{"id":78995,"date":"2026-11-18T12:01:37","date_gmt":"2026-11-18T10:01:37","guid":{"rendered":"https:\/\/www.iemn.fr\/?p=78995"},"modified":"2026-11-18T12:30:16","modified_gmt":"2026-11-18T10:30:16","slug":"78995","status":"publish","type":"post","link":"https:\/\/www.iemn.fr\/en\/a-la-une\/78995.html","title":{"rendered":"Ester MBINA &#8211; 19\/11\/2026 &#8211;  \u00ab\u00a0Performance Improvement of Flexible Organic Solar Cells Using Machine Learning\u00a0\u00bb"},"content":{"rendered":"<div id='layer_slider_1'  class='avia-layerslider main_color avia-shadow  avia-builder-el-0  el_before_av_heading  avia-builder-el-first  container_wrap sidebar_right'  style='height: 261px;'  ><div id=\"layerslider_58_r3sop6wsgkt\" data-ls-slug=\"homepageslider\" class=\"ls-wp-container fitvidsignore ls-selectable\" style=\"width:1140px;height:260px;margin:0 auto;margin-bottom: 0px;\"><div class=\"ls-slide\" data-ls=\"duration:6000;transition2d:5;\"><img loading=\"lazy\" decoding=\"async\" width=\"2600\" height=\"270\" src=\"https:\/\/www.iemn.fr\/wp-content\/uploads\/2019\/01\/sliders_news1.jpg\" class=\"ls-bg\" alt=\"\" srcset=\"https:\/\/www.iemn.fr\/wp-content\/uploads\/2019\/01\/sliders_news1.jpg 2600w, https:\/\/www.iemn.fr\/wp-content\/uploads\/2019\/01\/sliders_news1-300x31.jpg 300w, https:\/\/www.iemn.fr\/wp-content\/uploads\/2019\/01\/sliders_news1-768x80.jpg 768w, https:\/\/www.iemn.fr\/wp-content\/uploads\/2019\/01\/sliders_news1-1030x107.jpg 1030w, https:\/\/www.iemn.fr\/wp-content\/uploads\/2019\/01\/sliders_news1-1500x156.jpg 1500w, https:\/\/www.iemn.fr\/wp-content\/uploads\/2019\/01\/sliders_news1-705x73.jpg 705w\" sizes=\"auto, (max-width: 2600px) 100vw, 2600px\" \/><ls-layer style=\"font-size:14px;text-align:left;font-style:normal;text-decoration:none;text-transform:none;font-weight:700;letter-spacing:0px;border-style:solid;border-color:#000;background-position:0% 0%;background-repeat:no-repeat;width:180px;height:30px;left:0px;top:231px;line-height:32px;color:#ffffff;border-radius:6px 6px 6px 6px;padding-left:50px;background-color:rgba(0, 0, 0, 0.57);\" class=\"ls-l ls-ib-icon ls-text-layer\" data-ls=\"minfontsize:0;minmobilefontsize:0;\"><i class=\"fa fa-quote-right\" style=\"color:#ffffff;margin-right:0.8em;font-size:1em;transform:translateY( -0.125em );\"><\/i>ACTUALITES<\/ls-layer><\/div><\/div><\/div><div id='after_layer_slider_1'  class='main_color av_default_container_wrap container_wrap sidebar_right'  ><div class='container av-section-cont-open' ><div class='template-page content  av-content-small alpha units'><div class='post-entry post-entry-type-page post-entry-78995'><div class='entry-content-wrapper clearfix'>\n\n<style type=\"text\/css\" data-created_by=\"avia_inline_auto\" id=\"style-css-av-mwlybgs6-b4304537148b3bcc6bc972f382a28c0a\">\n#top .av-special-heading.av-mwlybgs6-b4304537148b3bcc6bc972f382a28c0a{\nmargin:0 0 10px 0;\npadding-bottom:4px;\n}\nbody .av-special-heading.av-mwlybgs6-b4304537148b3bcc6bc972f382a28c0a .av-special-heading-tag .heading-char{\nfont-size:25px;\n}\n.av-special-heading.av-mwlybgs6-b4304537148b3bcc6bc972f382a28c0a .av-subheading{\nfont-size:15px;\n}\n<\/style>\n<div  class='av-special-heading av-mwlybgs6-b4304537148b3bcc6bc972f382a28c0a