RLS spectra acquisition optimization with the RLS FS and RLS ExoMars Simulator

Introduction

The flight Raman instrument for the ExoMars 2022 mission, the Raman Laser Spectrometer (RLS) [1], was delivered and integrated in the Rosalind Franklin rover. In parallel, the RLS flight spare (FS) model is being used at INTA facilities to thoroughly characterize and understand the instrument performance, as well as to optimize the scientific return of the flight instrument through a proper parameterization. In addition, the RLS ExoMars simulator developed by the University of Valladolid (UVA), is used for thorough analysis of samples, emulating the operation mode of the RLS instrument (including  the automated adaptation to the sample spot), as well as the sample preparation and distribution system (SPDS) of the rover, in those aspects related to sample management.

In this work we present experiments and analysis performed with the different ground models, aimed at the optimization and characterization of the RLS performance, by addressing the following three issues: 1- optimizing the acquisition by studying the SNR of spectra with different configurations of the spectra acquisition algorithms implemented onboard the RLS instrument [2] (especially the number of accumulations), while also considering the influence of the instrument stability (laser power, CCD temperature, etc.). 2- optimizing the spectral quality of the acquired data by evaluating several on-ground spectral data-processing strategies. 3- evaluate and analyze the performance of the different models (FM, FS and Simulator) to understand the expected inter-correlation of the results obtained with them.

Experiments and previous work

Several samples with different emission efficiencies (diamond, calcite, serpentine, hematite and vermiculite), have been analyzed with the RLS FS and the RLS ExoMars Simulator, acquiring a relatively high number of acquisitions. The data has been processed to establish the spectral quality (measured as the SNR) as a function of the number of accumulations.

The RLS ExoMars simulator, on the other hand, has also been reworked to integrate the RAD1 spectrometer (RAman Demonstrator 1, a laboratory model with characteristics and design similar to RLS). This will bring the RLS ExoMars Simulator closer to the expected performance of the instrument that will fly to Mars.

Optimal n_a

The RLS instrument features an automated integration time (t_i) calculation algorithm, which allows optimizing this time for every spot in the sample. However, due to several reasons, the number of accumulations (n_a) has to be established from ground as a system parameter, so it will be the same for all the samples and spots obtained during one operational cycle (sol). It has been reported [2] that, for constant total acquisition times (t_i*n_a) in instruments such as RLS, the signal to noise ratio (SNR) of a spectrum increases more by maximizing t_i, than by maximizing n_a. This is in agreement with the implementation performed on the onboard software of RLS. Thus, the present study is centered in the characterization of the optimal number of accumulations (which is configurable from ground) for the RLS operation.

The results in this work have been used to infer the SNR evolution of the spectra as a function of the number of accumulations, which is critical to establish the on-ground parameter for n_a for an optimal acquisition. The results are obtained by comparing several SNR calculation methods, and have helped determining a good compromise in the selection of the n_a at a value of 30.

This analysis has shown noise levels behaving very close to the theoretical behavior, with the noise intensity following the expected inverse exponential decay with n_a. However, it has also provided insight as to how the spectra peak intensity is highly affected by the laser power or CCD temperature stability. In this sense it has been concluded that it will be necessary to perform some tests at arrival on Mars, in temperature conditions representative of the environment during Martian operations.

Data processing for optimal spectral quality

The analysis of the acquired data has also shown how the spectral quality of the spectra can be influenced by the data processing: CCD binning method, dark subtraction, baseline correction approach… but also how different intensity correction methods can help improve the quality of the spectra. For example, the correction with BZn [3] has been used to perform correction of the spectra (using different correction methods), showing how the data analysis strategy incorporated into the science processing pipelines of RLS during operation will need to include this correction. Other data processing strategies to be considered are the use of optimized binning methods (adjusted for each spectrum), or the dark correction of spectra, which also improves the spectral quality.

Correlation between instrument models

To have ground models which can be correlated with the instrument on Mars is critical for analysis during operations, but also to prepare and validate the science that will be obtained from the instrument once on Mars. The RLS FS instrument is identical to the flight instrument in every sense, which makes it very representative of the RLS FM, except in the operational conditions, as it is not possible to simulate the Martian conditions in a daily operational basis (and, for example, the CCD working temperature is kept at values higher than what will be expected on Mars).

On the other hand, the correlation of ground emulators such as the RLS ExoMars Simulator is of paramount importance to obtain results that are realistically correlated with the actual RLS instrument. The integration of the RAD1 spectrometer in the RLS ExoMars Simulator is a great step forward to achieve representative results that can ultimately be used to take decisions as to how to parameterize the flight instrument for Martian operation.

Acknowledgements

MINECO grants ESP2014-56138-C3-1-R, ESP2014-56138-C3-2-R, ESP2107-87690-C3-1-R, ESP2107-87690-C3-3-R.

References

[1] Rull, F. et al. Astrobiology 17, 627–654 (2017).

[2] Lopez-Reyes, G. et al. J. Raman Spectrosc. 48, (2017).

[3] A. Sanz-Arranz et al. J. Raman Spectrosc. 48, (2017).