{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_12_fig_12.jpg","caption":"Fig. 12. Atomic force microscope image of a  $\\sim 320\\mathrm{A}$  thick tungsten film deposited at  $425\\mathrm{K}$  after 125 AB cycles. The  $\\mathrm{WF}_6$  and  $\\mathrm{Si}_2\\mathrm{H}_6$  reactant exposures were sufficient for complete half-reactions. The light-to-dark range is  $25\\mathrm{\\AA}$","id":"train/atomic-layer-deposition/experimental-usecase/12/fig_12","sample_id":"atomic-layer-deposition/experimental-usecase/12/fig_12","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":4,"y":0,"width":641,"height":342}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/12/images/fig_12.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/12/images/fig_12.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/12/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/12/J.W. Klaus et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"12","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":650,"height":342,"image_format":"jpeg","image_sha256":"d4b9b612c83f0a5c0e7cad08214bd57cb131e40fa8a7541d0b43835c125187dd","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_13_fig_4.jpg","caption":"FIG. 4. Cross-sectional SEM micrograph of a  $20~\\mathrm{nm}$  Ti-Si-N film grown by MOALD at  $180^{\\circ}\\mathrm{C}$ . Step coverage of the  $\\mathrm{Ti - Si - N}$  film is approximately  $100\\%$  even on the  $0.3\\mu \\mathrm{m}$  diam hole with slightly negative slope and 10:1 aspect ratio.","id":"train/atomic-layer-deposition/experimental-usecase/13/fig_4","sample_id":"atomic-layer-deposition/experimental-usecase/13/fig_4","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":2,"y":0,"width":503,"height":527}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/13/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/13/images/fig_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/13/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/13/Jae-Sik Min et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"13","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":508,"height":530,"image_format":"jpeg","image_sha256":"42b0f3022b98e14b6599dc9d9310e526283e29cc136e12497f0e0a5cc424dba3","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_16_fig_8.jpg","caption":"FiG. 8. (a) TEM image of the MIM stack and (b) a close-up of the  $\\mathrm{ZrO_2}$  film.  $\\mathrm{ZrO_2}$  was deposited with the  $\\mathrm{Zr(Me_5Cp)(TEA) / O_3}$  process at  $300^{\\circ}\\mathrm{C}$ .","id":"train/atomic-layer-deposition/experimental-usecase/16/fig_8","sample_id":"atomic-layer-deposition/experimental-usecase/16/fig_8","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":21,"y":39,"width":620,"height":625},{"panel_id":"b","x":37,"y":700,"width":604,"height":286}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/16/images/fig_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/16/images/fig_8.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/16/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/16/Sanni Seppala et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"16","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":644,"height":992,"image_format":"jpeg","image_sha256":"1823f2f31571be9dd480ab4a3a14a6bd9b73f836c910ebe630f9cd3329304f9d","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_2_figure_5.jpg","caption":"Figure 5. SEM images of (a) 50- and (b) 110-nm-thick platinum films.","id":"train/atomic-layer-deposition/experimental-usecase/2/figure_5","sample_id":"atomic-layer-deposition/experimental-usecase/2/figure_5","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Subfigure (a) shows the surface morphology of a ~50 nm ALD-grown Pt film, characterized by a dense population of small, bright Pt nanoparticles distributed across the substrate. The grains appear closely packed, indicating early coalescence but with significant nanoscale roughness remaining.\"},{\"panel_id\":\"b\",\"text\":\"Subfigure (b) displays the morphology of a thicker ~110 nm Pt film, where the particles appear larger and more coalesced, consistent with grain growth through continued ALD cycling. The increased thickness promotes lateral grain expansion and partial merging, yielding a smoother but still nanostructured Pt layer. Together, the images illustrate the thickness-dependent evolution of Pt film morphology typical for ALD-grown noble metals.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"-Larger Pt grains are present in subfigure (b), indicating extended coalescence.\\n-The particle density visually decreases as grains merge with increasing thickness.\\n-The surface appears smoother at the microscale in subfigure (b) than in (a).\\n-Subfigure (a) shows more discrete, fine nanoparticles, while subfigure (b) shows more mature, merged grain structures.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Electrical conductivity, due to reduced electron scattering at grain boundaries.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"During the early stages of Pt ALD, individual adsorption–reduction events produce isolated nuclei that progressively grow laterally as cycles accumulate. With more cycles, precursor adsorption on existing Pt clusters becomes increasingly favorable, accelerating coalescence. Over time, these nuclei merge into larger grains, reducing the density of isolated particles and promoting film continuity. Subfigure (a) captures an intermediate stage where coalescence has begun but complete grain merging has not yet occurred.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The finer nanoparticle structure in the thinner film provides a higher surface-to-volume ratio and a larger density of exposed active sites, which can enhance catalytic activity. Although electrical connectivity may be lower, catalytic performance often benefits from high surface area rather than continuous film coverage.\"}]}]","bbox":[{"panel_id":"a","x":45,"y":-4,"width":623,"height":505},{"panel_id":"b","x":47,"y":530,"width":621,"height":494}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/images/figure_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/images/figure_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/Aaltonen et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"2","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":669,"height":1025,"image_format":"jpeg","image_sha256":"fbe01f5056199626e32fe5964c3b1239ad37cf8e634451b91a64fd671bd794cf","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_2_figure_6.jpg","caption":"Figure 6. AFM images of (a) 50- and (b) 110-nm-thick platinum films. Please note that the images have different  $\\mathcal{Z}$  -axis scales.","id":"train/atomic-layer-deposition/experimental-usecase/2/figure_6","sample_id":"atomic-layer-deposition/experimental-usecase/2/figure_6","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Subfigure (a) shows a 3D AFM height map of a ~50 nm ALD-grown Pt film. The topology consists of rounded hill-valley structures with moderate amplitude, indicating partially coalesced grains that have not yet reached full surface continuity.\"},{\"panel_id\":\"b\",\"text\":\"Subfigure (b) presents a ~110 nm Pt film whose AFM landscape exhibits sharper protrusions and a larger vertical scale, consistent with continued grain growth and columnar thickening typical of noble-metal ALD. The difference in z-axis scaling between (a) and (b) highlights that the thicker film exhibits significantly higher peak-to-valley contrast, reflecting its rougher nanoscale texture. Together, the images illustrate thickness-dependent morphological evolution characteristic of Pt ALD.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Subfigure (a) shows rounded, laterally spread hill-valley structures. Subfigure (b) exhibits sharper, taller protrusions. This change indicates that lateral grain expansion has saturated, and additional ALD cycles preferentially increase vertical grain height (columnar thickening).\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Local electric field distribution, which can affect charge injection, field emission, and catalytic reaction rates.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"In ALD, each reaction cycle deposits material in a surface-saturating manner, limiting uncontrolled vertical growth. However, local variations in nucleation density and surface diffusion lead to uneven lateral growth across grains. This produces a uniformly covered but moderately corrugated surface, as seen in subfigure (a), where height variations arise from partially coalesced grains rather than runaway growth.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"By correlating roughness trends with film thickness, engineers can tune ALD conditions to suppress excessive vertical growth when smooth films are required or enhance roughness when high surface area is desired. This morphological feedback loop is essential for tailoring Pt films for specific applications like catalysis or microelectronics.\"}]}]","bbox":[{"panel_id":"a","x":9,"y":0,"width":615,"height":417},{"panel_id":"b","x":7,"y":480,"width":617,"height":429}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/images/figure_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/Aaltonen et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"2","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":625,"height":911,"image_format":"jpeg","image_sha256":"bb0b2c4038d4d38e779f2cead807a6e9f18e6b0746cfdb6da5001f2c58630928","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_22_fig_5.jpg","caption":"Fig. 5. FESEM images of LiF films deposited at various temperatures on  $\\mathrm{SiO_2 / Si(111)}$  substrates. Deposition temperatures and film thicknesses are; a)  $250^{\\circ}\\mathrm{C}$ $111\\mathrm{nm}$ , b)  $275^{\\circ}\\mathrm{C}$ $97\\mathrm{nm}$ , c)  $300^{\\circ}\\mathrm{C}$ $88\\mathrm{nm}$ , d)  $325^{\\circ}\\mathrm{C}$ $73\\mathrm{nm}$ , and e)  $350^{\\circ}\\mathrm{C}$ $90\\mathrm{nm}$ . Lithd and  $\\mathrm{TiF_4}$  pulse lengths were  $2\\mathrm{s}$  and purge times between these pulses were  $4\\mathrm{s}$ .","id":"train/atomic-layer-deposition/experimental-usecase/22/fig_5","sample_id":"atomic-layer-deposition/experimental-usecase/22/fig_5","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"},{"panel_id":"e","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"d","x":657,"y":469,"width":614,"height":469},{"panel_id":"b","x":647,"y":0,"width":619,"height":467},{"panel_id":"c","x":6,"y":466,"width":647,"height":474},{"panel_id":"e","x":321,"y":252,"width":637,"height":474},{"panel_id":"a","x":6,"y":0,"width":637,"height":456}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/22/images/fig_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/22/images/fig_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/22/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/22/Miia Mantymaki et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"22","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":1272,"height":947,"image_format":"jpeg","image_sha256":"36ccda8e597b98e154ac6b402bf6153f1e4ca9164664dba6ece0aeff3d8d93de","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_22_fig_6.jpg","caption":"Fig. 6. AFM images of LiF films deposited at various temperatures. Deposition temperatures, films thicknesses, and rms roughnesses are; a)  $250^{\\circ}\\mathrm{C}$  111 nm 6.3 nm, b)  $275^{\\circ}\\mathrm{C}$  97 nm 9.5 nm, c)  $300^{\\circ}\\mathrm{C}$  88 nm 12.3 nm, d)  $325^{\\circ}\\mathrm{C}$  73 nm 15.9 nm, and e)  $350^{\\circ}\\mathrm{C}$  90 nm 20.6 nm. Notice that the vertical axis has a different scale in image e). Lithd and  $\\mathrm{TiF_4}$  pulse lengths were 2 s and purge times between these pulses were 4 s.","id":"train/atomic-layer-deposition/experimental-usecase/22/fig_6","sample_id":"atomic-layer-deposition/experimental-usecase/22/fig_6","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"},{"panel_id":"e","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"d","x":692,"y":320,"width":580,"height":353},{"panel_id":"b","x":668,"y":10,"width":604,"height":320},{"panel_id":"c","x":0,"y":345,"width":557,"height":345},{"panel_id":"e","x":338,"y":591,"width":590,"height":348},{"panel_id":"a","x":3,"y":5,"width":587,"height":321}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/22/images/fig_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/22/images/fig_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/22/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/22/Miia Mantymaki et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"22","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":1275,"height":945,"image_format":"jpeg","image_sha256":"a799c8351fc2a9ecca9413076019e54cbe35cdb41b2ebafffaf983263a814df7","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_25_figure_7.jpg","caption":"Figure 7. SEM micrographs of the  $\\mathrm{ZnO:Al / ZnO / ESL / ACIGS}$  stacks, where the ESL is (a)  $\\mathrm{ZnO}$ , (b) ZGO 0.12, (c) ZGO 0.33, (d)  $\\mathrm{SnO}_{y}$ , and (e) TGO 0.10. The scale bar represents  $200~\\mathrm{nm}$  in (a) and (b) and  $100~\\mathrm{nm}$  in (c)–(e). (f) Schematic illustration of the  $\\mathrm{ZnO:Al / ZnO / ESL / ACIGS}$  stack.","id":"train/atomic-layer-deposition/experimental-usecase/25/figure_7","sample_id":"atomic-layer-deposition/experimental-usecase/25/figure_7","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"},{"panel_id":"e","label":"image panel"},{"panel_id":"f","label":"device structure diagram"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"d","x":8,"y":197,"width":277,"height":183},{"panel_id":"b","x":285,"y":6,"width":278,"height":188},{"panel_id":"c","x":565,"y":5,"width":274,"height":187},{"panel_id":"e","x":285,"y":196,"width":278,"height":184},{"panel_id":"f","x":563,"y":196,"width":277,"height":186},{"panel_id":"a","x":9,"y":8,"width":276,"height":184}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/25/images/figure_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/25/images/figure_7.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/25/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/25/Adam Hultqvist et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"25","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":842,"height":380,"image_format":"jpeg","image_sha256":"859779510c12807ca1c1a92d8e344924efc298856f5f1f33ebb3b2bba2c5f5d7","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_26_figure_11.jpg","caption":"Figure 11. AFM image for  $89\\mathrm{-nm}$  ALD ITO film deposited on Si-100 at  $275^{\\circ}C$  using 600 cycles with  $5\\%$ $\\mathrm{SnO_2}$  cycles.","id":"train/atomic-layer-deposition/experimental-usecase/26/figure_11","sample_id":"atomic-layer-deposition/experimental-usecase/26/figure_11","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":0,"y":0,"width":397,"height":420}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_11.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_11.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/Jeffrey W. Elam et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"26","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":397,"height":420,"image_format":"jpeg","image_sha256":"c4a9ed1be9fd8d94c2c6987402084adc64b194b1453ea50003da5583ccee249a","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_26_figure_12.jpg","caption":"Figure 12. SEM plan-view (a) and cross-sectional (b) images of ALD ITO film deposited on Si(100) using 600 cycles with  $5\\%$ $\\mathrm{SnO_2}$  cycles at  $275^{\\circ}C$ .","id":"train/atomic-layer-deposition/experimental-usecase/26/figure_12","sample_id":"atomic-layer-deposition/experimental-usecase/26/figure_12","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":7,"y":6,"width":482,"height":334},{"panel_id":"b","x":7,"y":344,"width":481,"height":232}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_12.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_12.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/Jeffrey W. Elam et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"26","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":492,"height":580,"image_format":"jpeg","image_sha256":"a254805a2bc76a75128f8ae336a15913e26fbbfafecca6c775175669576fb752","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_26_figure_13.jpg","caption":"Figure 13. (a) Cross-sectional SEM image of AAO membrane coated conformally with  $9.7\\mathrm{-nm}$  ALD ITO film, and also with  $161\\mathrm{-nm}$  ALD ITO film deposited selectively on the AAO front surface visible on the left side of the image. (b) Higher-resolution SEM image of AAO membrane showing that the  $20\\mathrm{-nm}$  pores are sealed, and ITO nanocrystals decorate the inner walls of the  $200\\mathrm{-nm}$  pores.","id":"train/atomic-layer-deposition/experimental-usecase/26/figure_13","sample_id":"atomic-layer-deposition/experimental-usecase/26/figure_13","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":4,"y":0,"width":493,"height":422},{"panel_id":"b","x":531,"y":4,"width":445,"height":417}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_13.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_13.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/Jeffrey W. Elam et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"26","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":980,"height":425,"image_format":"jpeg","image_sha256":"21cbf9025442a856e62d855bda9b114c9cb3e5e6a00154782f95079c4005107f","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_27_fig_4.jpg","caption":"Fig. 4. XTEM image of TiN film deposited by remote PEALD method on contact hole approximately  $0.25\\mu \\mathrm{m}$  wide and  $2.5\\mu \\mathrm{m}$  deep.","id":"train/atomic-layer-deposition/experimental-usecase/27/fig_4","sample_id":"atomic-layer-deposition/experimental-usecase/27/fig_4","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The image shows a microscopic view of TiN-coated structures.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The image shows similar TiN thickness's on the sidewalls of the 3D structures as well as on top, indicating good conformality.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The aspect ratio is 10.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The top edges show chamfered edges indicating that edge effects do play a role in the deposition at these locations.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The process is likely radical driven as the sidewalls, that are not influenced by the ions, show similar thickness's as the top and bottom surface.