av-special-heading-h2  avia-builder-el-1  el_after_av_layerslider  el_before_av_hr  avia-builder-el-first  av-linked-heading'><h2 class='av-special-heading-tag'  itemprop=\"headline\"  >THESE : Ester MBINA &#8211; \u00ab\u00a0Performance Improvement of Flexible Organic Solar Cells Using Machine Learning\u00a0\u00bb<\/h2><div class=\"special-heading-border\"><div class=\"special-heading-inner-border\"><\/div><\/div><\/div>\n\n<style type=\"text\/css\" data-created_by=\"avia_inline_auto\" id=\"style-css-av-18u73nj-dad6a947580930e400fc42ba200e80f1\">\n#top .hr.av-18u73nj-dad6a947580930e400fc42ba200e80f1{\nmargin-top:5px;\nmargin-bottom:5px;\n}\n.hr.av-18u73nj-dad6a947580930e400fc42ba200e80f1 .hr-inner{\nwidth:100%;\n}\n<\/style>\n<div  class='hr av-18u73nj-dad6a947580930e400fc42ba200e80f1 hr-custom  avia-builder-el-2  el_after_av_heading  el_before_av_textblock  hr-left hr-icon-no'><span class='hr-inner inner-border-av-border-thin'><span class=\"hr-inner-style\"><\/span><\/span><\/div>\n<section  class='av_textblock_section av-jriy64i8-fd5f2e9d63bf552d6910d12f255eb26e'   itemscope=\"itemscope\" itemtype=\"https:\/\/schema.org\/BlogPosting\" itemprop=\"blogPost\" ><div class='avia_textblock'  itemprop=\"text\" >\n<style type=\"text\/css\" data-created_by=\"avia_inline_auto\" id=\"style-css-av-13ewzjw-68e036126b913e5028f77311dc66b825\">\n.av_font_icon.av-13ewzjw-68e036126b913e5028f77311dc66b825{\ncolor:#bfbfbf;\nborder-color:#bfbfbf;\n}\n.av_font_icon.av-13ewzjw-68e036126b913e5028f77311dc66b825 .av-icon-char{\nfont-size:60px;\nline-height:60px;\n}\n<\/style>\n<span  class='av_font_icon av-13ewzjw-68e036126b913e5028f77311dc66b825 avia_animate_when_visible av-icon-style- avia-icon-pos-left avia-icon-animate'><span class='av-icon-char' aria-hidden='true' data-av_icon='\ue8c9' data-av_iconfont='entypo-fontello' ><\/span><\/span>\n<h3>Esther MBINA<\/h3>\n<p><strong>19 novembre 2026 \u00e0 10h00<br \/>\n<\/strong>Amphitheatre of the IEMN-Laboratoire central - Villeneuve d'Ascq<\/p>\n<\/div><\/section>\n<section  class='av_textblock_section av-jtefqx33-26fa44c544b10063385818eda37e027b'   itemscope=\"itemscope\" itemtype=\"https:\/\/schema.org\/BlogPosting\" itemprop=\"blogPost\" ><div class='avia_textblock'  itemprop=\"text\" ><div  class='hr av-kjh3zw-4dff888f744b728a1aca9b3a0971493a hr-default  avia-builder-el-6  avia-builder-el-no-sibling'><span class='hr-inner'><span class=\"hr-inner-style\"><\/span><\/span><\/div>\n<h5><strong><span style=\"color: #800000;\">Jury :<\/span><\/strong><\/h5>\n<ul>\n<li>Pr. Larbi BOUBCHIR, Professeur \u00e0 l\u2019Universit\u00e9 Paris 8, Rapporteur ;<\/li>\n<li>Dr. Sylvain VEDRAINE, Professeur associ\u00e9 HDR \u00e0 l\u2019Universit\u00e9 de Limoges, Rapporteur ;<\/li>\n<li>Dr. Christine LARTIGAU-DAGRON, Professeure associ\u00e9e \u00e0 l\u2019Universit\u00e9 de Pau, Examinatrice ;<\/li>\n<li>Dr. Isabelle SEGUY, Chercheuse au LAAS-CNRS, Examinatrice ;<\/li>\n<li>Pr. Kamal LMIMOUNI, Professeur \u00e0 l\u2019Universit\u00e9 de Lille, Examinateur ;<\/li>\n<li>Pr. Yaovi GAGOU, Professeur \u00e0 l\u2019Universit\u00e9 de Picardie Jules Verne, Examinateur ;<\/li>\n<li>Dr. Beno\u00eet HEINRICH, Ing\u00e9nieur de recherche \u00e0 l\u2019Universit\u00e9 de Strasbourg, Invit\u00e9 ;<\/li>\n<li>Dr. Kekeli N\u2019KONOU, Professeur associ\u00e9 \u00e0 JUNIA, Encadrant ;<\/li>\n<li>Dr. Bruno GRANDIDIER, Directeur de recherche \u00e0 JUNIA-IEMN-CNRS, Directeur de th\u00e8se.