{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_26_figure_5.jpg","caption":"Figure 5. In situ QMS measurements for (a)  $m = 66$  from cyclopentadiene and (b)  $m = 48$  from ozone, measured during ITO ALD using  $5\\%$ $\\mathrm{SnO_2}$  cycles. The application of the  $\\mathrm{SnO_2}$  ALD cycles is indicated.","id":"train/atomic-layer-deposition/experimental-usecase/26/figure_5","sample_id":"atomic-layer-deposition/experimental-usecase/26/figure_5","subset":"bar-chart","split":"train","classification":[{"panel_id":"a","label":"bar chart"},{"panel_id":"b","label":"bar chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The bar chart shows the increasing intensity of the m=66 signal (C₅H₆) over time during repeated SnO₂ ALD cycles.\"},{\"panel_id\":\"b\",\"text\":\"The bar chart shows the fluctuating m=48 signal (ozone) intensity over time, with signal decay after each pulse.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Time (s) | m=66 Signal (AU) | m=48 Signal (AU) |\\n|----------|------------------|------------------|\\n| 0        | 0                | 0                |\\n| 100      | 2                | 2                |\\n| 200      | 4                | 3                |\\n| 300      | 6                | 5                |\\n| 400      | 9                | 6                |\\n| 500      | 12               | 7                |\\n| 600      | 15               | 3                |\\n| 700      | 13               | 2                |\\n| 800      | 14               | 1                |\\n| 900      | 15               | 0                |\\n| 1000     | 14               | 0                |\\n| 1100     | 0                | 0                |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Precursor injection timing , Byproduct removal, Reaction step resolution\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Cyclopentadiene signal increases with each cycle, while ozone shows decaying or pulsed signals that drop off more quickly.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"m=66 (cyclopentadiene), m=48 (ozone)\"}]}]","bbox":[{"panel_id":"a","x":9,"y":0,"width":557,"height":575},{"panel_id":"b","x":591,"y":12,"width":556,"height":565}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_5.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":"bar chart","caption_source":"content.json:image_caption"},"width":1147,"height":578,"image_format":"jpeg","image_sha256":"b8e96dd1bd0d529443fd4c9718b2718d33718dc42f5d9df400e7bd8fcd81c713","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_66_figure_5.jpg","caption":"Figure 5. Statistical distributions of  $H_{c}$  variation for (a)  $\\mathrm{ZnO}$ ; (b) oxygen-deficient  $\\mathrm{ZnO}$ ; and (c)  $\\mathrm{Ni}$ , (d)  $\\mathrm{Co}$ , and (e) Fe-doped  $\\mathrm{ZnO}$  layers.","id":"train/atomic-layer-deposition/experimental-usecase/66/figure_5","sample_id":"atomic-layer-deposition/experimental-usecase/66/figure_5","subset":"bar-chart","split":"train","classification":[{"panel_id":"a","label":"bar chart"},{"panel_id":"b","label":"bar chart"},{"panel_id":"c","label":"bar chart"},{"panel_id":"d","label":"bar chart"},{"panel_id":"e","label":"bar chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The bar chart shows the distribution of coercivity (H<sub>c</sub>) values for ZnO samples\"},{\"panel_id\":\"b\",\"text\":\"The bar chart shows the distribution of coercivity (H<sub>c</sub>) values for oxygen-deficient ZnO samples\"},{\"panel_id\":\"c\",\"text\":\"The bar chart shows the distribution of coercivity (H<sub>c</sub>) values for ZnO:Ni samples\"},{\"panel_id\":\"d\",\"text\":\"The bar chart shows the distribution of coercivity (H<sub>c</sub>) values for ZnO:Co samples\"},{\"panel_id\":\"e\",\"text\":\"The bar chart shows the distribution of coercivity (H<sub>c</sub>) values for ZnO:Fe samples\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| H<sub>c</sub> (Oe) | Relative frequency (%) |\\n|---------------|-------|\\n| 0-500           | 76% |\\n| 500-1000    | 19% |\\n| 1000-4000 | 5% |\"},{\"panel_id\":\"b\",\"text\":\"| H<sub>c</sub> (Oe) | Relative frequency (%) |\\n|---------------|-------|\\n| 0-500           | 70% |\\n| 500-1000    | 23% |\\n| 1000-4000 | 7% |\"},{\"panel_id\":\"c\",\"text\":\"| H<sub>c</sub> (Oe) | Relative frequency (%) |\\n|---------------|-------|\\n| 0-500           | 41% |\\n| 500-1000    | 40% |\\n| 1000-4000 | 19% |\"},{\"panel_id\":\"d\",\"text\":\"| H<sub>c</sub> (Oe) | Relative frequency (%) |\\n|---------------|-------|\\n| 0-500           | 35% |\\n| 500-1000    | 39% |\\n| 1000-4000 | 26% |\"},{\"panel_id\":\"e\",\"text\":\"| H<sub>c</sub> (Oe) | Relative frequency (%) |\\n|---------------|-------|\\n| 0-500           | 33% |\\n| 500-1000    | 36% |\\n| 1000-4000 | 21% |\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"The fraction of high magnetization values increase  in the following order, for Hc > 1000 Oe, \\n1. ZnO/Ni = 19%\\n2. ZnO/Co= 26% \\n3. ZnO/Fe=31%.