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":4,"width":547,"height":667}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/27/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/27/images/fig_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/27/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/27/Ju Youn KIM et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"27","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":553,"height":672,"image_format":"jpeg","image_sha256":"c6d88db15993326bead7cfb61844b04a3e53c3f73c4f8958df8a3bd280bb0044","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_28_figure_5.jpg","caption":"Figure 5. SEM (top) and AFM (bottom) images of the samples grown through (a), (c) the conventional ALD process and (b), (d) the ALD-DFM process on a  $\\mathrm{SiO}_2$  substrate.","id":"train/atomic-layer-deposition/experimental-usecase/28/figure_5","sample_id":"atomic-layer-deposition/experimental-usecase/28/figure_5","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":45,"y":27,"width":292,"height":285},{"panel_id":"b","x":391,"y":29,"width":274,"height":283},{"panel_id":"c","x":59,"y":370,"width":272,"height":268},{"panel_id":"d","x":390,"y":370,"width":277,"height":268}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/images/figure_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/images/figure_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/Woohyun Kim et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"28","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":666,"height":642,"image_format":"jpeg","image_sha256":"46e587451bfcbe8cbbd2d921045263bf1fd8ca18561f32b81a680d00917dea4c","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_28_figure_8.jpg","caption":"Figure 8. TEM images (left) and associated FFT analysis (right) of the samples grown through (a) the conventional ALD process and (b) the ALD-DFM process.","id":"train/atomic-layer-deposition/experimental-usecase/28/figure_8","sample_id":"atomic-layer-deposition/experimental-usecase/28/figure_8","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":52,"y":15,"width":617,"height":282},{"panel_id":"b","x":54,"y":347,"width":611,"height":263}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/images/figure_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/images/figure_8.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/Woohyun Kim et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"28","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":672,"height":614,"image_format":"jpeg","image_sha256":"b4f086a8f80deb6c58ad6f9c092dd1ce898c37aeb5578109a8d29a298bd16ab7","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_30_1b78016cb355e70dc8f9df6a158ffa618b7d7b3846cf68bd31bde9dd7bfc1aaf.jpg","caption":"","id":"train/atomic-layer-deposition/experimental-usecase/30/1b78016cb355e70dc8f9df6a158ffa618b7d7b3846cf68bd31bde9dd7bfc1aaf","sample_id":"atomic-layer-deposition/experimental-usecase/30/1b78016cb355e70dc8f9df6a158ffa618b7d7b3846cf68bd31bde9dd7bfc1aaf","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The image shows a microscopic view of a sample tested under the 'Side-by-Side Setup'.\"},{\"panel_id\":\"b\",\"text\":\"The image shows a microscopic view of a sample tested under the 'Face-to-Face Setup'.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Based on the images it seems that the surface is rough.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Usually ALD and ALE applications involve extremely thin features and thickness's for which optical microscopy is not suitable.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Not every location on the setup has the same color indicating that there is non-uniformity.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"yes the probe is damaged in the corners.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":306,"height":277},{"panel_id":"b","x":350,"y":0,"width":305,"height":277}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/30/images/1b78016cb355e70dc8f9df6a158ffa618b7d7b3846cf68bd31bde9dd7bfc1aaf.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/30/images/1b78016cb355e70dc8f9df6a158ffa618b7d7b3846cf68bd31bde9dd7bfc1aaf.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/30/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/30/Kaupo Kukli et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"30","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"not_found"},"width":655,"height":277,"image_format":"jpeg","image_sha256":"d7d8ce8e0ec628079599d1f711682c8ed12dbf270cb19903549dfe852dd2f631","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_30_figure_6.jpg","caption":"Figure 6. Bird-eye scanning electron microscopy images of  $\\mathrm{Fe_2O_3}$  (topmost panel) and  $\\mathrm{Fe_3O_4 - MgO}$  films  $2^{\\mathrm{nd}}$  to  $4^{\\mathrm{th}}$  panel from top) grown using different cycle sequences indicated by labels. The films were grown from  $\\mathrm{CpFeC_5H_4CHN(CH_3)_2}$  at  $425^{\\circ}\\mathrm{C}$ . Film thicknesses and  $\\mathrm{Mg}$  to  $\\mathrm{Fe}$  atomic ratios are also given by labels.","id":"train/atomic-layer-deposition/experimental-usecase/30/figure_6","sample_id":"atomic-layer-deposition/experimental-usecase/30/figure_6","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The pure Fe₂O₃ film shows large, faceted grains with sharp edges. The surface is rough and loosely packed. This morphology is typical of high-temperature hematite growth without modifiers.\"},{\"panel_id\":\"b\",\"text\":\"With a small amount of Mg added, the grains become smaller and less faceted. The surface looks more compact than in (a). Grain boundaries are still visible but less pronounced.\"},{\"panel_id\":\"c\",\"text\":\"At higher Mg content, the grain size is strongly reduced. The surface appears dense and uniform, with rounded features. Grain coarsening is clearly suppressed.\"},{\"panel_id\":\"d\",\"text\":\"At the highest Mg/Fe ratio, the surface consists of very fine, closely packed grains. Individual grain shapes are hard to distinguish. This indicates strong inhibition of surface diffusion and crystallite growth.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"},{\"panel_id\":\"c\",\"text\":\"\"},{\"panel_id\":\"d\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The roughest film is the non-doped film grown with Fe2O3\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Measurement: top-down SEM (“bird-eye”) surface imaging of the films\\n\\nScale bar: 1.00 µm in each panel\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The rough topography in (a) is consistent with growth dominated by separated columns/grains/pillars with voids between them. In (b), the higher packing density indicates more nucleation sites and/or tighter column spacing. In (c–d), the continuity and smoothness suggest the voided columnar morphology is suppressed, producing a more coalesced film surface.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"A smoother, more continuous film typically improves thickness uniformity and reduces extreme topographic variations that can complicate electrical contacting or multilayer integration. It can also reduce variability in local magnetic response caused by large voids/isolated pillars. In this study, morphology is treated as a key co-variable alongside phase content when interpreting magnetic performance trends.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"-Panel (a): large, plate-like, faceted grains\\n-Panel (d): fine, rounded grains with high packing density\\n-Panel (a): broader grain-size distribution\\n-Panel (d): narrow grain-size distribution\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It seems that iron oxide grows both lateraly as well as vertically, indicating there is no preferred orientation.\"}]},{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Changes in the roughness can be observed as a function of the Mg/Fe ratio.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Roughness increases as the Mg/Fe ratio increases. The highest level of roughness is produced at the highest Mg/Fe ratio.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Increasing the Mg/Fe ratio seems to result in lower thicknesses and therefore lower GPCs. This could be an effect of nucleation delay of iron oxide growing on top of Magnesium oxide.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Increasing MgO insertion suppresses grain growth, driving a transition from large faceted grains to finer, more rounded features. This indicates reduced surface diffusion and interrupted crystallite coalescence during growth. The effect strengthens monotonically with Mg/Fe ratio, consistent with Mg acting as a kinetic growth modifier.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"-Catalysis requiring high surface area\\n-Gas-sensing layers\\n-Electrochemical interfaces\\n-Diffusion-barrier coatings\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The non-doped film shows a pillar/grain-like rough surface with large voids. With small Mg addition (Mg/Fe ≈ 0.02), the surface becomes denser, and the voids shrink. At higher Mg/Fe (≈0.23–0.38), the films become more continuous and smoother, with fewer large surface openings.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The morphology of Fe2O3 films directly affects their functional properties, such as catalytic activity, magnetic behavior, and surface reactivity. By controlling MgO doping, the film’s surface continuity, pillar density, and thickness can be tuned, which can improve performance in applications where uniformity, smoothness, or specific surface features are critical.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"In subfigure a, more height differences can be observed, indicating that the film consists of more voids, resulting in a lower density.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The surface roughness decreases as the MgO content increases.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Atomic force microscopy could be used, with this method actual height differences are identified.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Reduced grain size with rounded grain boundaries.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Increased density of grains/pillars, Reduced void width, Smoother and more continuous surface, Decreased overall film thickness at high MgO levels\"}]}]","bbox":[{"panel_id":"a","x":4,"y":4,"width":394,"height":300},{"panel_id":"b","x":4,"y":312,"width":394,"height":292},{"panel_id":"c","x":6,"y":618,"width":390,"height":294},{"panel_id":"d","x":7,"y":923,"width":384,"height":283}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/30/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/30/images/figure_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/30/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/30/Kaupo Kukli et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"30","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":403,"height":1217,"image_format":"jpeg","image_sha256":"162c5d71a290bb63b1be2c5144e573d301958bd2891dbce6208465cb3d4452cb","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_30_figure_7.jpg","caption":"Figure 7. Cross section SEM/EDX images from a  $72~\\mathrm{nm}$  thick film grown from  $\\mathrm{CpFeC_5H_4CHN(CH_3)_2}$  at  $375^{\\circ}C$  usin 500 ALD cycles on 3D substrates. The elements detected are denoted by labels.","id":"train/atomic-layer-deposition/experimental-usecase/30/figure_7","sample_id":"atomic-layer-deposition/experimental-usecase/30/figure_7","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The cross-section SEM image shows a continuous film coating the 3D structure. The layer follows both the top surface and the slanted sidewalls. No obvious thinning or breaks are visible.\"},{\"panel_id\":\"b\",\"text\":\"The Si EDX map highlights the underlying substrate. The signal is strong below the film and drops sharply inside the coated region. This clearly separates substrate from deposited layer.\"},{\"panel_id\":\"c\",\"text\":\"The Fe EDX map outlines the film region across the full cross section. The signal follows the same geometry seen in the SEM image. This indicates uniform iron incorporation along sidewalls and horizontal surfaces.\"},{\"panel_id\":\"d\",\"text\":\"The O EDX map closely overlaps with the Fe distribution. Oxygen is present throughout the same region as iron. This confirms that the deposited layer is an iron oxide film rather than metallic Fe.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"},{\"panel_id\":\"c\",\"text\":\"\"},{\"panel_id\":\"d\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Si.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Silicon and oxygen are more or less homogeneously distributed in the areas where they are supposed to be. Iron on the other side shows fluctuations in density, even in the iron oxide layer.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Iron (Fe), and oxygen (O).\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The image shows the Fe content.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"No, the Fe appears to be closest to the surface of the film.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The film has a higher O content than the Fe, so the white color in the figure is more intense for the first case.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"In the figure it can be seen that iron atoms are present all through the sample, even in the silicon substrate. This indicates that iron diffuses into the substrate.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Uniform elemental incorporation across the 3D structure.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The Fe map shows signal not only near the trench opening but also extending down the trench, including the bottom region. Additionally, the O map shows a corresponding distribution, consistent with an iron oxide film rather than isolated contamination. Together with the SEM cross-section, this supports the conclusion that the film can be grown conformally into high aspect-ratio (~1:20) trenches using the same cycle timing as on planar substrates.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Conformal film growth along trench walls and bottoms, Uniform thickness of the iron oxide layer (~72 nm), Depth-resolved elemental distribution of Fe and O\"}]},{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"At the interface between the silicon substrate and iron oxide a different density of oxygen atoms is measured than in the bulk iron oxide. This indicates that there is likely a SiO2 interface layer.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The Oxygen signal (panel d) closely matches the spatial distribution of the Iron signal (panel c). Both signals are concentrated in the corrugated top layer, confirming that the deposited film is an iron oxide compound (likely Fe₂O₃ or Fe₃O₄) rather than pure iron.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The figure shows that ALD-grown iron oxide films can coat complex 3D structures uniformly, maintaining consistent thickness and conformality even in deep trenches. The EDX mappings confirm that both iron and oxygen are present throughout the film depth, indicating that the deposition process achieves complete coverage and stoichiometry. This suggests that such films could be reliably used in devices requiring high surface-area coatings or patterned 3D architectures.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"From the 3D structure upwards vertical lines can be identified which can be grainboundaries or local density fluctuations.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The film thickness appears continuous along both horizontal and inclined surfaces, with no visible thinning at corners or sidewalls. This indicates that precursor transport and surface reactions occur uniformly throughout the 3D geometry. Such behavior is consistent with self-limiting ALD chemistry enabling conformal coverage of high-aspect-ratio features.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Modern semiconductor devices, such as FinFETs or 3D NAND flash, rely on complex, non-planar architectures to increase transistor density and performance. The ability to deposit functional materials (dielectrics, conductors) uniformly over these vertical and recessed features ensures consistent electrical behavior and prevents defects like voids or electrical shorts.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"(a) Cross-section SEM image of the trench + film\\n\\n(b) Si EDX map (substrate/trench geometry)\\n\\n(c) Fe EDX map (film penetration/conformality)\\n\\n(d) O EDX map (oxide co-location with Fe)\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"ALD proceeds through self-limiting surface reactions, which helps maintain uniform growth even when reactant transport is constrained in narrow features. If precursor exposure and purge are sufficient, each surface (top, sidewalls, bottom) can reach similar surface saturation per cycle. The observed Fe/O presence along the trench depth is consistent with that expectation for ALD-type conformal coverage.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Many target applications for functional oxides (including magnetic or catalytic coatings) involve 3D, high-surface-area structures where performance depends on coating all internal surfaces. Conformal growth ensures the properties measured on planar samples can plausibly translate to 3D integration. This figure provides direct evidence that the process conditions used here can coat deep trenches rather than only planar substrates.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":5,"y":12,"width":321,"height":301},{"panel_id":"b","x":347,"y":4,"width":318,"height":304},{"panel_id":"c","x":6,"y":319,"width":324,"height":301},{"panel_id":"d","x":343,"y":321,"width":320,"height":297}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/30/images/figure_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/30/images/figure_7.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/30/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/30/Kaupo Kukli et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"30","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":667,"height":623,"image_format":"jpeg","image_sha256":"d5e0f2216e10c78772a9ebf2800cd114ff50381ed9754844acbede17b80cdbc6","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_33_figure_6.jpg","caption":"Figure 6 Plane-view SEM of ALD Cu on  $\\mathrm{SiO}_2$  substrate (a) and  $\\mathrm{Co / SiO}_2$  substrate (b).","id":"train/atomic-layer-deposition/experimental-usecase/33/figure_6","sample_id":"atomic-layer-deposition/experimental-usecase/33/figure_6","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":1,"y":32,"width":348,"height":346},{"panel_id":"b","x":352,"y":33,"width":338,"height":343}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/33/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/33/images/figure_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/33/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/33/BOOYONG S. LIM et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"33","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":694,"height":381,"image_format":"jpeg","image_sha256":"dae32a07cf36041c8f491709c49071e67caaabc7d9523219a07f4801cbfd000c","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_33_figure_7.jpg","caption":"Figure 7 Cross-sectional SEM of ALD Cu/Co inside holes with an aspect ratio of about 10:1.","id":"train/atomic-layer-deposition/experimental-usecase/33/figure_7","sample_id":"atomic-layer-deposition/experimental-usecase/33/figure_7","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":0,"y":0,"width":586,"height":678}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/33/images/figure_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/33/images/figure_7.