<\/li>\n<\/ul>\n<h5>Summary:<\/h5>\n<p>Les cellules solaires organiques (OSC, pour organic solar cells) pr\u00e9sentent un ensemble de propri\u00e9t\u00e9s remarquables, notamment une flexibilit\u00e9 m\u00e9canique, une semi-transparence, une grande l\u00e9g\u00e8ret\u00e9 et une compatibilit\u00e9 avec des proc\u00e9d\u00e9s de fabrication en solution \u00e0 basse temp\u00e9rature. Ces caract\u00e9ristiques les rendent particuli\u00e8rement adapt\u00e9es aux applications portables, aux textiles intelligents et \u00e0 l&rsquo;int\u00e9gration dans le b\u00e2timent. Cependant, leur optimisation se heurte \u00e0 un espace de param\u00e8tres vaste et fortement coupl\u00e9s comprenant la structure du donneur et de l&rsquo;accepteur, la morphologie nanom\u00e9trique de la couche active, les conditions de d\u00e9p\u00f4t, les interfaces et les couches de transport. L&rsquo;approche exp\u00e9rimentale classique par essais et erreurs \u00e9tant longue, co\u00fbteuse et limit\u00e9e, cette th\u00e8se propose de lever ce verrou en d\u00e9veloppant une plateforme d&rsquo;intelligence artificielle multi-\u00e9chelle pour guider leur d\u00e9veloppement.<\/p>\n<p>Le manuscrit s&rsquo;organise en quatre chapitres. Le chapitre 1 \u00e9tablit le contexte physique et mat\u00e9riel en introduisant le fonctionnement des OSC, l&rsquo;historique des architectures (des cellules de Schottky aux syst\u00e8mes modernes \u00e0 base de Y6) et les param\u00e8tres photovolta\u00efques cl\u00e9s servant de cibles de pr\u00e9diction, tout en d\u00e9crivant l&rsquo;espace chimique form\u00e9 par les \u00e9lectrodes, les couches interfaciales, les donneurs et les accepteurs. Le chapitre 2 d\u00e9taille la transition vers le domaine computationnel en passant en revue les m\u00e9thodes d&rsquo;apprentissage automatique (machine learning, ML) et profond (deep learning, DL). Il couvre l&rsquo;ensemble du flux de travail (donn\u00e9es, descripteurs, entra\u00eenement, validation, interpr\u00e9tation) et r\u00e9pond aux d\u00e9fis des jeux de donn\u00e9es OSC (taille restreinte, h\u00e9t\u00e9rog\u00e9n\u00e9it\u00e9 des mesures, besoin d&rsquo;interpr\u00e9tabilit\u00e9) pour proposer un cadre de d\u00e9cision guidant le choix des mod\u00e8les.<\/p>\n<p>Le chapitre 3 pr\u00e9sente une plateforme ML interpr\u00e9table pour le criblage de paires donneur\/accepteur (D\/A). \u00c0 partir de 1 575 paires exp\u00e9rimentales, l&rsquo;algorithme HistGradientBoosting obtient un coefficient de d\u00e9termination de R2 de 0,79. L&rsquo;analyse par SHAP et Boruta montre que les performances d\u00e9pendent non seulement des niveaux \u00e9lectroniques, mais aussi de la g\u00e9om\u00e9trie mol\u00e9culaire et des interactions des groupes fonctionnels. Ce mod\u00e8le est ensuite d\u00e9ploy\u00e9 en criblage virtuel \u00e0 haut d\u00e9bit pour identifier des paires donnant lieu \u00e0 un rendement de conversion \u00e9nerg\u00e9tique (PCE) sup\u00e9rieur \u00e0 20%.