level:
- Slit becoming an optical fiber link increases significantly the opto-meachnical design flexibility, simplifying also instrument alignment and integration. The telescope and the spectrometer(s) can be accommodated separately on smaller , simpler and more stable mechanical structures which leads also to higher mechanical stability and ruggedness. Instrument can be set up on several smaller optical benches> higher stability.
- to eliminated speckles from calibration stage (coming from diffusor) using de-coherence.
- to smoothen, scramble or depolarize polarised incoming light. Opens the possibility to design the instrument without the polarization scrambler or to reduce the strength of a necessary polarization scrambler. This usually leads also to an increased image quality.
- to be used as integral field slicer
- integral field expander in case of a 2D (Figure 4) waveguide array in order to accommodate a micro shutter array with low fill factor along one direction/axis (see Figure 4)
- could be used also to filter light
- works as slit-homogeniser to increase spectral signature stability.
- entrance aperture surface of the input waveguide array and/or output aperture surface of the output waveguide array could instead of being flat be manufactured with optical power in order to play the role of additional curved optical surface(s) (e.g.lens).
The DMA of the SSA have several functions:
- Scene/object selection or attenuation ( e.g. de-clouding, blocking of bright objects, scene intensity equalisation, scene/object selection ) > reduction of straylight (increase of the radiometric accuracies), increase of the dynamic range of the detection chain.
- Slit closure (during CCD frame transfer) to reduce noise on detector or preventing direct sun illumination
- On-ground and in-flight monitoring functions (e.g. co-registration), calibration and correction functions
Figure 4: 2D multimode fiber bundle is split along one axis (y-axis in order to accommodate shutter arrays with low fill factor along the same axis. The space between the linear shutter arrays could be used for the shutter/mirror micro-actuators or suspension mechanism. This bundle arrangement could also be used in combination with relay optics between the fiber and the shutter/mirror arrays.
 Murray, G. J., Allington-Smith, J. R., "Strategies for spectroscopy on Extremely Large Telescopes - II. Diverse-field spectroscopy," MNRAS 399, 209-218 (2009).
 Poppett, C. L., Allington-Smith, J. R., Murray, G. J., “Strategies for spectroscopy on extremely large telescopes - III. Remapping switched fibre systems”, Monthly Notices of the Royal Astronomical Society, Volume 399, Issue 1, pp. 443-452.
Francesco Pepe, David Ehrenreich& Michael R. Meyer,“Instrumentation for the detection and characterization of exoplanets”, Review Paper, doi:10.1038/nature13784
 G. Avila ; P. Singh, “Optical fiber scrambling and light pipes for high accuracy radial velocities measurements“, Proc. SPIE 7018, Advanced Optical and Mechanical Technologies in Telescopes and Instrumentation, 70184W (July 14, 2008); doi:10.1117/12.789509
 Patent pending: WO 2012/115793, PCT/US2012/024574, OPTICAL HOMOGENIZING ELEMENTS TO REDUCE SPECTRAL NOISE IN HYPERSPECTRAL IMAGING SYSTEM.
For technical questions please contact:
Tel: +31 565 3592