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/33/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/33/BOOYONG S. LIM et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"33","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":586,"height":678,"image_format":"jpeg","image_sha256":"c059c49380314b5eb7043c6e0694ae72508fed8c7d7c03fc76cfdca9c1c2fdf5","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_36_fig_7.jpg","caption":"Fig. 7 Atomic force microscopy images of the conventional HEMT and the  $\\mathrm{Ga}_{2}\\mathrm{O}_{3}$  MOS-HEMT","id":"train/atomic-layer-deposition/experimental-usecase/36/fig_7","sample_id":"atomic-layer-deposition/experimental-usecase/36/fig_7","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows an atomic force microscopy image of the conventional HEMT.\"},{\"panel_id\":\"b\",\"text\":\"The figure shows an atomic force microscopy image of the Ga2O3 MOS-HEMT.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The conventional HEMT with a RMS value of 0.532 nm.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The Ga2O3 MOS-HEMT.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The Ga2O3 MOS-HEMT has the lowest RMS value compared to the conventional AlGaN HEMT.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"No, this could mean that both surfaces are smooth.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":9,"width":464,"height":474},{"panel_id":"b","x":489,"y":12,"width":461,"height":471}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/36/images/fig_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/36/images/fig_7.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/36/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/36/Huan-Yu Shih et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"36","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":950,"height":483,"image_format":"jpeg","image_sha256":"aa54b0c0b76bac9d52a5df9d740af4f7bbbdb7c262ce42b588a9e7d7cb782d9d","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_4_fig_6.jpg","caption":"Fig. 6. AFM images of the  $\\mathrm{HfO_2}$  films grown onto Si(100) at a)  $200^{\\circ}\\mathrm{C}$ , and b)  $300^{\\circ}\\mathrm{C}$ .","id":"train/atomic-layer-deposition/experimental-usecase/4/fig_6","sample_id":"atomic-layer-deposition/experimental-usecase/4/fig_6","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":12,"y":15,"width":641,"height":435},{"panel_id":"b","x":12,"y":479,"width":641,"height":449}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/4/images/fig_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/4/images/fig_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/4/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/4/Kukli et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"4","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":653,"height":928,"image_format":"jpeg","image_sha256":"7d82ab8e4ba2e3b583f778c2a001c75227148c1ce9148eb84779841160ea22f5","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_40_fig_5.jpg","caption":"Fig. 5. Plan-view TEM micrographs of (a) TiN and (b) Ti-Si-N prepared by PEALD at  $350^{\\circ}\\mathrm{C}$","id":"train/atomic-layer-deposition/experimental-usecase/40/fig_5","sample_id":"atomic-layer-deposition/experimental-usecase/40/fig_5","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Plan-view TEM micrograph of a TiN film deposited by PEALD at 350 °C, showing a nanocrystalline microstructure with visible grain contrast at the ~10 nm scale.\"},{\"panel_id\":\"b\",\"text\":\"Plan-view TEM micrograph of a Ti–Si–N film deposited by PEALD at 350 °C, showing a much more homogeneous contrast with no clearly resolved grains at the ~10 nm scale.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Polycrystalline with clear grain contrast and lattice fringes.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Silicon addition (≈11 at.%) completely suppresses crystallization, transforming the film from polycrystalline (a) to fully amorphous (b).\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Fully amorphous – completely featureless, no grains or lattice fringes at 10 nm scale.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":5,"y":5,"width":569,"height":473},{"panel_id":"b","x":5,"y":519,"width":569,"height":462}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/40/images/fig_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/40/images/fig_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/40/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/40/Park et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"40","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":575,"height":983,"image_format":"jpeg","image_sha256":"dec607a840f61348abe5b16d99fc6da76f02a56078ec3b0ca88d97864488c260","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_41_fig_3.jpg","caption":"FIG. 3. High resolution TEM image showing a  $20\\mathrm{nm}$  thick  $\\mathrm{Al}_2\\mathrm{O}_3$  film on  $c$ -Si after a  $30\\mathrm{min}$  annealing at  $425^{\\circ}\\mathrm{C}$  in a  $\\mathrm{N}_2$  environment.","id":"train/atomic-layer-deposition/experimental-usecase/41/fig_3","sample_id":"atomic-layer-deposition/experimental-usecase/41/fig_3","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The image shows a cross-sectional view of a layered material structure, including Al2O3, SiOx, and c-Si from top to bottom.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The colors of the layers indicate the density which say something on the different materials.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The crystallic Si substrate shows ordered dots in a constant array, whereas the alumina layer is unorganized.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The crystaline Si has the highest density, followed by the alumina, with the silicon oxide having the lowest density.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The silicon oxide thickness is around 2 nm.\"}]}]","bbox":[{"panel_id":"a","x":4,"y":1,"width":529,"height":529}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/41/images/fig_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/41/images/fig_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/41/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/41/Hoex et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"41","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":536,"height":531,"image_format":"jpeg","image_sha256":"5d9012c95bf5492c6dc3304929933e6eea400d9afb009088432dc2f226910b79","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_42_figure_4.jpg","caption":"Figure 4. Scanning electron micrographs of deposited copper films. (a) Plan view of copper nanoparticles deposited using a 1 s pulse length. (b) Plan view of crystalline copper deposited using a 4 s pulse length. (c) Plan view of crystalline copper deposited using a 6 s pulse length. (d) A profile of crystalline copper deposited using a 6 s pulse length.","id":"train/atomic-layer-deposition/experimental-usecase/42/figure_4","sample_id":"atomic-layer-deposition/experimental-usecase/42/figure_4","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":0,"y":0,"width":558,"height":412},{"panel_id":"b","x":576,"y":0,"width":554,"height":412},{"panel_id":"c","x":0,"y":427,"width":559,"height":412},{"panel_id":"d","x":578,"y":427,"width":552,"height":412}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/42/images/figure_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/42/images/figure_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/42/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/42/Coyle et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"42","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":1130,"height":839,"image_format":"jpeg","image_sha256":"7856f25246500ea3766152283e6c024866284887da783e38230ca64bb98f27e2","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_44_figure_1.jpg","caption":"Figure 1. FE-SEM images of surface morphology of ALD thin films synthesized at  $170^{\\circ}\\mathrm{C}$  for (a) 700 cycles (sample # PbTe-9), (b) 3000 cycles (sample # PbTe-11), (c) 1000 cycles (sample # PbSe-9), and (d) 4000 cycles (sample # PbSe-11). Sample numbers are indicated by  $*$  in Tables I and II.","id":"train/atomic-layer-deposition/experimental-usecase/44/figure_1","sample_id":"atomic-layer-deposition/experimental-usecase/44/figure_1","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":5,"y":0,"width":500,"height":378},{"panel_id":"b","x":505,"y":0,"width":500,"height":380},{"panel_id":"c","x":3,"y":377,"width":504,"height":368},{"panel_id":"d","x":505,"y":377,"width":496,"height":370}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/images/figure_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/images/figure_1.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/Zhang et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"44","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":1005,"height":753,"image_format":"jpeg","image_sha256":"6724c13befd60da85a625f4ab67a73fa0b3df4eb6f6ac09638ae163c7930417c","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_44_figure_2.jpg","caption":"Figure 2. FE-SEM micrograph showing a cross-sectional view of cleavage side of ALD at  $170^{\\circ}\\mathrm{C}$ ; (a) 3000 cycles (sample # PbTe-11, Table I), and (b) 4000 cycles (sample # PbSe-12, Table II).","id":"train/atomic-layer-deposition/experimental-usecase/44/figure_2","sample_id":"atomic-layer-deposition/experimental-usecase/44/figure_2","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":3,"y":0,"width":504,"height":374},{"panel_id":"b","x":505,"y":3,"width":496,"height":365}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/images/figure_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/images/figure_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/Zhang et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"44","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":1005,"height":375,"image_format":"jpeg","image_sha256":"36984da539b931b555851216eba63f15ff473edab18017c1fe4449e4dd63dce1","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_44_figure_3.jpg","caption":"Figure 3. TEM cross-sectional images of ALD deposited PbTe film (sample # PbTe-9, Table I); (a) low magnification overview of the film, nucleating in a Volmer-Weber type of island growth mode, and (b) high resolution microstructure of individual PbTe crystallite.","id":"train/atomic-layer-deposition/experimental-usecase/44/figure_3","sample_id":"atomic-layer-deposition/experimental-usecase/44/figure_3","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":3,"y":6,"width":497,"height":487},{"panel_id":"b","x":504,"y":2,"width":497,"height":491}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/images/figure_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/images/figure_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/Zhang et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"44","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":1003,"height":500,"image_format":"jpeg","image_sha256":"9beb0e5035c4b621004130d79d54f455195ccb924bb02b231364e82fe3d63e6b","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_44_figure_5.jpg","caption":"Figure 5. HAADF-STEM image and the corresponding EDS elemental maps of Pb, Se and Te from individual films of ALD PbTe and ALD PbSe. The orange window marks the area mapped and the yellow window represents the drift correction are used.","id":"train/atomic-layer-deposition/experimental-usecase/44/figure_5","sample_id":"atomic-layer-deposition/experimental-usecase/44/figure_5","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":0,"y":0,"width":669,"height":370},{"panel_id":"b","x":8,"y":395,"width":660,"height":423}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/images/figure_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/images/figure_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/Zhang et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"44","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":670,"height":817,"image_format":"jpeg","image_sha256":"29c524f864a4c38446cdcd31464990067e30d1d5a65a2320783e427cb140bae8","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_44_figure_6.jpg","caption":"Figure 6. FE-SEM images of basic unit cell for nanolaminate superlattice structures consisting of 5000 ALD cycles (PbTe) plus 4000 cycles (PbSe) on top of it (a) surface morphology (b) cross-sectional view.","id":"train/atomic-layer-deposition/experimental-usecase/44/figure_6","sample_id":"atomic-layer-deposition/experimental-usecase/44/figure_6","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":4,"y":1,"width":585,"height":427},{"panel_id":"b","x":592,"y":1,"width":576,"height":422}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/images/figure_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/Zhang et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"44","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":1172,"height":430,"image_format":"jpeg","image_sha256":"ff391a30f50910c74f074236efc041d5d02d29a4ee5638624624484298d94f08","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_44_figure_7.jpg","caption":"Figure 7. HAADF-STEM image and the corresponding EDS elemental maps of Pb, Se and Te from the PbTe/PbSe clearly delineates two distinct layers of a basic nanolaminate composite with ALD PbSe on top of ALD PbTe. The orange window marks the area mapped and the yellow window represents the drift correction are used.","id":"train/atomic-layer-deposition/experimental-usecase/44/figure_7","sample_id":"atomic-layer-deposition/experimental-usecase/44/figure_7","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":0,"y":1,"width":679,"height":325}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/images/figure_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/images/figure_7.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/44/Zhang et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"44","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":681,"height":328,"image_format":"jpeg","image_sha256":"d0fa443cadd3d73c8ae95e5ce6ebdf5d2c0560e0ff01b8f1cf9d82549fc12cab","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_50_fig_8.jpg","caption":"Fig. 8 Transmission electron microscopy images: (a) bright-field (BF) TEM of two ZnO:B films  $(m = 30$  and  $m = 19)$  and a ZnO:Al film  $(m = 18)$ ; and (b) high-angle annular dark field (HAADF) STEM images of ZnO:B  $(m = 30)$  and ZnO:Al  $(m = 18)$ . The B- and Al-doped layers are indicated by arrows.","id":"train/atomic-layer-deposition/experimental-usecase/50/fig_8","sample_id":"atomic-layer-deposition/experimental-usecase/50/fig_8","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":9,"y":14,"width":870,"height":210},{"panel_id":"b","x":9,"y":256,"width":867,"height":163}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/images/fig_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/images/fig_8.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/Garcia-Alonso et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"50","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":878,"height":428,"image_format":"jpeg","image_sha256":"8932befd3a2b775e90de975c4e260737ed791e7034cd97f56b6ac367fced669f","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_54_figure_9.jpg","caption":"Figure 9. SEM images of the Cu-TiN-Si stack: (a) the smooth as-deposited Cu film and (b) the roughened Cu film after annealing at  $700^{\\circ}\\mathrm{C}$ . (c) The as-deposited closed Cu film has become discontinuous after annealing, exposing areas of the underlying surface.","id":"train/atomic-layer-deposition/experimental-usecase/54/figure_9","sample_id":"atomic-layer-deposition/experimental-usecase/54/figure_9","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":8,"y":15,"width":600,"height":583},{"panel_id":"b","x":8,"y":630,"width":600,"height":574},{"panel_id":"c","x":8,"y":1237,"width":600,"height":583}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/54/images/figure_9.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/54/images/figure_9.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/54/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/54/Knoops et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"54","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":608,"height":1820,"image_format":"jpeg","image_sha256":"6e44fc1ad59251918225a336d7129c084ca0941cdc405e7e29653487bcdfa45e","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_57_figure_2.jpg","caption":"Figure 2. (a) Cross-sectional field-emission scanning electron microscopy (FESEM) images of CsI films deposited on Si with 400 cycles (left) and 3300 cycles (right). (b) Top-down FESEM images of CsI films deposited on Si with 300 cycles at different deposition temperatures  $T_{\\mathrm{Dep}}$ . (c) GIXRD patterns of CsI films on Si deposited at different temperatures. CsI reference pattern ICDD 6-311 is indicated as red bars. (d) Relative atomic force microscopy (AFM) roughness  $R_{\\mathrm{q}}$  calculated as absolute  $R_{\\mathrm{q}}$  divided by the film thickness. (e) AFM images of CsI films deposited at 150 and  $250^{\\circ}\\mathrm{C}$ . Unless otherwise evident, data are from films deposited with 300 cycles at  $150^{\\circ}\\mathrm{C}$ , pulse durations of 1.5 and  $0.5\\mathrm{s}$  for Cs(btsa) and  $\\mathrm{SnI_4}$ , respectively, and  $1.0\\mathrm{s}$  purge durations.","id":"train/atomic-layer-deposition/experimental-usecase/57/figure_2","sample_id":"atomic-layer-deposition/experimental-usecase/57/figure_2","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"stacked spectra chart"},{"panel_id":"d","label":"line chart"},{"panel_id":"e","label":"heatmap"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure displays SEM images of CsI films deposited at various temperatures.\"},{\"panel_id\":\"b\",\"text\":\"The figure shows SEM images of CsI films deposited at different temperatures.\"},{\"panel_id\":\"c\",\"text\":\"The figure presents Grazing Incidence X-ray Diffraction (GIXRD) patterns for Cesium Iodide (CsI) films deposited on Silicon at temperatures ranging from 150°C to 350°C. All deposited films exhibit diffraction peaks that perfectly match the reference pattern for cubic CsI (ICDD 6-311)\"},{\"panel_id\":\"d\",\"text\":\"This figure plots the relative surface roughness (Rq normalized by film thickness) of ALD-grown films against deposition temperature, showing roughness decreases to a minimum around 225-250°C before increasing at higher temperatures.\"},{\"panel_id\":\"e\",\"text\":\"The figure compares AFM images of CsI films deposited at different temperatures and thicknesses.