2.1

Design of the instrument

SALSA sits in an ISO6 clean room facility that preserves it from general environmental variations and dusts. Figure 1 shows a functional representation of the SALSA instrument composed of different subassemblies. The emission subassembly is based on a supercontinuum laser source (FIANIUM WL-SC-400-8) which outputs a very wide spectrum light signal from 400 nm to 2.4 μm with a total integrated power of 8 W. A dichroic beam splitter (FIANIUM Splitter-950-01) splits the broadband spectrum into two channels: a visible channel (from 400 to 975 nm) and an infrared channel (from 975 nm to 2.4 μm).

Figure 1

Spectral and Angular Light Scattering Apparatus (SALSA) is composed of three subassemblies: Emission, Measurement and Detection.

The optical components used on both channels are functionally identical but differ on their spectral operation ranges. The two channels are used independently and asynchronously, if one is activated then the other one stay closed. Two tunable volume hologram filters (Photon Etc. LLTF Contrast VIS HP8 and LLTF Contrast SWIR HP8 respectively) set the working wavelength and reduce the spectral width of the signal between 2 and 3.5 nm. Electro-mechanical shutters (SH1 and SH2) block the light beam and transmit it only during acquisitions. Order sorting filters (OSF1 and OSF2) clean the signal from parasitic harmonic orders induced by the volume hologram filters. Mirrors (M1, M2 and M3) spatially align the optical beams and beam splitters (BS1 and BS2) collect a small portion of the signal, around 10%, as reference signals. Then the main signal goes through a set of four neutral optical density stages (ODs). The combination of these densities results in a total OD between 0 and 10, adjustable by a pitch of 1. The only structural difference between the visible channel and the infrared channel is a removable recombination mirror (RM) which reflects the visible channel towards the optical densities. This mirror is mounted on a translation stage, when performing acquisitions in the visible spectrum the mirror reflects the light beam towards the output of the emission subassembly, while it clears the way for the infrared channel when performing infrared acquisitions. The light beam is finally collected by a parabolic off-axis reflective collimator (ORC). The reference signals collected by the above-mentioned beam splitters (BS1 and BS2) are also sent to reflective collimators (RC R1 and RC R2) connected to photodiodes (VIS_PD and IR_PD, respectively FEMTO DE-OE-200-SI and DE-OE-200-In1) via multimode optical fibers (FOL R1 and FOL R2).

The measurement subassembly is connected to the emission subassembly with a multimode fiber optical link (FOLl, core diameter 50 μm). A reflective collimator (TRC) outputs the optical beam which illuminates the sample under characterization (TFF). The incidence angle is set by a rotation stage that holds the sample. After being transmitted, reflected, or scattered, light is collected by a second reflective collimator (RRC) mounted on a rotation arm centered on the sample. Therefore, by adjusting the angular position θ of the mechanical arm, it is possible to measure scattered light at any angle around the sample and in the plane of incidence. The measurement subassembly is connected to the detection subassembly with a multimode fiber optical link (FOL2, core diameter 105 μm).

The detection subassembly includes a Czerny-Turner monochromator (ANDOR Shamrock SR-193i-B1-SIL), two gratings are mounted back-to-back on a turret in the spectrograph. The first one is an 1800-lines-per-milimeters holographic grating optimized for a range from 360 to 1100 nm. The second one is a 760-lines-per-milimeters optimized for 850 nm to 2 μm. The spectrograph owns two output ports, on each is mounted a scientific grade camera: a CCD detector on port1 (ANDOR Newton DU920P-BVF) with 255 x 1024 pixels of 26 x 26 μm each and an InGaAs linear detector on port2 (ANDOR DU491A) with 1 x 1024 pixels of 500 x 25 μm each. A sliver-coated flip mirror (FM) leads the light beam to one of the two ports. The CCD camera detects visible light, while InGaAs is sensitive to the infrared spectrum.

Figure 2

SALSA instrument in ISO6 clean room in the Espace Photonique of Institut Fresnel. Measurement subassembly: a reflective collimator (right) outputs the optical beam which illuminates the sample (middle), the scattered light is collected by a second reflective collimator mounted on a rotating arm (left).

2.2

Operation of the instrument

The measurement procedure has been already presented in our previous papers [5]–[7]. For measuring the amount of scattered light by any sample at a wavelength λ₀ and an angle θ₀, the rotating arm is set at the chosen angle. The value of λ₀ determines which channel (visible or infrared) to use in the emission and detection subassemblies. The involved LLTF filter, the recombination mirror, the spectrometer grating, and the detection camera are tuned to the chosen λ₀ value. Initial values are set for the τ (exposition time) and OD (overall optical densities) parameters. Then the electromechanical shutter is opened to let the light beam illuminate the sample. The acquisitions on the reference photodiode and the camera are synchronized thanks to this single shutter. The signal on the camera is expected to be around 70% of its saturation level. The effective level is adjusted by tuning the τ and OD parameters simultaneously. Several acquisitions might be executed until the signal achieves a convenient level. The CDD camera has numerous lines of pixels and the light beam illuminates only a small area on the detector. The illuminated area is small enough to acquire simultaneously the signal and the noise using different zones of the detector matrix. On the other hand, the InGaAs camera has only one line of pixels which collect the signal. It is not possible to acquire the signal and the noise simultaneously. The noise level is measured immediately after the signal with another acquisition of the same duration but with no illumination of the sample. Depending on the position of the rotating arm, SALSA measures either the scattered light or the specular transmitted or reflected light. The scattering function estimated is the Angle Resolved Scattering (ARS), a spectrally and angularly resolved function which describes the angular distribution of the scattered light around the sample. The specular functions estimated are the transmittance and the reflectance. In any case a calibration is required. A Spectralon white diffuse reflectance sample defines the standard levels for the scattering measurements, while a baseline (with no sample) acquisition defines the reference level for specular functions measurements.