<\/p>\n<p>Le chapitre 4 \u00e9tend la mod\u00e9lisation au dispositif complet gr\u00e2ce au r\u00e9seau de neurones \u00e0 graphe (GNN) inform\u00e9 par la physique, nomm\u00e9 HELIOS. Ce mod\u00e8le int\u00e8gre le donneur, l&rsquo;accepteur, le solvant, l&rsquo;additif, la couche de transport d&rsquo;\u00e9lectrons (ETL) et la couche de transport de trous (HTL), en s&rsquo;appuyant sur une attention crois\u00e9e D\/A, une matrice d&rsquo;adjacence physique et le respect de contraintes thermodynamiques (comme la limite de la tension en circuit ouvert Voc). \u00c9valu\u00e9 sur 1 062 dispositifs complets, HELIOS atteint un R2 moyen de 0,66 pour le PCE avec une erreur absolue moyenne de 1,86%. Un criblage virtuel sur 4 millions de configurations (sur 2,3 milliards possibles) a permis d&rsquo;isoler 98 dispositifs avec un PCE pr\u00e9dit sup\u00e9rieur \u00e0 19,5%, dont un dispositif champion atteignant un PCE de 20,24% (combinaison du donneur PM1, de l&rsquo;accepteur G Dimer, de l&rsquo;ETL H75, de la HTL 2PACz, du tolu\u00e8ne comme solvant et du o-xyl\u00e8ne comme additif). L&rsquo;analyse de sensibilit\u00e9 montre que le choix des additifs et des couches de transport peut faire varier le PCE de plus d&rsquo;un point pour un m\u00eame couple D\/A.<\/p>\n<p>Enfin, la th\u00e8se se conclut par un bilan des r\u00e9sultats, une analyse critique des limites de la plateforme et des perspectives sur l&rsquo;avenir des plateformes de d\u00e9couverte de mat\u00e9riaux pilot\u00e9es par l&rsquo;IA pour le photovolta\u00efque organique.<\/p>\n<h5>Abstract:<\/h5>\n<p>Organic solar cells (OSCs) exhibit a remarkable combination of properties, including mechanical flexibility, semi-transparency, lightweight nature, and compatibility with low-temperature solution processing. These features make them particularly well-suited for portable applications, smart textiles, and building-integrated photovoltaics. However, their optimization is hindered by a vast and strongly coupled parameter space, which includes the donor and acceptor structures, the nanometric morphology of the active layer, deposition conditions, interfaces, and transport layers. Because the conventional trial-and-error experimental approach is time-consuming, expensive, and limited, this thesis addresses this bottleneck by developing a multi-scale artificial intelligence platform to guide their development.<\/p>\n<p>The manuscript is organized into four chapters. Chapter 1 establishes the physical and material context by introducing the operating principles of OSCs, the history of device architectures (from Schottky cells to modern Y6-based systems), and the key photovoltaic parameters used as prediction targets, while outlining the chemical space defined by electrodes, interfacial layers, donors, and acceptors. Chapter 2 details the transition to the computational domain by reviewing machine learning (ML) and deep learning (DL) methodologies. It covers the complete workflow (data collection, descriptor engineering, training, validation, interpretation) and addresses the specific challenges of OSC datasets (limited sample sizes, measurement heterogeneity, need for physical interpretability) to establish a decision framework that guides the selection oof regression model.