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"},{\"panel_id\":\"c\",\"text\":\"\"},{\"panel_id\":\"d\",\"text\":\"| Temperature (°C) | 2θ (°) | Assigned Plane |\\n|-----------------|-------|----------------|\\n| 150              | 28.5   | (200)          |\\n| 150              | 33.0   | (211)          |\\n| 150              | 36.0   | (220)          |\\n| 150              | 47.5   | (310)          |\\n| 150              | 56.0   | (222)          |\\n| 150              | 69.0   | (321)          |\"},{\"panel_id\":\"e\",\"text\":\"| Deposition temperature (°C) | Relative roughness Rq (%) |\\n|-----------------------------|---------------------------|\\n| 150                         | 5                          |\\n| 170                         | 7                          |\\n| 190                         | 10                         |\\n| 210                         | 15                         |\\n| 230                         | 22                         |\\n| 250                         | 35                         |\\n| 270                         | 55                         |\\n| 290                         | 80                         |\\n| 305                         | 95                         |\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Roughness decreases with temperature up to ~225°C, then increases. The minimum occurs due to enhanced adatom surface diffusion at moderate temperatures, which smoothens growth, while higher temperatures may induce excessive decomposition or 3D islanding, increasing roughness.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Yes, the minimal roughness at 225°C is beneficial for a gate dielectric. A smoother film reduces interfacial scattering and charge trapping, improving channel mobility and threshold voltage stability. However, the electrical quality of the dielectric itself must also be verified at this temperature.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Cesium Iodide (CsI). The peaks align perfectly with the ICDD 6-311 reference bars for CsI.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"To save thermal budget/energy. Since there is no gain in crystal quality at 350°C, the lower temperature is preferred to protect the substrate and reduce costs.\"}]}]","bbox":[{"panel_id":"a","x":34,"y":1,"width":637,"height":198},{"panel_id":"b","x":32,"y":242,"width":642,"height":219},{"panel_id":"c","x":717,"y":1,"width":462,"height":437},{"panel_id":"d","x":698,"y":425,"width":473,"height":292},{"panel_id":"e","x":717,"y":724,"width":456,"height":181}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/57/images/figure_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/57/images/figure_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/57/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/57/Weia et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"57","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":1178,"height":914,"image_format":"jpeg","image_sha256":"23b99e485fb98f6e1465fde5708f13e6dc925739fcd47d0b5aa9f1121d2da1b5","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_57_figure_5.jpg","caption":"Figure 5. Panels (a), (c), (e), (g), and (i) show top-down SEM images of samples where  $\\mathrm{PbI}_2$  was deposited on top of the CsI film. Panels (b), (d), (f), (h), and (j) show cross-sectional SEM images of the same samples. Panel (k) shows a cross-sectional FESEM image of a hole with 10:1 AR in a patterned Si substrate. Similar to the other samples, also on this sample, CsI film was deposited first, followed by  $\\mathrm{PbI}_2$  deposition. CsI films (100 nm) were deposited at  $150^{\\circ}\\mathrm{C}$  with 300 cycles. The pulse durations were  $1.5\\mathrm{s}$  for  $\\mathrm{Cs(btsa)}$  and  $1.0\\mathrm{s}$  for  $\\mathrm{SnI}_4$  and purge durations were  $1.0\\mathrm{s}$ . In the  $\\mathrm{PbI}_2$  deposition, the durations of the precursor pulses were 0.8 and  $2.0\\mathrm{s}$  for  $\\mathrm{Pb(btsa)}_2$  and  $\\mathrm{SnI}_4$ , respectively, whereas purge durations were  $1.0\\mathrm{s}$ .","id":"train/atomic-layer-deposition/experimental-usecase/57/figure_5","sample_id":"atomic-layer-deposition/experimental-usecase/57/figure_5","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"},{"panel_id":"e","label":"image panel"},{"panel_id":"f","label":"image panel"},{"panel_id":"g","label":"image panel"},{"panel_id":"h","label":"image panel"},{"panel_id":"i","label":"image panel"},{"panel_id":"j","label":"image panel"},{"panel_id":"k","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":12,"y":3,"width":270,"height":197},{"panel_id":"b","x":287,"y":0,"width":276,"height":206},{"panel_id":"c","x":4,"y":195,"width":274,"height":201},{"panel_id":"d","x":283,"y":199,"width":280,"height":193},{"panel_id":"e","x":4,"y":395,"width":282,"height":191},{"panel_id":"f","x":285,"y":387,"width":280,"height":205},{"panel_id":"g","x":4,"y":585,"width":276,"height":191},{"panel_id":"h","x":283,"y":585,"width":280,"height":195},{"panel_id":"i","x":6,"y":780,"width":276,"height":180},{"panel_id":"j","x":285,"y":778,"width":278,"height":186},{"panel_id":"k","x":561,"y":1,"width":276,"height":963}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/57/images/figure_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/57/images/figure_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/57/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/57/Weia et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"57","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":842,"height":964,"image_format":"jpeg","image_sha256":"b488eaf5047659f74d1336b23bd3351b8932df72cf6a01b474631bc6070f40f4","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_64_fig_3.jpg","caption":"FIG. 3. AFM images of  $2\\mu \\mathrm{m}\\times 2\\mu \\mathrm{m}$  regions of (a) a  $45\\mathrm{-nm}$  platinum film grown at  $250^{\\circ}\\mathrm{C}$  with a total  $\\mathbb{Z}$  -range of  $6.1~\\mathrm{nm}$  and b) a  $50 - \\mathrm{nm}$  platinum film grown at  $300^{\\circ}\\mathrm{C}$  with a total  $\\mathbb{Z}$  -range of  $9.7~\\mathrm{nm}$  .The  $\\mathbb{Z}$  -scale for both images spans  $10\\mathrm{nm}$","id":"train/atomic-layer-deposition/experimental-usecase/64/fig_3","sample_id":"atomic-layer-deposition/experimental-usecase/64/fig_3","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"AFM image of a 45-nm platinum film grown at 250 °C over a 2 µm × 2 µm region, with a z-range of 6.1 nm and RMS roughness of 0.8 nm. The surface shows relatively small grains.\"},{\"panel_id\":\"b\",\"text\":\"AFM image of a 50-nm platinum film grown at 300 °C over a 2 µm × 2 µm region, with a z-range of 9.7 nm and RMS roughness of 1.2 nm. The surface shows larger grains and increased roughness compared to the 250 °C film.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Parameter | Value |\\n|---|---|\\n| Film thickness | 45 nm |\\n| Deposition temperature | 250 °C |\\n| Scan area | 2 µm × 2 µm |\\n| Z-range | 6.1 nm |\\n| RMS surface roughness | 0.8 nm |\\n| Grain characteristics | Smaller grains compared to 300 °C film |\"},{\"panel_id\":\"b\",\"text\":\"| Parameter | Value |\\n|---|---|\\n| Film thickness | 50 nm |\\n| Deposition temperature | 300 °C |\\n| Scan area | 2 µm × 2 µm |\\n| Z-range | 9.7 nm |\\n| RMS surface roughness | 1.2 nm |\\n| Grain characteristics | Larger grains compared to 250 °C film |\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Increasing the deposition temperature to 300 °C results in larger grain sizes and slightly higher surface roughness compared to films grown at 250 °C. However, the films remain smoother than similar films deposited using air as the oxygen source in previous studies. This improvement is likely due to the higher nucleation density achieved with the higher oxygen partial pressure, which promotes more uniform film growth and reduces roughness.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The RMS surface roughness increases from 0.8 nm at 250 °C to 1.2 nm at 300 °C.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"250 °C (45 nm film): Smaller grains, RMS roughness 0.8 nm, z-range 6.1 nm, 300 °C (50 nm film) Larger grains, RMS roughness 1.2 nm, z-range 9.7 nm\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":603,"height":607},{"panel_id":"b","x":0,"y":659,"width":603,"height":602}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/64/images/fig_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/64/images/fig_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/64/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/64/Aaltonen et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"64","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":603,"height":1294,"image_format":"jpeg","image_sha256":"d2ff9a89360322c3f7ec8aedb2a56fc203b7813bf8a54e5fddcf4e9a045eceb7","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_9_d0c00dec3627c492e7e1f2f448ee4f5d8fa71fbdb9057d123f4c3d482cdd5f4a.jpg","caption":"","id":"train/atomic-layer-deposition/experimental-usecase/9/d0c00dec3627c492e7e1f2f448ee4f5d8fa71fbdb9057d123f4c3d482cdd5f4a","sample_id":"atomic-layer-deposition/experimental-usecase/9/d0c00dec3627c492e7e1f2f448ee4f5d8fa71fbdb9057d123f4c3d482cdd5f4a","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"line chart"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"},{"panel_id":"e","label":"image panel"},{"panel_id":"f","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This panel shows a cross-sectional electron microscopy image of a nanoscale device structure, illustrating excellent thickness control and conformal film coverage on high-aspect-ratio features such as metal and high-κ regions on silicon. It highlights the capability of low-temperature ALD to uniformly coat complex 3D architectures at the ~100 nm scale.\"},{\"panel_id\":\"b\",\"text\":\"This schematic illustrates the discrete, self-limiting nature of atomic layer deposition, where film thickness increases in well-defined nanometer- or sub-nanometer-scale steps with each process cycle. It emphasizes precise thickness control arising from the sequential, surface-controlled reaction mechanism.\"},{\"panel_id\":\"c\",\"text\":\"This panel presents an SEM image of a dense array of high-aspect-ratio nanostructures uniformly coated by ALD. The image demonstrates excellent conformality and uniformity over large areas and complex geometries, which is difficult to achieve with conventional deposition techniques.\"},{\"panel_id\":\"d\",\"text\":\"This panel shows a transmission electron microscopy image of a single, elongated nanostructure uniformly coated along its entire length. The consistent coating thickness confirms the ability of ALD to provide conformal coverage on isolated, curved, and one-dimensional nanoscale objects.\"},{\"panel_id\":\"e\",\"text\":\"This cross-sectional SEM image highlights multilayer film growth on a structured substrate, showing sharp interfaces and uniform layer thickness even within rough or porous regions. The image underscores the advantage of low-temperature ALD for depositing high-quality films without damaging underlying materials.\"},{\"panel_id\":\"f\",\"text\":\"This panel demonstrates a flexible electronic device incorporating ALD-processed layers, bent without mechanical failure. It illustrates how low-temperature, conformal ALD enables integration of functional thin films into flexible and wearable electronic applications.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"| Discrete process step|Nanometer-level thickness (schematic)|\\n|----------------------|--------------------------------------|\\n|0|~0 nm|\\n|1|Increasing|\\n|2|Increasing|\\n|3|Increasing|\\n|4|Increasing|\\n|...|...|\\n|N|~Linear increase|\"},{\"panel_id\":\"c\",\"text\":\"\"},{\"panel_id\":\"d\",\"text\":\"\"},{\"panel_id\":\"e\",\"text\":\"\"},{\"panel_id\":\"f\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"GaP nanowires, CIGS absorber layers, and FinFET semiconductor fins.\"}]},{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The figure illustrates the need for highly conformal, pinhole-free thin films in advanced technologies like FinFET transistors, nanowire devices, CIGS solar cells, and flexible OLEDs.  These structures often involve intricate geometries or temperature-sensitive substrates, making precise, low-temperature deposition critical.  Such regulated coatings directly influence device performance, reliability, and scalability.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"For constant electrical performance, FinFET fins must have homogeneous coverage on three-dimensional geometries, For accurate surface passivation, GaP nanowires require nanoscale ALD layers,  Rough CIGS absorber layer — required pinhole-free buffer layers for optimal solar cell operation, Flexible OLED substrates — depend on low-temperature, mechanically stable films.\"}]}]","bbox":[{"panel_id":"a","x":91,"y":2,"width":232,"height":214},{"panel_id":"b","x":372,"y":2,"width":199,"height":213},{"panel_id":"c","x":94,"y":233,"width":226,"height":195},{"panel_id":"d","x":372,"y":232,"width":198,"height":196},{"panel_id":"e","x":94,"y":442,"width":221,"height":197},{"panel_id":"f","x":372,"y":444,"width":197,"height":194}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/9/images/d0c00dec3627c492e7e1f2f448ee4f5d8fa71fbdb9057d123f4c3d482cdd5f4a.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/9/images/d0c00dec3627c492e7e1f2f448ee4f5d8fa71fbdb9057d123f4c3d482cdd5f4a.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/9/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/9/H.C.M. Knoops et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"9","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"not_found"},"width":570,"height":642,"image_format":"jpeg","image_sha256":"48cec1748865ecfa809582df6e1268e984b131af40b0c6ae2cd6ba4c2d67d8de","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_10_figure_2.jpg","caption":"Figure 2. (a) Photograph of a 4-in. Si(100) wafer covered with  $10\\mathrm{nm}$  of a-Si:H with the letters  $\\mathrm{^{\\circ}TU / e^{\\circ}}$  prepared using the direct-write ALD process of  $\\mathrm{In}_2\\mathrm{O}_3{:}\\mathrm{H}$ . The number of ALD cycles was 400 and the thickness of the  $\\mathrm{In}_2\\mathrm{O}_3{:}\\mathrm{H}$  was  $\\sim 35 \\mathrm{nm}$ . (b) XPS signals for the  $\\mathrm{In}3\\mathrm{d}_{5 / 2}$  and In  $3\\mathrm{d}_{3 / 2}$  binding energy measured for two distinctive points inside (black) and outside (red) the patterned area.","id":"train/atomic-layer-deposition/simulation-usecase/10/figure_2","sample_id":"atomic-layer-deposition/simulation-usecase/10/figure_2","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"multi spectra chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Photograph of a 4-inch Si(100) wafer coated with 10 nm a-Si:H, showing the letters \\\"TU/e\\\" patterned by direct-write ALD of In2O3:H (approximately 35 nm thick after 400 ALD cycles), demonstrating successful area-selective deposition at macroscopic scale.\"},{\"panel_id\":\"b\",\"text\":\"XPS spectra of the In 3d region comparing activated (black) and non-activated (red) areas. The activated area shows strong In 3d5/2 (approximately 444.5 eV) and In 3d3/2 (approximately 452 eV) peaks, while the non-activated area shows negligible indium signal, confirming excellent selectivity.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"| Binding energy (eV) | activated area (10^5 counts/s) | non-activated area (10^5 counts/s) |\\n|---|---|---|\\n| 439 | 0.05 | 0.05 |\\n| 441 | 0.10 | 0.05 |\\n| 442 | 0.15 | 0.05 |\\n| 443 | 0.50 | 0.05 |\\n| 444 | 1.80 | 0.05 |\\n| 445 | 2.90 | 0.05 |\\n| 446 | 1.60 | 0.05 |\\n| 447 | 0.50 | 0.05 |\\n| 448 | 0.20 | 0.05 |\\n| 449 | 0.15 | 0.05 |\\n| 450 | 0.30 | 0.05 |\\n| 451 | 1.00 | 0.05 |\\n| 452 | 1.90 | 0.05 |\\n| 452.5 | 2.10 | 0.05 |\\n| 453 | 1.80 | 0.05 |\\n| 454 | 0.70 | 0.05 |\\n| 455 | 0.25 | 0.05 |\\n| 457 | 0.10 | 0.05 |\\n| 459 | 0.10 | 0.05 |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"It compares XPS spectra from two locations on the same sample.\\n\\n, Black curve: activated area\\n\\n, Red curve: non-activated area\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The activated area shows strong peaks, while the non-activated area stays close to the baseline across the binding-energy window. This indicates that indium-containing material is present (or much thicker) only on the activated region. In other words, the data support strong area selectivity in the deposition process.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"XPS is surface-sensitive; if nucleation had occurred on the non-activated area, even a thin initial film would typically produce detectable indium-related features. The near-baseline response implies the surface remains largely uncovered by the deposited In₂O₃:H (or is below detection / extremely thin). This is consistent with a long nucleation delay on the non-activated, H-terminated Si surface.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The optical photo shows that a macroscopic pattern is produced, but it does not by itself prove chemical selectivity. The XPS comparison verifies that the patterned regions differ strongly in composition (indium signal present vs absent/very low). Together, they provide both a visual and chemical validation of selective growth suitable for direct-write patterning.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":310,"height":232},{"panel_id":"b","x":311,"y":4,"width":356,"height":267}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/10/images/figure_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/10/images/figure_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/10/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/10/Alfredo Mameli et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"10","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":667,"height":278,"image_format":"jpeg","image_sha256":"fce7ca7cce68a711e92532e21f236dc132980edf45f26a990b2760cfe8d55435","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_10_figure_3.jpg","caption":"Figure 3. (a) Photograph of  $\\mathrm{In}_2\\mathrm{O}_3\\cdot \\mathrm{H}$  lines being 3.0 and  $0.8\\mathrm{mm}$  wide as prepared by the direct-write ALD process with 400 cycles. XPS line scans for the patterns depicted in (a) showing the atomic percentages related to (b)  $\\mathrm{In}_2\\mathrm{O}_3$  (In  $3\\mathrm{d}_5$ $\\mathrm{I}_2$  and O 1s) and (c) the Si substrate (Si 2p).","id":"train/atomic-layer-deposition/simulation-usecase/10/figure_3","sample_id":"atomic-layer-deposition/simulation-usecase/10/figure_3","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"multiple scatter plot"},{"panel_id":"c","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Optical photo of a patterned/printed region on a substrate, used to show a visible contrast between deposited and non-deposited areas along a line.\"},{\"panel_id\":\"b\",\"text\":\"EDX line-scan composition across the pattern: In and O (from In₂O₃:H) atomic percentages vs distance (mm). In and O are high on the deposited regions (near the ends of the scan) and drop to ~0 over the non-deposited middle region, indicating strong lateral selectivity.\"},{\"panel_id\":\"c\",\"text\":\"EDX line-scan Si atomic percentage vs distance. Si is low where In/O are high (film-covered regions) and high in the central region where In/O are near zero (exposed substrate), complementing panel (b).