3.1

Light scattering characterization

A dedicated bandpass filter centered at 1273 nm has been used for the qualification of the instrument, this sample is called F0 in this paper. This filter has been designed and manufactured by CNES and CILAS ArianeGroup and is intended for a space application. It is composed of around 100 alternating layers of SiO₂ and Nb₂O₅. The calibration of the setup requires the measurement of the ARS of a perfect white reflectance Lambertian diffuser (LABSPHERE Spectralon). The ARS analytical formulation for this material is the following:

Where ρ(λ) is the albedo of the sample at wavelength λ, the angular cosine dependance is a specific feature used to check the correct alignment of the setup. The ARS measured on this sample is compared to a cosine shape, a good agreement assesses a right optical alignment of the sample and the rotating arm.

Figure 3

Comparison between the ARS cosine model and the normalized ARS measured on the Spectralon sample at 1490 nm.

Figure 4

Angular response of the instrument to the direction defined by the transmitted specular beam.

The angular resolution is granted by the constitution of the setup and the use of a high-precision rotation stage. The angular response presents a relative decrease of 8 decades at 1° from the specular beam position. Therefore, we can confidently assert that the angular resolution of the setup is below 1°.

ARS measurements on the sample F0 show both spectral and angular variations. Recent works carried out in our group have led to a complete and powerful theoretical model implementation which describe such scattering from optical thin-films [10]. The comparison between the model and the measurements presents an excellent and unprecedented agreement. Not to forget, the sample under characterization is very complex which makes it difficult to characterize experimentally and numerically. The model also highlights the spectral and angular peaks observed experimentally. An acquisition without sample depicts the Signature of the instrument, i.e., the lowest detectable values. The level of the Signature shows that the setup should be able to measure down to 10⁻⁸ sr⁻¹, closed to the air particles scattering limit. F0 has been characterized at some specific wavelengths, Figure 5 and Figure 6 depict the measurement and model data at 1115 nm and 1270 nm with a fine angular resolution (1 nm). The same comparison has been carried along all the near-infrared accessible spectrum with a lower angular resolution and the agreement with the model remains excellent.

Figure 5

ARS measured on F0 at 1115 nm. Comparison with Model and Signature.

Figure 6

ARS measured on F0 at 1270 nm. Comparison with Model and Signature.

3.2

Specular transmittance characterization

The same F0 filter has been subjected to specular transmittance characterization, illustrated on Figure 7. The measurement data shows an excellent agreement with the design data: the high rejection levels are well defined, as well as the narrow spectral peaks and the steep edges. These results validate the asserted that the spectral resolution is lower than 1nm. Once again the Signature of the instrument is very low, around 10⁻¹², which is 6 decades below the state-of-the-art [11].

Figure 7

Transmittance of the dedicated F0 filter from 400 to 1650 nm. Comparison between the design, the measurement, and the Signature of the instrument.

Figure 8

Transmittance of F0 superposed with a silver monolayer density (Ag2) compared with single F0 and single Ag2 on the [1025-1275 nm] band. The resulting transmittance is close to the Signature level.

The rejection of the above-mentioned optical filter is considerably larger than the noise level. To validate the detection limits, a 60 nm thick silver monolayer neutral density, called Ag2, is superimposed on F0. This optical density has a transmittance around 10⁻³ in the near infrared range. The total transmittance of the two components superposition is measured between 1025 nm and 1275 nm, roughly centered on one of the two main deep rejection bands. The resulting transmittance value is the one of the F0 filter lowered by a factor of 10⁻³. The lowest level achieved in this configuration is 10⁻¹¹, close to the Signature limit.

3.3

Improve the spectral resolution

The measurement campaign for F0 characterization over the whole available spectrum counts 1301 data points, from 400 nm to 1.7 μm by step of 1 nm. The resulting resolution is 1 nm, but it can be improved with an adequate data post-processing. The combination of a monochromator with an array-based detector shows a great interest in terms of spectral discrimination. The light beam arriving at the monochromator has a small spectral bandwidth (between 2 and 3.5 nm) dispersed spatially by the grating. The signal acquired by the camera is spread over a horizontal line of pixels Considering the dispersion of the grating it is possible to discretize the wavelengths closed to the working one using different pixels of the camera. This configuration cleans the signal from non-desired spectral contributions and increases the spectral resolution. The instrument can consequently resolve narrow spectral peaks (<1 nm). Figure 9 illustrates this spectral dispersion on the raw signals detected by the cameras, the blue points represent baseline acquisitions (no sample), the red points represent acquisitions with sample. The presence of a sample under characterization modifies the signals and provides information about the spectral function locally. The spatial information along the pixel line is transcribed into spectral information. Figure 10 depicts a high-resolution processing, resulting in a transmittance resolution of 0.1 nm obtained from a 1 nm resolution acquisition. The comparison with the high-resolution processing is excellent and suggests that the narrow spectral peaks are even better resolved. The transmittance levels are displayed only between 10⁻⁴ and 1 on vertical axis and from 1075 nm to 1300 nm for ease of reading.

Figure 9

Raw signals acquired on the 1024 pixels line. The spatial position corresponds to a spectral information around the working wavelength (left 850 nm, right 1350 nm).

Figure 10

Top: transmittance of sample F0 with 1 nm spectral resolution. Bottom: transmittance of sample F0 with high-resolution processing. The high-resolution processing is coherent with the classical processing and improves the spectral information.