<\/p>\n<p>Chapter 3 presents an interpretable ML platform for screening donor\/acceptor (D\/A) pairs. Evaluated on 1,575 experimental pairs, the HistGradientBoosting algorithm achieves a determination coefficient R2 of 0.79. SHAP and Boruta feature selection reveal that performance depends not only on electronic energy levels, but also on molecular geometry and functional group interactions with the environment. This model is then deployed for high-throughput virtual screening to identify D\/A combinations with a predicted power conversion efficiency (PCE) above 20%.<\/p>\n<p>Chapter 4 extends the modeling framework to the complete device using a physics-informed graph neural network (GNN) named HELIOS. This model simultaneously considers the donor, acceptor, solvent, additive, electron transport layer (ETL), and hole transport layer (HTL), leveraging D\/A cross-attention, a physical adjacency matrix, and thermodynamic constraints (such as the open-circuit voltage Voc limit). Evaluated on 1,062 complete devices, HELIOS achieves an average R2 of 0.66 for PCE with a mean absolute error of 1.86%. Virtual screening across 4 million configurations (out of 2.3 billion possibilities) identified 98 devices with a predicted PCE exceeding 19.5%, including a top champion device reaching a predicted PCE of 20.24% (combining PM1 donor, G Dimer acceptor, H75 ETL, 2PACz HTL, toluene solvent, and o-xylene additive). Multi-component sensitivity analysis demonstrates that selecting different transport layers and additives can alter the PCE by over one percentage point for the exact same D\/A pair.<\/p>\n<p>Finally, the thesis concludes with a synthesis of the main results, a critical evaluation of the current platform&rsquo;s limitations, and prospective views on the future of AI-driven material discovery platforms for organic photovoltaics.<\/p>\n<\/div><\/section>","protected":false},"excerpt":{"rendered":"","protected":false},"author":2,"featured_media":71239,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3,8],"tags":[],"class_list":["post-78995","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-a-la-une","category-actualites"],"_links":{"self":[{"href":"https:\/\/www.iemn.fr\/en\/wp-json\/wp\/v2\/posts\/78995","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.iemn.fr\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.iemn.fr\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.iemn.fr\/en\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.iemn.fr\/en\/wp-json\/wp\/v2\/comments?post=78995"}],"version-history":[{"count":7,"href":"https:\/\/www.iemn.fr\/en\/wp-json\/wp\/v2\/posts\/78995\/revisions"}],"predecessor-version":[{"id":78997,"href":"https:\/\/www.iemn.fr\/en\/wp-json\/wp\/v2\/posts\/78995\/revisions\/78997"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.iemn.fr\/en\/wp-json\/wp\/v2\/media\/71239"}],"wp:attachment":[{"href":"https:\/\/www.iemn.fr\/en\/wp-json\/wp\/v2\/media?parent=78995"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.iemn.fr\/en\/wp-json\/wp\/v2\/categories?post=78995"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.iemn.fr\/en\/wp-json\/wp\/v2\/tags?post=78995"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}