\"}]","data_extraction":"[{\"panel_id\":\"b\",\"text\":\"| Distance (mm) | In (Atomic %) | O (In2O3:H) (Atomic %) |\\n|---|---|---|\\n| 0.0 | 8.0 | 12.0 |\\n| 0.8 | 31.0 | 51.5 |\\n| 1.6 | 31.0 | 51.0 |\\n| 2.4 | 31.0 | 55.0 |\\n| 3.2 | 25.0 | 41.0 |\\n| 4.0 | 1.5 | 2.0 |\\n| 4.8 | 0.0 | 0.0 |\\n| 5.6 | 0.0 | 0.0 |\\n| 6.4 | 0.0 | 0.0 |\\n| 7.2 | 0.0 | 0.0 |\\n| 8.0 | 0.0 | 0.0 |\\n| 8.8 | 0.0 | 0.0 |\\n| 9.6 | 0.0 | 0.0 |\\n| 10.4 | 0.0 | 0.0 |\\n| 11.2 | 0.0 | 0.0 |\\n| 12.0 | 0.0 | 0.0 |\\n| 12.8 | 0.0 | 0.0 |\\n| 13.6 | 0.0 | 0.0 |\\n| 14.4 | 0.0 | 0.5 |\\n| 15.2 | 14.0 | 24.0 |\\n| 16.0 | 30.0 | 51.0 |\\n| 16.8 | 12.0 | 20.0 |\\n| 17.6 | 0.5 | 0.5 |\\n| 18.4 | 0.5 | 0.5 |\"},{\"panel_id\":\"c\",\"text\":\"| Distance (mm) | Si (Atomic %) |\\n|---|---|\\n| 0.0 | 47.0 |\\n| 0.8 | 2.0 |\\n| 1.6 | 2.0 |\\n| 2.4 | 2.0 |\\n| 3.2 | 13.0 |\\n| 4.0 | 66.0 |\\n| 4.8 | 72.0 |\\n| 5.6 | 73.0 |\\n| 6.4 | 73.0 |\\n| 7.2 | 73.5 |\\n| 8.0 | 74.0 |\\n| 8.8 | 73.0 |\\n| 9.6 | 75.0 |\\n| 10.4 | 74.0 |\\n| 11.2 | 74.0 |\\n| 12.0 | 74.0 |\\n| 12.8 | 74.5 |\\n| 13.6 | 74.0 |\\n| 14.4 | 73.0 |\\n| 15.2 | 35.0 |\\n| 16.0 | 3.0 |\\n| 16.8 | 37.0 |\\n| 17.6 | 72.0 |\\n| 18.4 | 73.0 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It quantifies the lateral chemical contrast over a macroscopic distance, not just a single point. The steep transitions between In/O-rich and Si-rich regions provide a practical measure of pattern edge sharpness and selectivity. This supports claims that the process can produce patterned films over device-relevant length scales.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"(b) Atomic % of In and O (In₂O₃:H) vs distance\\n\\n, (c) Atomic % of Si vs distance\\n\\n, x-axis (both): distance (mm) along the measured line scan\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It implies that nucleation (and thus measurable film growth) is strongly suppressed on the non-activated surface over that lateral region. If significant nucleation had occurred, In and O would remain detectible across a wider span. The sharp localization of In/O is consistent with a large nucleation delay difference between activated and non-activated areas.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Where In and O are high in (b), Si is low in (c), indicating those regions are covered by the In₂O₃:H film. In the middle section of the scan, In and O drop to ~0 while Si rises to a high level, indicating exposed substrate. This complementary behavior supports strong spatial selectivity of the deposition.\"}]}]","bbox":[{"panel_id":"c","x":3,"y":519,"width":504,"height":363},{"panel_id":"b","x":0,"y":208,"width":507,"height":308},{"panel_id":"a","x":85,"y":-1,"width":425,"height":208}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/10/images/figure_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/10/images/figure_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/10/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/10/Alfredo Mameli et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"10","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":511,"height":881,"image_format":"jpeg","image_sha256":"0600ff8fe2a3f6dd1d547d24d35e9a5794b03425f27c5a3f61332c6a58398ae6","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_10_figure_6.jpg","caption":"Figure 6. Pattern prepared by direct-write ALD of a  $35~\\mathrm{nm}$  thick  $\\mathrm{In}_2\\mathrm{O}_3{:}\\mathrm{H}$  film on a flexible stainless steel foil covered with  $20~\\mathrm{nm}$  of a-Si:H. This serves as a first demonstrator of the capability of the direct-write ALD process for large-area and flexible electronics.","id":"train/atomic-layer-deposition/simulation-usecase/10/figure_6","sample_id":"atomic-layer-deposition/simulation-usecase/10/figure_6","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Photograph of a gloved hand holding a patterned, translucent thin-film sample showing the “TU/e” logo as a visible contrast feature, consistent with selective deposition/patterning of a transparent conductive oxide film.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"A gloved hand holding a thin-film sample\\n\\n, A visible “TU/e” patterned region on the sample\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The visible contrast indicates that the film (or its thickness/composition) differs between the patterned and non-patterned regions. In this paper’s context, that is consistent with selective deposition occurring only on the activated areas. The pattern being visible at macroscopic scale suggests the selectivity is strong enough to be observed without microscopy.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Many oxide TCO films transmit visible light when sufficiently thin and smooth, so the substrate remains visible through them. At the same time, differences in thickness, refractive index, or free-carrier absorption between coated and uncoated regions can create a noticeable color/contrast. That makes patterned regions detectable even when the film is broadly transparent.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It provides an immediate demonstration that the method can produce large-area, user-visible patterns, not just nanoscale selectivity at a single test spot. This supports potential use in patterned TCO features for electronics and optoelectronics.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":2,"width":499,"height":559}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/10/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/10/images/figure_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/10/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/10/Alfredo Mameli et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"10","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":500,"height":564,"image_format":"jpeg","image_sha256":"4b20731074011a398d81c81ee9a2f99a9004cf086922508ad157a68e195141a0","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_17_figure_2.jpg","caption":"Figure 2: Transmission electron microscopic image of  $\\mathrm{Al}_2\\mathrm{O}_3$ -coated MWCNTs. The insert shows a cylindrical  $\\mathrm{Al}_2\\mathrm{O}_3$  layer (A) coating a particular multi-walled CNTs and the multi walls (B) of this multi-walled nanotube.","id":"train/atomic-layer-deposition/simulation-usecase/17/figure_2","sample_id":"atomic-layer-deposition/simulation-usecase/17/figure_2","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":4,"y":3,"width":995,"height":705}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/17/images/figure_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/17/images/figure_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/17/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/17/David Sibanda et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"17","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":1003,"height":714,"image_format":"jpeg","image_sha256":"0d6438c34fb058f6b85450a4eaf82a6a7d6e396b5a5b1ce920859b7d987ca423","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_32_fig_10.jpg","caption":"Fig. 10 - Snapshots of the flow field developed during the TMA exposure, inside the reactor chamber: a) 10 ms, b) 20 ms, c) 30 ms, d) 60 ms, after the start of the TMA exposure.","id":"train/atomic-layer-deposition/simulation-usecase/32/fig_10","sample_id":"atomic-layer-deposition/simulation-usecase/32/fig_10","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":8,"y":14,"width":446,"height":610},{"panel_id":"b","x":722,"y":14,"width":456,"height":613},{"panel_id":"c","x":9,"y":697,"width":450,"height":606},{"panel_id":"d","x":727,"y":697,"width":439,"height":615}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/images/fig_10.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/images/fig_10.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/reactorGP Gakis et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"32","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":1178,"height":1358,"image_format":"jpeg","image_sha256":"6c39412697f22b0f1062c7f03197ee0f137631f0d957e5fa4bae43721c02574f","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_32_fig_13.jpg","caption":"Fig. 13 - Snapshots of the flow field inside the reactor chamber: a)  $10\\mathrm{ms}$ , b)  $20\\mathrm{ms}$ , c)  $60\\mathrm{ms}$ , d)  $200\\mathrm{ms}$  after the start of the  $\\mathrm{H}_2\\mathrm{O}$  pulse.","id":"train/atomic-layer-deposition/simulation-usecase/32/fig_13","sample_id":"atomic-layer-deposition/simulation-usecase/32/fig_13","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":9,"y":10,"width":441,"height":603},{"panel_id":"b","x":727,"y":12,"width":445,"height":611},{"panel_id":"c","x":8,"y":670,"width":441,"height":608},{"panel_id":"d","x":704,"y":670,"width":450,"height":604}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/images/fig_13.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/images/fig_13.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/reactorGP Gakis et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"32","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":1172,"height":1339,"image_format":"jpeg","image_sha256":"98154f0bed104f15681226dc6f6f91fc537d910ee8e0427bfe4ec0c240dad330","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_32_fig_5.jpg","caption":"Fig. 5 - a) Flow field predictions for the process parameters of Table 1, b) flow field predictions at 30 sccm top inlet flow.","id":"train/atomic-layer-deposition/simulation-usecase/32/fig_5","sample_id":"atomic-layer-deposition/simulation-usecase/32/fig_5","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":8,"y":8,"width":392,"height":543},{"panel_id":"b","x":602,"y":8,"width":398,"height":544}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/images/fig_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/images/fig_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/reactorGP Gakis et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"32","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":1000,"height":622,"image_format":"jpeg","image_sha256":"bea7e2eaef9d2bb1301f1ace9d021449c996723da21cc894e40aebd9b20ed6bf","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_36_figure_1.jpg","caption":"Figure 1. Surface morphologies observed using SEM and AFM: (a, e) bare Ti substrate, (b, f) ODTS-coated Ti, (c, g) ODPA-coated Ti, and (d, h) DDPA-coated Ti. Inset images show the WCA of the samples.","id":"train/atomic-layer-deposition/simulation-usecase/36/figure_1","sample_id":"atomic-layer-deposition/simulation-usecase/36/figure_1","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"},{"panel_id":"e","label":"image panel"},{"panel_id":"f","label":"image panel"},{"panel_id":"g","label":"image panel"},{"panel_id":"h","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The SEM image shows a smooth bare Ti surface with minimal visible texture; a water-contact-angle inset is included.\"},{\"panel_id\":\"b\",\"text\":\"The SEM image of ODTS-coated Ti shows a granular surface texture, with a corresponding WCA inset.\"},{\"panel_id\":\"c\",\"text\":\"The SEM image of ODPA-coated Ti exhibits a relatively uniform morphology compared with (b).\"},{\"panel_id\":\"d\",\"text\":\"The SEM image of DDPA-coated Ti shows a fine, homogeneous surface texture; a WCA inset is included.\"},{\"panel_id\":\"e\",\"text\":\"The AFM image of bare Ti shows limited height contrast and sparse surface features.\"},{\"panel_id\":\"f\",\"text\":\"The AFM image of ODTS-coated Ti shows a higher density of bright surface protrusions.\"},{\"panel_id\":\"g\",\"text\":\"The AFM image of ODPA-coated Ti shows many small, closely spaced nanoscale features.\"},{\"panel_id\":\"h\",\"text\":\"The AFM image of DDPA-coated Ti shows a uniform distribution of nanoscale features, with a displayed height scale up to ~20 nm.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"},{\"panel_id\":\"c\",\"text\":\"\"},{\"panel_id\":\"d\",\"text\":\"\"},{\"panel_id\":\"e\",\"text\":\"\"},{\"panel_id\":\"f\",\"text\":\"\"},{\"panel_id\":\"g\",\"text\":\"\"},{\"panel_id\":\"h\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"DDPA seems to form a more homogeneous layer compared with the other coatings, resulting in a fine and consistent texture. This uniformity probably helps achieve more predictable surface chemistry and potentially better coverage. It's likely why the WCA inset might differ from the granular ODTS one — smoother, more even coatings often give different wetting behavior.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The bare Ti substrate appears quite smooth in the SEM, with minimal surface texture or defects visible at the 1 μm scale. This clean, flat morphology is what you'd expect from a polished or etched metal prior to functionalization. It's a good baseline for comparing how the coatings alter the surface.\"}]},{\"panel_id\":\"f\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The AFM highlights a higher density of bright protrusions that correspond to elevated nanoscale structures. These likely stem from the way ODTS molecules assemble or aggregate during coating. It's noticeably rougher than the bare substrate, which suggests the coating introduces significant topography.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Much rougher — shows clear granular clusters instead of the smooth appearance of bare Ti.\"}]}]","bbox":[{"panel_id":"a","x":1,"y":5,"width":265,"height":278},{"panel_id":"b","x":267,"y":3,"width":268,"height":280},{"panel_id":"c","x":532,"y":6,"width":268,"height":277},{"panel_id":"d","x":800,"y":0,"width":266,"height":288},{"panel_id":"e","x":6,"y":317,"width":260,"height":265},{"panel_id":"f","x":271,"y":311,"width":260,"height":269},{"panel_id":"g","x":533,"y":311,"width":262,"height":269},{"panel_id":"h","x":795,"y":313,"width":269,"height":269}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/36/images/figure_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/36/images/figure_1.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/36/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/36/Seunggi Seo et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"36","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":1067,"height":584,"image_format":"jpeg","image_sha256":"5d800116015d9e34fb62598191b348d1e843b73c76818e6a046e3dcee2aadb4d","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_5_figure_4.jpg","caption":"Figure 4. (a,b) Top-view SEM image of 35 EBID  $\\mathrm{SiO_2}$  seed layers with an area of  $500 \\times 500 \\mathrm{nm}^2$  and a thickness of  $\\sim 1 \\mathrm{nm}$ . The high-magnification image in (b) shows a single pattern. (c,d) Top-view SEM image of a similar sample after 80 ALD cycles of  $\\mathrm{ZnO}$ , deposited at 100  $^\\circ \\mathrm{C}$ . The high-magnification image in (d) shows a single pattern on which the polycrystalline  $\\mathrm{ZnO}$  is clearly visible.","id":"train/atomic-layer-deposition/simulation-usecase/5/figure_4","sample_id":"atomic-layer-deposition/simulation-usecase/5/figure_4","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Top-view SEM image of 35 EBID SiO2 seed layers with  an area of 500 × 500 nm2 and a thickness of ∼1 nm before ZnO ALD deposition\"},{\"panel_id\":\"b\",\"text\":\"Top-view SEM image of 35 EBID SiO2 seed layers with  an area of 500 × 500 nm2 and a thickness of ∼1 nm  after ZnO ALD deposition\"},{\"panel_id\":\"c\",\"text\":\"High magnification of image a - before ALD ZnO\"},{\"panel_id\":\"d\",\"text\":\"High magnification of image b  - after ALD ZnO\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"},{\"panel_id\":\"c\",\"text\":\"\"},{\"panel_id\":\"d\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"35 (Can be calculated from counting the patterns in images a and c)\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"EBID-grown SiO2 and the c-Si native oxide  behave differently because of a high concentration of carbon impurities (∼26 at. %) that was measured for the SiO2 deposited by EBID. Besides OH groups that will obviously be present on a SiO2 surface, it has been suggested that electron irradiation of the TEOS ligands can lead to the formation of Si−C and Si−CH3 and Si−H bonds. The  presence of these surface groups can inhibit DEZ adsorption, which can explain the slight difference in thickness and\\nmorphology that were observed between the ZnO deposited on EBID SiO2 and native SiO2. In addition, the thinner SiO2 EBID patterns may also not form a completely closed film.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The selectivity of the an ALD process can be calculated using the thickness as measured by the SE and the coverage as imaged by the SEM in the figure shown. After 80 ALD cycles of ZnO at a deposition temperature of 100 °C, the selectivity was measured to be 0.96 by SE and 0.83 by SEM.\"}]}]","bbox":[{"panel_id":"c","x":288,"y":9,"width":271,"height":221},{"panel_id":"b","x":7,"y":240,"width":268,"height":254},{"panel_id":"a","x":5,"y":6,"width":275,"height":226},{"panel_id":"d","x":289,"y":240,"width":270,"height":261}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/5/images/figure_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/5/images/figure_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/5/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/5/Alfredo Mameli et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"5","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":561,"height":506,"image_format":"jpeg","image_sha256":"66ba9f113c5cbedc8874db7242a61c53aaa0e95aed54ff1bc987e34954506890","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_5_figure_5.jpg","caption":"Figure 5. (a)  $\\mathrm{ZnO}$  nucleation curves on  $\\mathrm{SiO}_2$  (closed symbols) and a-  $\\mathrm{Si / H}$  (open symbols) as measured by in situ SE for two different temperatures:  $100^{\\circ}\\mathrm{C}$  (square) and  $250^{\\circ}\\mathrm{C}$  (circle). (b) Selectivity as a function of the number of ALD cycles, for deposition temperatures of  $100^{\\circ}\\mathrm{C}$  (square) and  $250^{\\circ}\\mathrm{C}$  (circles). The horizontal dotted line at  $S = 1$  indicates perfect selectivity. Figure 7. Statistical density of  $\\mathrm{ZnO}$  islands on the non-GA, after 80  $\\mathrm{ZnO}$  ALD cycles for different deposition temperatures. The insets show color-coded top-view SEM images of the NGA ( $500 \\times 500 \\mathrm{nm}^2$ ).","id":"train/atomic-layer-deposition/simulation-usecase/5/figure_5","sample_id":"atomic-layer-deposition/simulation-usecase/5/figure_5","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The image panel shows four SEM images of ZnO islands at different deposition temperatures\"},{\"panel_id\":\"b\",\"text\":\"A line chart displaying the relationship between deposition temperature and the number of ZnO islands formed corresponding to the SEM images.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of ZnO islands |  |\\n|---------------------|---|\\n| 150                 |  |\\n| 50                  |  |\\n| 20                  |  |\\n| 0                   |  |\"},{\"panel_id\":\"b\",\"text\":\"| Deposition temperature (°C) | Number of ZnO islands |\\n|-----------------------------|------------------------|\\n| 100                         | 150                    |\\n| 150                         | 50                     |\\n| 200                         | 20                     |\\n| 250                         | 0                      |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"For the same number of cycles, the density of ZnO islands is one order of magnitude larger at 100 °C, compared to the one at 250 °C. The higher density of nucleation at 100 °C seems consistent with undesired reactions with physisorbed precursor or coreactant molecules on the NGA. Because the rate of desorption of  physisorbed species decreases with decreasing the deposition temperature, such physisorbed species may act as spurious nucleation sites and effectively decrease the selectivity of the process at low\\ndeposition temperature.