Signal analysis of NEMS sensors at the output of a chromatography column

Auteurs :
Publication MaxEnt 2014


Signal analysis of NEMS sensors at the output of a chromatography column


application/pdf Signal analysis of NEMS sensors at the output of a chromatography column


54.35 KB


Creative Commons None (All Rights Reserved)


Scientific sponsors


Logistic sponsors


Funding sponsors

<resource  xmlns:xsi=""
        <identifier identifierType="DOI">10.23723/9603/11322</identifier><creators><creator><creatorName>Francois Bertholon</creatorName></creator><creator><creatorName>Olivier Harant</creatorName></creator><creator><creatorName>Christian Jutten</creatorName></creator><creator><creatorName>Pierre Grangeat</creatorName></creator><creator><creatorName>Bertrand Bourlon</creatorName></creator><creator><creatorName>Laurent Gerfault</creatorName></creator></creators><titles>
            <title>Signal analysis of NEMS sensors at the output of a chromatography column</title></titles>
        <resourceType resourceTypeGeneral="Text">Text</resourceType><dates>
	    <date dateType="Created">Sat 30 Aug 2014</date>
	    <date dateType="Updated">Mon 2 Oct 2017</date>
            <date dateType="Submitted">Thu 15 Mar 2018</date>
	    <alternateIdentifier alternateIdentifierType="bitstream">dfc379bac4fc7e846f878a9418e095a1f17c6072</alternateIdentifier>
            <description descriptionType="Abstract"></description>

Signal analysis of NEMS sensors at the output of a chromatography column F. Bertholon*, O. Harant*, Ch. Jutten+ , B. Bourlon*, L. Gerfault*, P.Grangeat* * Université Grenoble Alpes, CEA, Leti, Minatec Campus, 17 Rue des martyrs 38000, Grenoble, France. + Université Grenoble Alpes, GIPSA lab, 11 Rue des Mathématiques, 38400 Saint-Martin-d'Hères, France Abstract Gas chromatography is a technique to separate chemical components in gas state [1]. A chromatogram is a succession of peaks each corresponding to the output from the column of molecules of the same component. We consider new devices where nano-chromatography columns carved on silicium chip are coupled with sensors called NEMS, Nano Electro-Mechanical Systems. Those gravimetric sensors are vibrating cantilevers, covered with a chemical layer for molecular adsorption. The resonance frequency is controlled by a Phase Locked Loop control [2]. The output signal is the instantaneous frequency of the vibrating beam. This resonance frequency decreases when molecules are adsorbed on the cantilever. The frequency shift is proportional to the mass adsorbed. In this paper, we consider an inverse problem approach to retrieve the original gas mixture composition from the output signal based on Bayesian source separation [3] and model inversion. The gas chromatography (GC) column and the NEMS sensors are here described in a molecular point of view through a stochastic model based on the random walk molecular model proposed by Giddings and Eyring [4,5]. Our proposed general model describes the shape of the signal output as the probability distribution of the retention time of the molecules within the column. Each acquisition is settled with some prior parameters defined by the known column and chemical properties such as the prior distributions for a molecule to stay within the mobile phase or to remain fixed on the stationary phase, or the prior distribution to be absorbed on the cantilever and then to be released in the mobile phase. Those parameters control the retention time distribution for each random walk of each molecule within the column. The inference on the unknown profile parameters is processed according to a Bayesian scheme [6,7] based on the joint probability density function of the output retention times. Such inference requires first to identify each peak by locating their relative position. The position and width of each peak which define the parameters of our model are then estimated with a Maximum Likelihood Estimation to fully characterize the shape function of each peak. Then the mixture coefficients can be estimated. Indeed, each shape function corresponds to one gas and constitutes a base vector of elementary components. The scalars of this vector space which need to be determined are the proportions of each constituent inside the mixture. This defines a hierarchical model with 2 levels: component and GC signal. Then we propose a Hierarchical Bayesian source separation method to estimate the basis vector and the scalars of a given chemical sample in this basis. Finally we provide experimental evaluation on the analysis of a mixture of hydrocarbons. Références [1] J.M. Miller H.M. McNAir, Basic Gas Chromatography, 2nd ed.: Wiley, 2009. [2] Eric Colinet, Laurent Durraffourg, and al, "Self-oscillation conditions of a resonant-nano- electromechanical mass sensor," Journal of applied physics, no. 105, 2009. [3] Leonardo Tomazeli Duarte, Saïd Moussaoui, and Christian Jutten, "Source separation in chemical analysis," IEEE Signal Processing Magazine, pp. 135-146, May 2014. [4] J.Calvin Giddings and Henry Eyring, "A molecular dynamic theory of chromatography," The Journal of Physical Chemistry, vol. 59, no. 5, pp. 416-421,1955 [Online], May 1955. [5] Atila Felinger, "Molecular dynamic theories in chromatography," Journal of Chromatography A, vol. 1184, pp. 20-41, January 2008. [6] Rémi Pérenon, "Traitement de l’information en mode comptage appliqué aux détecteurs spectrométriques.," Université Joseph Fourier, Grenoble, France, PhD thesis 2013. [7] Pascal Szacherski, "Reconstruction de profils protéiques pour le recherches biomarqueurs," Université Bordeaux 1, Bordeaux, Phd thesis 2012.