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"1. ALD of Al2O3 on graphene: lower deposition temperatures resulted in\\nhigher Al2O3 coverage on a chemically inert graphene substrate\\n2. Using self-assembled monolayers (SAM) as NGA and titanium as the GA,  the selectivity for Al2O3 ALD is hampered by the physisorption of TMA on the SAM.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"A statistical analysis of the density of ZnO islands was conducted on several 500 × 500 nm2 areas of the a-Si/H substrate in order to explain the decrease in selectivity for low deposition temperature\"}]}]","bbox":[{"panel_id":"b","x":69,"y":3,"width":599,"height":148},{"panel_id":"a","x":5,"y":148,"width":662,"height":309}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/5/images/figure_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/5/images/figure_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/5/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/5/Alfredo Mameli et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"5","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":669,"height":461,"image_format":"jpeg","image_sha256":"d32e18ca37d1218cbf64dfa5ad9bb51b3f2da4390a2d3e5487a058e6a84b665e","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_5_figure_8.jpg","caption":"Figure 8. (a,b) Cross-sectional TEM images of a  $\\mathrm{ZnO}$  thin film that was selectively deposited (using 80 ALD cycles at  $250^{\\circ}\\mathrm{C}$ ) on a  $\\sim 1 \\mathrm{nm}$  thick  $\\mathrm{SiO}_2$  EBID pattern with a width of  $\\sim 500 \\mathrm{nm}$ . Low-magnification images show a pristine NGA outside the patterned GA, indicating that no  $\\mathrm{ZnO}$  deposition occurred on these surfaces. (c) High-magnification images showing the patterned GA and (d) region  $250 \\mathrm{nm}$  away from the pattern, confirming the selectivity of the process. (e,f) EDX elemental mappings of the same pattern at two different magnifications also show the presence of  $\\mathrm{ZnO}$  on the  $\\mathrm{SiO}_2$  GA and the absence of  $\\mathrm{ZnO}$  on the a $\\mathrm{Si / H}$  NGA.","id":"train/atomic-layer-deposition/simulation-usecase/5/figure_8","sample_id":"atomic-layer-deposition/simulation-usecase/5/figure_8","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"},{"panel_id":"e","label":"image panel"},{"panel_id":"f","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"d","x":175,"y":249,"width":164,"height":111},{"panel_id":"b","x":2,"y":156,"width":338,"height":89},{"panel_id":"c","x":3,"y":249,"width":170,"height":112},{"panel_id":"e","x":346,"y":150,"width":208,"height":93},{"panel_id":"f","x":345,"y":251,"width":210,"height":108},{"panel_id":"a","x":3,"y":5,"width":554,"height":145}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/5/images/figure_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/5/images/figure_8.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/5/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/5/Alfredo Mameli et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"5","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":558,"height":361,"image_format":"jpeg","image_sha256":"029880dd2b48c4a32992fef26feabddfee3225896469db869f0b0b785dcf0c57","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_53_figure_1.jpg","caption":"Figure 1. TEM images of  $\\mathrm{SiO}_2$  and  $\\mathrm{Si}_3\\mathrm{N}_4$  ALD thin films.","id":"train/atomic-layer-deposition/simulation-usecase/53/figure_1","sample_id":"atomic-layer-deposition/simulation-usecase/53/figure_1","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":4,"y":3,"width":518,"height":408}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/53/images/figure_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/53/images/figure_1.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/53/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/53/Ciaran A. Murray et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"53","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":522,"height":414,"image_format":"jpeg","image_sha256":"34c216f5300c05dfb823d1367eb4e29b4a6d9410766f57dce4c959787f4fb2ed","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_11_figure_5.jpg","caption":"Figure 5. Low-magnification HAADF-STEM image of a  $\\mathrm{ZnO}$  covered NW together with high-magnification images of the top, center, and bottom regions (a) before and (b) after ALE. (c) Averaged  $\\mathrm{ZnO}$  thicknesses as measured every micrometer along the NWs before and after ALE. The standard deviation of the measurements is taken as the error. (d) High-magnification TEM image of the  $\\mathrm{ZnO}$  layer after the ALE process. Lattice fringes are observable up to the top surface, indicating that no surface amorphization occurs during the ALE process.","id":"train/atomic-layer-etching/experimental-usecase/11/figure_5","sample_id":"atomic-layer-etching/experimental-usecase/11/figure_5","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"multiple scatter plot"},{"panel_id":"d","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows a comparison of a sample before ALE (Atomic Layer Etching) process using scanning electron microscopy (SEM) images at different magnifications.\"},{\"panel_id\":\"b\",\"text\":\"The figure shows a comparison of a sample  after ALE (Atomic Layer Etching) process using scanning electron microscopy (SEM) images at different magnifications.\"},{\"panel_id\":\"c\",\"text\":\"The line chart compares the averaged thickness of the sample at different distances from the top before and after ALE treatment.\"},{\"panel_id\":\"d\",\"text\":\"The figure shows a high-resolution transmission electron microscopy (TEM) image of the sample's surface after ALE treatment.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"},{\"panel_id\":\"c\",\"text\":\"| Distance from Top (µm) | Thickness Before ALE (nm) | Thickness After ALE (nm) |\\n|-------------------------|---------------------------|---------------------------|\\n| 1                      | 60                        | 45                        |\\n| 2                      | 60                        | 45                        |\\n| 3                      | 60                        | 45                        |\\n| 4                      | 60                        | 45                        |\\n| 5                      | 60                        | 45                        |\\n| 6                     | 60                         | 45                        |\"},{\"panel_id\":\"d\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"In the high-resolution TEM image of the top ZnO layers after the ALE process,  lattice fringes are visible up to the top surface layer, demonstrating that the ZnO surface retains its crystallinity upon ALE. This result indicates that no significant damage or amorphization occurs during the ALE process. The observed isotropic etch profile and the fact that the ZnO surface is not damaged or amorphized are in line with a negligible role of ions in the ALE mechanism, illustrating that the O2 plasma ALE half-reaction is predominantly driven by radicals\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The film thickness reduction (15 ± 2 nm) is comparable at all regions of the NW from the STEM-measured thickness differences (before and after ALE).  An averaged EPC of 1.3 ± 0.2 Å is obtained\\nacross the full length of the NW. This EPC value is in excellent agreement with the EPC measured by in situ SE (i.e., 1.31 Å/cycle). The result clearly demonstrates the accurate etch control and the isotropic nature of the ALE process.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"No, TEM images.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"1. 3D substrate consisting of a regular array of vertical NWs\\n2. NWs are conformally covered with a 60 ± 2 nm thick polycrystalline ZnO layer\\n3. NWs are then subjected to 120 cycles of the ALE process at 250 °C\\n4. NWs are 7 μm long\"}]}]","bbox":[{"panel_id":"c","x":23,"y":519,"width":274,"height":229},{"panel_id":"b","x":292,"y":47,"width":260,"height":447},{"panel_id":"a","x":0,"y":59,"width":282,"height":461},{"panel_id":"d","x":280,"y":511,"width":287,"height":254}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/11/images/figure_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/11/images/figure_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/11/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/11/Isotropic Atomic Layer Etching of ZnO Using Acetylacetone and O2 Plasma.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"11","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":564,"height":770,"image_format":"jpeg","image_sha256":"71d2b731daadebbf9536fcaa2ba9d63e4a1a76a3a092a3d0313dd2ed0bf8c8e4","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_22_figure_10.jpg","caption":"Figure 10. AFM height images of an exfoliated  $\\mathrm{MoS}_2$  flake on a thermal Si oxide substrate. (a) Exfoliated flake prior to etching. (b) Exfoliated flake following 90 cycles of  $\\mathrm{MoS}_2$  ALE at  $200^{\\circ}\\mathrm{C}$ . (c) Exfoliated flake after an additional 90 cycles of  $\\mathrm{MoS}_2$  ALE at  $250^{\\circ}\\mathrm{C}$ . Initial etching at  $200^{\\circ}\\mathrm{C}$  shows roughening of edge sites and removal of small regions across the surface. At  $250^{\\circ}\\mathrm{C}$ , roughened edges are removed.","id":"train/atomic-layer-etching/experimental-usecase/22/figure_10","sample_id":"atomic-layer-etching/experimental-usecase/22/figure_10","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":11,"y":8,"width":327,"height":339},{"panel_id":"b","x":338,"y":6,"width":328,"height":340},{"panel_id":"c","x":663,"y":4,"width":410,"height":344}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/images/figure_10.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/images/figure_10.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/Thermal Atomic Layer Etching of MoS2 Using MoF6 and H2O.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"22","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":1073,"height":347,"image_format":"jpeg","image_sha256":"66406ba56b0d76d1d61c989a02b5343093cefd45a8247746d1dcf51b60f32301","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_22_figure_8.jpg","caption":"Figure 8. TEM images of  $\\mathrm{MoS}_2$  -coated CNT. (a) Prior to etching, the CNT displayed  $\\sim 6 - 10$  layers of  $\\mathrm{MoS}_2$  as depicted by the arrows. (b) After the first 60 cycles of etching, roughly  $2\\mathrm{MoS}_2$  layers were removed. (c) Additional 60 etching cycles showed the removal of another 2 layers, leaving only 2-6 total  $\\mathrm{MoS}_2$  layers on the CNT.","id":"train/atomic-layer-etching/experimental-usecase/22/figure_8","sample_id":"atomic-layer-etching/experimental-usecase/22/figure_8","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":2,"y":0,"width":470,"height":316},{"panel_id":"b","x":2,"y":323,"width":469,"height":311},{"panel_id":"c","x":5,"y":641,"width":467,"height":306}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/images/figure_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/images/figure_8.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/Thermal Atomic Layer Etching of MoS2 Using MoF6 and H2O.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"22","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":475,"height":947,"image_format":"jpeg","image_sha256":"33c53a9e234057d38066f44bbfab7700e3b774b912837d5e279621cb96325072","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_30_fig_3.jpg","caption":"Fig. 3. Cross-sectional TEM images of GaN processed with (a) ALE and (b) continuous plasma process.","id":"train/atomic-layer-etching/experimental-usecase/30/fig_3","sample_id":"atomic-layer-etching/experimental-usecase/30/fig_3","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The TEM cross section shows the GaN surface after ALE, with the GaN layer above the AlGaN layer. The etched front appears relatively smooth, with small height variations marked near about 1.3 to 1.4 nm.\"},{\"panel_id\":\"b\",\"text\":\"The TEM cross section shows the GaN surface after a continuous plasma process, again above AlGaN. The etched front appears much rougher and more nonuniform, with larger height variations marked near about 6.8 to 18 nm.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"In (a) the etched front is relatively smooth and the height variation is small. In (b) the etched front is much more jagged and the height variation is larger. This supports the paper’s point that ALE can reduce roughening compared with a continuous plasma process by separating surface modification and removal steps.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"They indicate the material layers in the cross section, with GaN above AlGaN in both subfigures.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The ALE process in subfigure (a).\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It can reduce plasma induced damage related to rough topography.\\n\\nIt improves interface and surface quality for subsequent device layers.\\n\\nIt helps maintain dimensional control in features such as recessed gates.\"}]}]","bbox":[{"panel_id":"b","x":2,"y":223,"width":663,"height":179},{"panel_id":"a","x":2,"y":2,"width":665,"height":179}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/images/fig_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/images/fig_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/Atomic layer etching of GaN and AlGaN.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"30","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":669,"height":408,"image_format":"jpeg","image_sha256":"a90c864c680d7de2a9c546c3e50b2dd5aedd495c042a537ffb911c04addb5009","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_30_fig_4.jpg","caption":"Fig. 4. (Color online) AFM images of AlGaN surface. (a) Initial, (b)  $\\mathrm{Cl}_2$  ALE, and (c)  $\\mathrm{BCl}_3 / \\mathrm{Cl}_2$  ALE.","id":"train/atomic-layer-etching/experimental-usecase/30/fig_4","sample_id":"atomic-layer-etching/experimental-usecase/30/fig_4","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"AFM image of the initial AlGaN surface, with RMS roughness 0.4 nm over a 5 x 5 μm area.\"},{\"panel_id\":\"b\",\"text\":\"AFM image after Cl2 ALE, showing a rougher surface with RMS roughness 1.8 nm.\"},{\"panel_id\":\"c\",\"text\":\"AFM image after BCl3 and Cl2 ALE, showing a smoother surface with RMS roughness 0.3 nm.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Figure | Description |\\n| --- | --- |\\n| (a) | 5 x 5 µm image with RMS = 0.4 nm |\\n| (b) | Image with RMS = 1.8 nm |\\n| (c) | Image with RMS = 0.3 nm |\"}]","vqa":"[{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The initial surface in (a) has moderate roughness at 0.4 nm. After Cl2 ALE in (b) the roughness increases to 1.8 nm, indicating strong roughening. After BCl3 and Cl2 ALE in (c) the roughness decreases to 0.3 nm, indicating a smoother surface than the initial case.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"BCl3 and Cl2 ALE, shown in panel (c) with RMS roughness 0.3 nm.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"It is the RMS surface roughness of the AlGaN surface in nm for that scan area.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It can help maintain uniform electric fields in near surface device regions.\\n\\nIt can improve interface quality for later dielectric or metal deposition.\\n\\nIt can reduce variability in critical dimensions for patterned features.\"}]}]","bbox":[{"panel_id":"c","x":452,"y":6,"width":216,"height":212},{"panel_id":"b","x":229,"y":8,"width":214,"height":211},{"panel_id":"a","x":4,"y":3,"width":215,"height":212}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/images/fig_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/Atomic layer etching of GaN and AlGaN.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"30","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":669,"height":220,"image_format":"jpeg","image_sha256":"272336ec2da19f90ade2bb3995a7948554750aaae7e7f2d93533ee3d80bbc828","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_34_fig10.jpg","caption":"FIG.10. Pore area analysis of the AAO substrate (a) intrinsic, (b) 100 cycles pALE, (c) 300 cycles pALE by ImageJ, and (d) comparison of pore diameter after pALE.","id":"train/atomic-layer-etching/experimental-usecase/34/fig10","sample_id":"atomic-layer-etching/experimental-usecase/34/fig10","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"bar chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"show SEM micrographs of the intrinsic AAO, AAO after 100 pALE cycles, and AAO after 300 pALE cycles, respectively, with red overlays indicating the pore areas extracted using ImageJ.\"},{\"panel_id\":\"b\",\"text\":\"show SEM micrographs of the intrinsic AAO, AAO after 100 pALE cycles, and AAO after 300 pALE cycles, respectively, with red overlays indicating the pore areas extracted using ImageJ.\"},{\"panel_id\":\"c\",\"text\":\"show SEM micrographs of the intrinsic AAO, AAO after 100 pALE cycles, and AAO after 300 pALE cycles, respectively, with red overlays indicating the pore areas extracted using ImageJ.\"},{\"panel_id\":\"d\",\"text\":\"provides a bar chart comparison of the average pore diameters for the three conditions, showing relatively consistent pore sizes with slight variations after ALE cycles.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"},{\"panel_id\":\"c\",\"text\":\"\"},{\"panel_id\":\"d\",\"text\":\"| Condition|Pore Diameter (nm)|Error Bar (approx.)|\\n|----------|------------------|-------------------|\\n|AAO|~200|±100|\\n|100 cycles|~220|±120|\\n|300 cycles|~270|±130|\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Monitoring pore diameter evolution provides insight into etching uniformity, selectivity, and process control. By comparing intrinsic AAO with samples subjected to different pALE cycles, researchers can determine whether the etch rate is consistent and whether the process maintains pore integrity. The increase in pore diameter without significant distortion indicates a controlled etching environment, which is important for applications requiring precise nanostructure tuning, such as filtration membranes, sensing substrates, or templated nanofabrication.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The pore diameter increases from approximately 200 nm for intrinsic AAO to around 270 nm after 300 pALE cycles.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Slight widening of pore openings, Smoother pore-wall edges, More uniform pore boundaries, Increased average pore diameter with higher cycle numbers\"}]}]","bbox":[{"panel_id":"a","x":6,"y":19,"width":653,"height":240},{"panel_id":"b","x":8,"y":268,"width":650,"height":243},{"panel_id":"c","x":5,"y":521,"width":655,"height":240},{"panel_id":"d","x":5,"y":769,"width":655,"height":492}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/images/fig10.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/images/fig10.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/CF4 plasma-based atomic layer etching of Al2O3.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"34","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":666,"height":1267,"image_format":"jpeg","image_sha256":"d5ad206834ae04dbe7faae172898b423809068d2c822c9bb840ab42a86f07039","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_34_fig_2.jpg","caption":"FIG. 2. TEM-EDS element mapping of  $\\mathrm{CF_4}$  plasma-treated  $\\mathrm{Al}_2\\mathrm{O}_3$  film at (a) 0.07 Torr-10 min and at (b) 2 Torr-5 min. Dotted line represents the original  $\\mathrm{Al}_2\\mathrm{O}_3$  film surface.","id":"train/atomic-layer-etching/experimental-usecase/34/fig_2","sample_id":"atomic-layer-etching/experimental-usecase/34/fig_2","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This panel presents a cross-sectional TEM image of the CF₄ plasma-treated Al₂O₃ film at 0.07 Torr for 10 minutes, accompanied by EDS elemental maps (Al, O, F, Si). The maps reveal partial etching and fluorine incorporation near the surface, while the dotted line marks the original Al₂O₃ thickness.\"},{\"panel_id\":\"b\",\"text\":\"This panel shows the same set of analyses for a higher-pressure CF₄ treatment (2 Torr for 5 minutes). The TEM image displays a significantly reduced etching depth, and the EDS maps indicate a more uniform elemental distribution with less fluorine penetration compared to the low-pressure condition.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Low-pressure CF₄ plasma produces deeper etching and stronger chemical modification of Al₂O₃, making it suitable for applications requiring high reactivity and material removal. In contrast, high-pressure plasma results in limited fluorine penetration and minimal etching, which is beneficial for gentle surface treatment. Thus, pressure acts as a key tuning parameter for precise etch-depth control\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.07 Torr for 10 minutes shows significantly deeper fluorine penetration.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Increased F incorporation at low pressure, Greater etching depth at low pressure, More uniform Al/O signal at high pressure, Clear surface modification identified by TEM contrast\"}]}]","bbox":[{"panel_id":"a","x":54,"y":1,"width":1341,"height":321},{"panel_id":"b","x":63,"y":324,"width":1333,"height":295}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/images/fig_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/images/fig_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/CF4 plasma-based atomic layer etching of Al2O3.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"34","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":1403,"height":622,"image_format":"jpeg","image_sha256":"20b0a394b4b5ec0138683178f2a51e299ea275d4e199fa640a821cdbe72969d2","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_34_fig_9.jpg","caption":"FIG. 9. Surface images of the AAO substrate (a)–(c) intrinsic, (d)–(f) after 100 cycles pALE, and (g)–(i) after 300 cycles pALE.","id":"train/atomic-layer-etching/experimental-usecase/34/fig_9","sample_id":"atomic-layer-etching/experimental-usecase/34/fig_9","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"},{"panel_id":"e","label":"image panel"},{"panel_id":"f","label":"image panel"},{"panel_id":"g","label":"image panel"},{"panel_id":"h","label":"image panel"},{"panel_id":"i","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This panel shows a tilted SEM image of the pristine porous template at 0 ALD cycles, revealing a highly ordered, interconnected pore network with sharp edges and uniform feature size, serving as the reference morphology before film deposition.\"},{\"panel_id\":\"b\",\"text\":\"This top-view SEM image at higher magnification illustrates the open pore structure of the uncoated template, with well-defined pore openings and thin pore walls, indicating minimal surface roughness prior to ALD processing.\"},{\"panel_id\":\"c\",\"text\":\"This high-magnification SEM image highlights the local pore geometry of the uncoated structure, showing irregular pore interiors and sharp junctions that emphasize the complexity of the high-aspect-ratio surface to be coated.\"},{\"panel_id\":\"d\",\"text\":\"This tilted SEM image after 100 ALD cycles shows that the overall porous architecture is preserved while pore walls become visibly thicker, demonstrating conformal film growth without structural collapse.\"},{\"panel_id\":\"e\",\"text\":\"This top-view SEM image after 100 ALD cycles reveals a reduction in pore diameter and smoother pore edges, indicating uniform coating of the internal pore surfaces by the ALD film.\"},{\"panel_id\":\"f\",\"text\":\"This high-magnification image further confirms conformal deposition after 100 cycles, with increased wall thickness and rounded pore contours resulting from continuous ALD film growth.\"},{\"panel_id\":\"g\",\"text\":\"This tilted SEM image after 300 ALD cycles shows substantial thickening of the pore walls while maintaining the long-range order of the porous structure, highlighting the scalability of conformal growth with increasing cycle number.\"},{\"panel_id\":\"h\",\"text\":\"This top-view SEM image after 300 cycles demonstrates significant pore narrowing and enhanced uniformity, reflecting cumulative ALD film deposition on all exposed surfaces.\"},{\"panel_id\":\"i\",\"text\":\"This high-magnification SEM image illustrates near-cylindrical pores with smooth, thick walls after 300 cycles, confirming excellent thickness control and conformality even after extensive ALD processing.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Description|\\n|------------|\\n|SEM image at 20k tilt, 0 cycles|\"},{\"panel_id\":\"b\",\"text\":\"| Description|\\n|------------|\\n|SEM image at 30k, 0 cycles|\"},{\"panel_id\":\"c\",\"text\":\"| Description|\\n|------------|\\n|SEM image at 50k, 0 cycles|\"},{\"panel_id\":\"d\",\"text\":\"| Description|\\n|------------|\\n|SEM image at 20k tilt, 100 cycles|\"},{\"panel_id\":\"e\",\"text\":\"| Description|\\n|------------|\\n|SEM image at 30k, 100 cycles|\"},{\"panel_id\":\"f\",\"text\":\"| Description|\\n|------------|\\n|SEM image at 50k, 100 cycles|\"},{\"panel_id\":\"g\",\"text\":\"| Description|\\n|------------|\\n|SEM image at 20k tilt, 300 cycles|\"},{\"panel_id\":\"h\",\"text\":\"| Description|\\n|------------|\\n|SEM image at 30k, 300 cycles|\"},{\"panel_id\":\"i\",\"text\":\"| Description|\\n|------------|\\n|SEM image at 50k, 300 cycles|\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The controlled widening and smoothing of AAO pores after 100 and 300 pALE cycles demonstrate that the process enables highly selective material removal while preserving structural order. These improvements in pore uniformity, circularity, and surface smoothness make pALE-treated AAO substrates suitable for applications such as nanopatterning, filtration templates, and nanoscale device fabrication where precise geometry control is critical.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Compared to 0 cycles, the 100-cycle images show moderately enlarged and smoother pores, while the 300-cycle images exhibit significantly wider, more circular, and highly uniform pore structures.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Increased pore diameter, Rounded and smoother pore edges, Reduced surface irregularities, More consistent pore spacing and uniformity\"}]}]","bbox":[{"panel_id":"a","x":76,"y":45,"width":349,"height":254},{"panel_id":"b","x":76,"y":304,"width":347,"height":257},{"panel_id":"c","x":76,"y":568,"width":347,"height":256},{"panel_id":"d","x":434,"y":48,"width":342,"height":251},{"panel_id":"e","x":433,"y":306,"width":347,"height":251},{"panel_id":"f","x":434,"y":571,"width":344,"height":251},{"panel_id":"g","x":789,"y":45,"width":342,"height":253},{"panel_id":"h","x":788,"y":303,"width":344,"height":242},{"panel_id":"i","x":789,"y":570,"width":341,"height":249}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/images/fig_9.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/images/fig_9.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/CF4 plasma-based atomic layer etching of Al2O3.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"34","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":1136,"height":828,"image_format":"jpeg","image_sha256":"04b2b400989d9e306c4471dc3817efcd616bf6e9ab726e38d93cb7eb4223bcbb","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_36_fig_13.jpg","caption":"Fig. 13. TEM images of the trench pattern sample as  $\\mathrm{TiO_2}$  was deposited: (a) perspective of the pattern, (b) enlargement of the top of the pattern, (c) middle of the pattern at a depth of  $1400\\mathrm{nm}$  from the top, and (d) the bottom of the pattern. Images (e)–(j) are TEM images after eight cycles of the process. Images show samples after the cyclic process with  $\\mathrm{O_2}$  plasma treatment for  $30~\\mathrm{s}$  [(e)–(g)] and  $60~\\mathrm{s}$  [(h)–(j)].","id":"train/atomic-layer-etching/experimental-usecase/36/fig_13","sample_id":"atomic-layer-etching/experimental-usecase/36/fig_13","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"},{"panel_id":"e","label":"image panel"},{"panel_id":"f","label":"image panel"},{"panel_id":"g","label":"image panel"},{"panel_id":"h","label":"image panel"},{"panel_id":"i","label":"image panel"},{"panel_id":"j","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":0,"y":3,"width":257,"height":823},{"panel_id":"b","x":295,"y":3,"width":361,"height":267},{"panel_id":"c","x":295,"y":279,"width":359,"height":271},{"panel_id":"d","x":295,"y":561,"width":361,"height":265},{"panel_id":"e","x":672,"y":8,"width":345,"height":252},{"panel_id":"f","x":672,"y":283,"width":345,"height":258},{"panel_id":"g","x":670,"y":566,"width":345,"height":259},{"panel_id":"h","x":1033,"y":1,"width":345,"height":256},{"panel_id":"i","x":1033,"y":286,"width":338,"height":254},{"panel_id":"j","x":1042,"y":568,"width":333,"height":249}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig_13.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig_13.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/Cyclic C4F8 and O2 plasma etching of TiO2.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"36","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":1383,"height":828,"image_format":"jpeg","image_sha256":"c92ccce87442d38af8309be96aed0a6b0f512ac0f20515b8af7b39d4f0938dc6","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_36_fig_2.jpg","caption":"Fig. 2. Cross-sectional TEM images: (a) as deposited and (b) after  $\\mathrm{CF_4}$  plasma treatment for  $30~\\mathrm{s}$ .","id":"train/atomic-layer-etching/experimental-usecase/36/fig_2","sample_id":"atomic-layer-etching/experimental-usecase/36/fig_2","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":6,"y":13,"width":496,"height":670},{"panel_id":"b","x":530,"y":10,"width":376,"height":679}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/Cyclic C4F8 and O2 plasma etching of TiO2.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"36","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":908,"height":686,"image_format":"jpeg","image_sha256":"93fcc7267cc34391d83e4dee4bf28ac714ba1c8b9429ec51522223c080fe3629","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_36_fig_6.jpg","caption":"Fig. 6. TEM images: (a) as  $\\mathrm{TiO_2}$  deposited, (b) after  $\\mathrm{C_4F_8}$  plasma treatment, and (c) after  $\\mathrm{O_2}$  plasma treatment.","id":"train/atomic-layer-etching/experimental-usecase/36/fig_6","sample_id":"atomic-layer-etching/experimental-usecase/36/fig_6","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":4,"y":1,"width":414,"height":269},{"panel_id":"b","x":418,"y":8,"width":437,"height":260},{"panel_id":"c","x":860,"y":4,"width":346,"height":264}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/Cyclic C4F8 and O2 plasma etching of TiO2.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"36","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":1214,"height":272,"image_format":"jpeg","image_sha256":"dd0d49c279f7529112c126a5eb4682cfe8b535aed2c0ddd92f51e5964640cc9a","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_36_fig_7.jpg","caption":"Fig. 7. TEM images of  $\\mathrm{TiO_2}$  film after  $\\mathrm{C_4F_8}$  plasma treatment for (a)  $30~\\mathrm{s}$  and (b)  $60~\\mathrm{s}$ .","id":"train/atomic-layer-etching/experimental-usecase/36/fig_7","sample_id":"atomic-layer-etching/experimental-usecase/36/fig_7","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":2,"y":7,"width":383,"height":397},{"panel_id":"b","x":389,"y":7,"width":513,"height":396}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig_7.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/Cyclic C4F8 and O2 plasma etching of TiO2.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"36","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":905,"height":405,"image_format":"jpeg","image_sha256":"30ec45744cd3eec889abea7a9e7fdb7107e166a309cdbb38a4c6e363bbb90732","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_4_fig_4.jpg","caption":"FIG. 4. AFM micrograph of a photolithographically patterned ITO grating using ALE. The ITO grating has  $1\\mu \\mathrm{m}$  lines with  $3\\mu \\mathrm{m}$  period. Conventional photoresist was used as an etch mask. The etched areas are visibly smoother than unetched areas. 100 cycles using  $20\\mathrm{W}$  RF power was used. (a) shows the 3D view and (b) shows the conventional 2D AFM measurement of the same region. The scan area was  $10\\times 10\\mu \\mathrm{m}$","id":"train/atomic-layer-etching/experimental-usecase/4/fig_4","sample_id":"atomic-layer-etching/experimental-usecase/4/fig_4","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":0,"y":30,"width":670,"height":263},{"panel_id":"b","x":7,"y":359,"width":632,"height":454}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/4/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/4/images/fig_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/4/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/4/Atomic layer etching of indium tin oxide.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"4","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":678,"height":817,"image_format":"jpeg","image_sha256":"9e39757d836e4002d65b7fdf84e3193c37307b9b454a1603e615508f355b61f3","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_43_fig_4.jpg","caption":"Fig. 4. Cross-section TEM image of ALD  $\\mathrm{Al_2O_3}$  film on a trench pattern: (a) as-deposited and (b) after 30 ALE cycles at  $250^{\\circ}\\mathrm{C}$ .","id":"train/atomic-layer-etching/experimental-usecase/43/fig_4","sample_id":"atomic-layer-etching/experimental-usecase/43/fig_4","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":0,"y":0,"width":536,"height":508},{"panel_id":"b","x":573,"y":0,"width":532,"height":508}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/43/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/43/images/fig_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/43/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/43/Surface reaction during thermal atomic layer etching of aluminum oxide.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"43","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":1105,"height":508,"image_format":"jpeg","image_sha256":"2b4215d7c918936bdf52963057f52f134ea8f67d570f63f524a653a27268373d","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_9_figure_2.jpg","caption":"Figure 2. (a) Scanning electron microscopy (SEM) images of InGaAs/InAlAs vertical nanowires (VNWs) after  $\\mathrm{BCl}_3 / \\mathrm{SiCl}_4 / \\mathrm{Ar}$  RIE and after 300 and 600 cycles of HF/DMAC thermal ALE at  $250^{\\circ}C$  b) SEM images of InGaAs/InAlAs VNWs before and after 250 cycles of HF/DMAC thermal ALE at a higher reaction temperature of  $300^{\\circ}C$  . VNWs after 500 cycles were not shown because the InAlAs in the VNWs was completely etched. (c) XRR measurement of a planar InGaAs film after  $300^{\\circ}C$  HF/DMAC thermal ALE, confirming the higher etch rate of  $\\sim 0.18\\mathrm{\\AA}/$  cycle. (d) Summary of changes in VNW radius versus number of cycles for InGaAs and InAlAs VNWs at 250 and  $300^{\\circ}C$","id":"train/atomic-layer-etching/experimental-usecase/9/figure_2","sample_id":"atomic-layer-etching/experimental-usecase/9/figure_2","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"multi spectra chart"},{"panel_id":"d","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The image panel consists of series of ccanning electron microscopy (SEM) images of InGaAs/InAlAs vertical nanowires (VNWs) during different stages of HF/DMAC ALE process at 250 °C\"},{\"panel_id\":\"b\",\"text\":\"The image panel consists of series of ccanning electron microscopy (SEM) image at different ALE stages at  300 °C\"},{\"panel_id\":\"c\",\"text\":\"The line chart shows the change in X-ray reflectivity intensity with respect to omega-2θ (arcsec) for different cycles of ALE.\"},{\"panel_id\":\"d\",\"text\":\"The line chart illustrates the change in radius (ΔRadius) over the number of ALE cycles for InAlAs and InGaAs at different temperatures.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Condition       | InGaAs Thickness | InAlAs Thickness |\\n|-----------------|-------------------|-------------------|\\n| After RIE       | 28 nm            | 28nm              |\\n| 300 cycles      | 24 nm            | 18 nm            |\\n| 600 cycles      | 24 nm            | 10 nm            |\"},{\"panel_id\":\"b\",\"text\":\"| Condition       | InGaAs Thickness | InAlAs Thickness |\\n|-----------------|-------------------|-------------------|\\n| After RIE       | 35 nm            | 35 nm               |\\n| 250 cycles      | 24 nm            | 4 nm             |\"},{\"panel_id\":\"c\",\"text\":\"| Cycles    | Density edge intensity |\\n|-----------|--------------------|\\n| Pristine  |10^9  |\\n| 200       |10^7     |\\n| 450       |10^5     |\"},{\"panel_id\":\"d\",\"text\":\"| Material  | Temperature | Growth Rate (Å/cycle) |\\n|-----------|------------|-------------------------|\\n| InAlAs    | 300°C      | 0.62                  |\\n| InGaAs    | 300°C      | —                     |\\n| InAlAs    | 250°C      | —                     |\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The partial etching selectivity between InGaAs and InAlAs  might occur because fluorination of AlAs (Gibbs free energy ΔG° = −86.0 kcal/mol) is thermochemically more favorable than the fluorination of GaAs (ΔG° = −17.4 kcal/mol) at 250 °C. This behavior is also consistent with the faster oxidation of InAlAs versus InGaAs at 300−500 °C.Fluorination/oxidation of InAlAs is probably more favorable than for InGaAs because Al−F/Al-O bonds are stronger than Ga−F/Ga-O bonds.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The radial etch rates of InGaAs and InAlAs VNWs at 300 °C are as follows. \\n1. After the first 250 cycles, InGaAs is 0.24 and InAlAs is 0.62 Å/cycle on average. \\n2. After the second 250 cycles,  InGaAs  is 0.10 and InAlAs decreases to 0.41 Å/cycle on average.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The negligible etching of InGaAs after 300 cycles of thermal ALE at 250 °C may be caused by the low temperature of 250 °C.\"}]}]","bbox":[{"panel_id":"c","x":-3,"y":358,"width":442,"height":375},{"panel_id":"b","x":461,"y":11,"width":375,"height":320},{"panel_id":"a","x":11,"y":6,"width":459,"height":354},{"panel_id":"d","x":432,"y":360,"width":411,"height":368}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/9/images/figure_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/9/images/figure_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/9/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/9/In Situ Thermal Atomic Layer Etching for Sub5 nm InGaAs.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"9","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":847,"height":775,"image_format":"jpeg","image_sha256":"d1188a21a822a1865fe9be05cc48bdf89d8f70a6e31c9f944c3afbeb16695b6c","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_9_figure_3.jpg","caption":"Figure 3. SEM side-view images of InGaAs/InAlAs fins after (a) fin RIE and after subsequent thermal ALE with HF/DMAC under various conditions: (b)  $P_{\\mathrm{DMAC}} = 40 \\mathrm{mTorr}$  for 250 and 600 cycles at  $250^{\\circ}\\mathrm{C}$ , (c)  $P_{\\mathrm{DMAC}} = 150 \\mathrm{mTorr}$  for 500 cycles at  $250^{\\circ}\\mathrm{C}$ , and (d)  $P_{\\mathrm{DMAC}} = 40 \\mathrm{mTorr}$  for 250 and 500 cycles at  $300^{\\circ}\\mathrm{C}$ . (e,f) AFM measurement of RMS surface roughness and topography of a planar InGaAs surface after sequential thermal ALE (HF/DMAC,  $P_{\\mathrm{DMAC}} = 40 \\mathrm{mTorr}$ ,  $300^{\\circ}\\mathrm{C}$ ). In all cases, the InGaAs wafer was etched by  $\\mathrm{BCl}_3 / \\mathrm{SiCl}_4 / \\mathrm{Ar}$  RIE prior to the thermal ALE treatment.","id":"train/atomic-layer-etching/experimental-usecase/9/figure_3","sample_id":"atomic-layer-etching/experimental-usecase/9/figure_3","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"},{"panel_id":"e","label":"line chart"},{"panel_id":"f","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The image panel shows the surface morphology of InGaAs after RIE before ALE\"},{\"panel_id\":\"b\",\"text\":\"The image panel shows the surface morphology of InGaAs after 250 and 600 cycles of ALE at 40m Torr and 250 °C\"},{\"panel_id\":\"c\",\"text\":\"The image panel shows the surface morphology of InGaAs after 500 cycles of ALE at 140m Torr and 250 °C\"},{\"panel_id\":\"d\",\"text\":\"The image panel shows the surface morphology of InGaAs after 250 and 600 cycles of ALE at 40m Torr and 300 °C\"},{\"panel_id\":\"e\",\"text\":\"The line chart illustrates the change in RMS surface roughness of etched InGaAs surface with increasing number of cycles under HF/DMAC at 40 mT and 300°C.\"},{\"panel_id\":\"f\",\"text\":\"The image panel includes three AFM images showing the surface topography of InGaAs after RIE and after different cycles of thermal ALE at 40 mT and 300°C.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"\"},{\"panel_id\":\"c\",\"text\":\"\"},{\"panel_id\":\"d\",\"text\":\"\"},{\"panel_id\":\"e\",\"text\":\"| Number of Cycles | RMS Surface Roughness (nm) |\\n|-----------------|--------------------------|\\n| 0               | 0.5                      |\\n| 150             | 0.5                      |\\n| 200             | 0.48                      |\\n| 250             | 0.7                      |\"},{\"panel_id\":\"f\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Thermal ALE can preserve or possibly improve sidewall quality even after hundreds of cycles. The improvement in fin sidewall roughness, in particular, can have a significant impact on FinFET performance as the AFM image shows that the surface topography is preserved after 150 and 200 cycles.\"}]},{\"panel_id\":\"f\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]},{\"panel_id\":\"e\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"At lower DMAC pressure and lower temperature, the fin sidewalls become rougher after 250 and 600 cycles (b). In the cases of (c) or (d), with a higher pressure DMAC or higher temperature, the quality of the sidewall and planar surfaces significantly improves. Overall, process (image d) with the higher\\ntemperature of 300 °C for 250 cycles produces the smoothest fin sidewall and planar surfaces.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The slight increase of Rrms after 250 thermal ALE cycles may occur because InGaAs and InAlAs are ternary materials. When they are etched, the surface can suffer from preferential etching of one metal over the other. XPS analysis  confirmed that the In/Ga ratio decreased from 1.0 to 0.8−0.9 after 600 thermal ALE cycles at 250−300 °C as expected from the preferential etching of InAs compared with GaAs. There was also increased Al content perhaps resulting from partial conversion of InGaAs/InAlAs to AlAs during the DMAC reaction.\"}]}]","bbox":[{"panel_id":"d","x":715,"y":2,"width":392,"height":264},{"panel_id":"b","x":181,"y":2,"width":347,"height":272},{"panel_id":"c","x":530,"y":0,"width":186,"height":281},{"panel_id":"e","x":7,"y":306,"width":340,"height":299},{"panel_id":"f","x":340,"y":302,"width":758,"height":292},{"panel_id":"a","x":9,"y":6,"width":168,"height":268}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/9/images/figure_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/9/images/figure_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/9/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/9/In Situ Thermal Atomic Layer Etching for Sub5 nm InGaAs.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"9","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":1106,"height":639,"image_format":"jpeg","image_sha256":"9db4debd535cc6f1a25e26c4c246f1dd3fee63db03773418057401e913ef9c72","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_17_fig_2.jpg","caption":"Fig. 2. (a) Twelve substrate regions for microscopic simulations. (b) Twelve substrate positions for the investigation of the flow distribution.","id":"train/atomic-layer-etching/simulation-usecase/17/fig_2","sample_id":"atomic-layer-etching/simulation-usecase/17/fig_2","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":0,"y":9,"width":356,"height":351},{"panel_id":"b","x":483,"y":0,"width":120,"height":360}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/17/images/fig_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/17/images/fig_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/17/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/17/Multiscale computational fluid dynamics modeling of thermal atomic layer etching.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"17","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":603,"height":411,"image_format":"jpeg","image_sha256":"c4aad4cd56b3cd9195484873f7bd9bc2b6baeb59cbf33a8a26d5e4716fdc69e4","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_23_fig_8.jpg","caption":"FIG. 8. TEM cross sections of a  $4 \\times 50 \\mathrm{nm}$  InGaAs nanowire covered by  $4.7 \\mathrm{nm}$  ALD-grown  $\\mathrm{Al}_2\\mathrm{O}_3$  and  $20 \\mathrm{nm}$  of ALD tungsten. Reprinted with permission from Lu et al., in International Electron Devices Meeting, San Francisco, 1-5 December 2018 (IEDM, Montgomery Village, 2018), p. 895. Copyright 2018, IEEE.","id":"train/atomic-layer-etching/simulation-usecase/23/fig_8","sample_id":"atomic-layer-etching/simulation-usecase/23/fig_8","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"b","x":414,"y":5,"width":250,"height":400},{"panel_id":"a","x":6,"y":4,"width":406,"height":398}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/23/images/fig_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/23/images/fig_8.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/23/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/23/Review Paper -- Thermal atomic layer etching A review.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"23","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":667,"height":406,"image_format":"jpeg","image_sha256":"30663a1ececfacb558c60061d01e2c009ff6a13d27ca7a63ac8ee9641ad01f2e","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_3_fig_15.jpg","caption":"FIG. 15. (Color online) Profiles resulting from etching the gate structure with a (a)–(c) continuous etching process, and (d)–(f) the optimized ALE process. Time increases from left to right. Frames are taken at equal over-etch (as a percentage of the time required to expose the bottom  $\\mathrm{SiO}_2$ ), not at equal etch times. The etch times listed for the ALE process (d)–(f) are active (plasma on) times, and ignore any purge or dwell times necessary for a functional ALE process.","id":"train/atomic-layer-etching/simulation-usecase/3/fig_15","sample_id":"atomic-layer-etching/simulation-usecase/3/fig_15","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"},{"panel_id":"d","label":"image panel"},{"panel_id":"e","label":"image panel"},{"panel_id":"f","label":"image panel"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"d","x":0,"y":503,"width":516,"height":497},{"panel_id":"b","x":497,"y":61,"width":474,"height":411},{"panel_id":"c","x":957,"y":66,"width":455,"height":413},{"panel_id":"e","x":511,"y":511,"width":457,"height":482},{"panel_id":"f","x":975,"y":515,"width":439,"height":478},{"panel_id":"a","x":5,"y":55,"width":510,"height":447}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/3/images/fig_15.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/3/images/fig_15.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/3/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/3/Atomic layer etching of 3D structures in silicon Self-limiting.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"3","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":1422,"height":1000,"image_format":"jpeg","image_sha256":"93831789b5acba8f13bba6e4a117976473f1870a9bafcddddd9087e8a9e45861","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_32_fig_3.jpg","caption":"FIG. 3. (a) The perspective view of Connolly surface constructed on  $\\mathrm{BCl}_3$  adsorbed  $\\mathrm{Al}_2\\mathrm{O}_3(100)$  using a probe atom with a radius of  $2.77\\mathrm{\\AA}$ ; (b) The time-resolved variations of  $\\mathrm{Al}_2\\mathrm{O}$  molar ratio in  $\\mathrm{Al}_2\\mathrm{O}_3(100)$  surface during the Ar bombardment (XPS measurements); (c) Atomic ASAs obtained from the above Connolly surface (unit:  $\\mathring{\\mathrm{a}}^2$ ). The oxygen, aluminium, boron, and chlorine atoms are shown in red, violet, bronze, and green, respectively.","id":"train/atomic-layer-etching/simulation-usecase/32/fig_3","sample_id":"atomic-layer-etching/simulation-usecase/32/fig_3","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"line chart"},{"panel_id":"c","label":"molecular structure diagram"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Perspective rendering of the Connolly surface for a BCl3 adsorbed Al2O3(100) model, constructed using a probe radius of 2.77 Å (Ar van der Waals radius).\"},{\"panel_id\":\"b\",\"text\":\"XPS based Al to O molar ratio versus Ar irradiation time, showing an initial decrease followed by recovery as irradiation continues.\"},{\"panel_id\":\"c\",\"text\":\"Atomic accessible surface areas (ASA) computed from the Connolly surface, with Cl atoms showing the largest ASAs (tens of Å^2 per atom) compared with other surface atoms.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"\"},{\"panel_id\":\"b\",\"text\":\"| Ar Irradiation Time (sec) | A/O Molar Ratio |\\n|---|---|\\n| Ref. | 0.55 |\\n| 75 | 0.45 |\\n| 100 | 0.48 |\\n| 125 | 0.55 |\"},{\"panel_id\":\"c\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"A Connolly surface for the BCl3 adsorbed Al2O3(100) structure, generated using a 2.77 Å probe radius.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The chlorine atoms have the largest accessible surface areas, roughly 25 to 50 Å^2 per Cl atom.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The very large ASAs on Cl indicate that Cl dominates the outermost accessible surface, so Ar is most likely to strike Cl first. Because Cl is directly bonded to surface Al in the adsorption structure, these impacts can transfer energy efficiently into Al containing units, supporting the idea that Al associated moieties detach early.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The Al to O ratio drops at the early stage (around 75 s) and then increases again at longer irradiation times (100 to 125 s), moving back toward the reference value.\"}]}]","bbox":[{"panel_id":"a","x":1,"y":1,"width":314,"height":340},{"panel_id":"b","x":316,"y":1,"width":348,"height":345},{"panel_id":"c","x":32,"y":348,"width":626,"height":457}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/32/images/fig_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/32/images/fig_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/32/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/32/Understanding time-resolved processes in atomic-layer etching of ultra-thin Al2O3 film.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"32","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:image_caption"},"width":667,"height":811,"image_format":"jpeg","image_sha256":"d561210cf3896c1b4b9705b96b527201deeda52d6f6ec67d70b55ce80e50104f","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_4_figure_6.jpg","caption":"Figure 6. BF-TEM images of a) the reference unetched  $25 \\text{nm}$  TiN film on  $22 \\text{nm} \\text{SiO}_2$  film, b) after complete etching of  $25 \\text{nm}$  TiN film by 600 cycles of each  $3 \\text{s}$  long  $\\text{CCl}_4$  pulse separated by  $6 \\text{s}$  of  $\\text{N}_2$  purges.","id":"train/atomic-layer-etching/simulation-usecase/4/figure_6","sample_id":"atomic-layer-etching/simulation-usecase/4/figure_6","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"BF-TEM image of the reference unetched TiN/SiO₂ system, showing a 25 nm TiN film on 22 nm SiO₂. Spin-on carbon was used to enhance material contrast.\"},{\"panel_id\":\"b\",\"text\":\"BF-TEM image of the same sample after 600 cycles of 3 s CCl₄ pulses with 6 s N₂ purges, showing complete removal of the TiN film and a clean SiO₂ surface without residues.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Description |\\n|---|\\n| Reference unetched 25 nm TiN film on 22 nm SiO₂. Shows the original TiN layer for comparison. |\"},{\"panel_id\":\"b\",\"text\":\"| Description |\\n|---|\\n| TiN film completely removed after 600 cycles of 3 s CCl₄ pulses with 6 s N₂ purges. The SiO₂ surface appears clean and free of residues. |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"25 nm.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Before etching, the SiO₂ substrate is covered by a 25 nm TiN film, which is clearly visible in panel (a). After etching (panel b), the TiN film is completely removed, leaving the SiO₂ surface clean and smooth, with no visible residues from either the TiN layer or the etch process. The spin-on carbon used during TEM imaging highlights the material contrast but does not affect the interpretation of the substrate’s condition.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Provides high-resolution cross-sectional visualization of thin films, Confirms complete removal of the TiN layer, Allows assessment of the substrate’s surface integrity post-etch, Highlights any residual contamination or defects from the etch process\"}]}]","bbox":[{"panel_id":"a","x":4,"y":3,"width":247,"height":572},{"panel_id":"b","x":255,"y":2,"width":387,"height":578}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/4/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/4/images/figure_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/4/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/4/Combining Experimental and DFT Investigation.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"4","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":653,"height":583,"image_format":"jpeg","image_sha256":"a5c390c29ef6ef70190e9c91ba824e5797cc8c5e43cd0b2ba84169ab986b2cc9","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_4_figure_8.jpg","caption":"Figure 8. Cross-sectional bright field transmission electron micrograph (BF-TEM) of 3D structures: a) about  $3.3 \\text{nm}$  TiN film deposited on  $\\text{SiO}_2$  fins with lateral cavities, b) the same after about  $0.8 \\text{nm}$  TiN is etched by the  $\\text{NbF}_5\\text{-}\\text{CCl}_4$  etch-process, and c) after removing the TiN film completely. The etching was performed at  $460^{\\circ}\\text{C}$ .","id":"train/atomic-layer-etching/simulation-usecase/4/figure_8","sample_id":"atomic-layer-etching/simulation-usecase/4/figure_8","subset":"image-panel","split":"train","classification":[{"panel_id":"a","label":"image panel"},{"panel_id":"b","label":"image panel"},{"panel_id":"c","label":"image panel"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"BF-TEM cross-section of pristine SiO₂ fin structures coated with a ~3.3 nm TiN layer, showing high conformality.\"},{\"panel_id\":\"b\",\"text\":\"BF-TEM cross-section after partial etching (~0.8 nm) of the TiN layer using NbF₅–CCl₄ at 460 °C, showing minimal surface defects.\"},{\"panel_id\":\"c\",\"text\":\"BF-TEM cross-section after complete removal of the TiN film, showing intact SiO₂ fins and confirming etch selectivity.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"|Description|\\n|-----------|\\n| TEM image of SiO₂ fin structure coated with ~3.3 nm pristine TiN film, showing high conformality. |\"},{\"panel_id\":\"b\",\"text\":\"|Description|\\n|-----------|\\n| TEM image after ~0.8 nm TiN conformally etched by the NbF₅–CCl₄ process, with no significant surface defects. |\"},{\"panel_id\":\"c\",\"text\":\"|Description|\\n|-----------|\\n| TEM image after complete removal of the TiN film (over‑etch), with SiO₂ fins remaining intact, confirming etch selectivity. |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"No significant increase in surface defects was observed after ~0.8 nm TiN removal.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Selective removal of TiN\\nPreservation of SiO₂ fins\\nHigh conformality maintained\\nPotential for advanced 3D nanostructure processing\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The conformality of TiN films plays a critical role in the effectiveness of gas‑phase etching. Because the TiN layer uniformly coats even non‑line‑of‑sight features, the NbF₅–CCl₄ etch process can remove material consistently across complex geometries. This structural property enhances isotropic etching and ensures selectivity, making conformal TiN deposition essential for reliable thinning and removal.\"}]}]","bbox":[{"panel_id":"a","x":1,"y":7,"width":454,"height":447},{"panel_id":"b","x":463,"y":5,"width":451,"height":454},{"panel_id":"c","x":923,"y":7,"width":451,"height":447}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/4/images/figure_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/4/images/figure_8.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/4/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/4/Combining Experimental and DFT Investigation.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"4","first_classification_panel_id":"a","first_classification_label":"image panel","caption_source":"content.json:img_caption"},"width":1375,"height":461,"image_format":"jpeg","image_sha256":"15f30b9b7fead45f5369e70682a9591f2c13c81b4e87853a35b78f5f299740df","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}