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It is known that the MOKE is proportional to spin−orbit interaction and net spin polarization; hence, it preferentially senses magnetic electrons (i.e., in our case 3d electrons). Therefore, these results unambiguously demonstrate that the observed magnetic behavior of TM-doped ZnO layers stems from the magnetic properties of the dopant; hence, they give evidence that the applied ALD scheme resulted in the formation of a (diluted) magnetic semiconductor\"}]},{\"panel_id\":\"c\",\"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\":\"Magnetization for all TM-doped ZnO layers increases substantially\\u0001the regions with small magnetization decrease to about 33−41% (in dependence on the\\ndopant) compared to more than 70% for the undoped layers. Respectively, the frequencies of medium and large magnetization values increase.\"}]}]","bbox":[{"panel_id":"a","x":40,"y":0,"width":798,"height":140},{"panel_id":"b","x":40,"y":141,"width":797,"height":144},{"panel_id":"c","x":33,"y":287,"width":800,"height":144},{"panel_id":"d","x":37,"y":430,"width":800,"height":163},{"panel_id":"e","x":40,"y":595,"width":798,"height":213}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/66/images/figure_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/66/images/figure_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/66/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/66/Paskaleva et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"66","first_classification_panel_id":"a","first_classification_label":"bar chart","caption_source":"content.json:image_caption"},"width":841,"height":811,"image_format":"jpeg","image_sha256":"ac351459c727715169939cee62a57259cb96e6fb73ea306663b0c6f42bb61da0","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_66_figure_6.jpg","caption":"Figure 6. Statistical distributions of Kerr angle variations for (a)  $\\mathrm{ZnO}$ ; (b) oxygen-deficient  $\\mathrm{ZnO}$ ; and (c)  $\\mathrm{Ni}$ , (d)  $\\mathrm{Co}$ , and (e) Fe-doped  $\\mathrm{ZnO}$  layers.","id":"train/atomic-layer-deposition/experimental-usecase/66/figure_6","sample_id":"atomic-layer-deposition/experimental-usecase/66/figure_6","subset":"bar-chart","split":"train","classification":[{"panel_id":"a","label":"bar chart"},{"panel_id":"b","label":"bar chart"},{"panel_id":"c","label":"bar chart"},{"panel_id":"d","label":"bar chart"},{"panel_id":"e","label":"bar chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The bar chart shows the distribution of θ_Kerr values for undoped ZnO\"},{\"panel_id\":\"b\",\"text\":\"The bar chart shows the distribution of θ_Kerr values for oxygen deficient ZnO\"},{\"panel_id\":\"c\",\"text\":\"The bar chart shows the distribution of θ_Kerr values for transition metal doped ZnO:Ni categorized into three ranges: 0<|±θ_Kerr|<100 mdeg, 100<|±θ_Kerr|<1000 mdeg, and |±θ_Kerr|>1000 mdeg.\"},{\"panel_id\":\"d\",\"text\":\"The bar chart shows the distribution of θ_Kerr values for transition metal doped ZnO:Co\"},{\"panel_id\":\"e\",\"text\":\"The bar chart shows the distribution of θ_Kerr values for transition metal doped ZnO:Fe\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| θ_Kerr Range (mdeg) | Relative Frequency (%) |\\n|---|---|\\n| 0–5   | 5.5 |\\n| 5–10  | 5.0 |\\n| 10–15 | 4.5 |\\n| 15–20 | 4.0 |\\n| 20–25 | 3.5 |\\n| 25–30 | 3.0 |\\n| 30–35 | 2.5 |\\n| 35–40 | 2.0 |\\n| 40–45 | 1.5 |\\n| 45–50 | 1.0 |\\n| 50–60 | 0.8 |\\n| 60–70 | 0.4 |\\n| 70–80 | 0.2 |\\n| 80–90 | 0.1 |\\n| 90-100 | 0 |\"},{\"panel_id\":\"b\",\"text\":\"| θ_Kerr Range (mdeg) | Relative Frequency (%) |\\n|---|---|\\n| 0–5   | 6.5 |\\n| 5–10  | 6.0 |\\n| 10–15 | 5.5 |\\n| 15–20 | 5.0 |\\n| 20–25 | 4.0 |\\n| 25–30 | 3.5 |\\n| 30–35 | 3.0 |\\n| 35–40 | 2.5 |\\n| 40–45 | 2.0 |\\n| 45–50 | 1.5 |\\n| 50–60 | 1.0 |\\n| 60–70 | 0.5 |\\n| 70–80 | 0.2 |\\n| 80–90 | 0.1 |\\n| 90–100| 0 |\"},{\"panel_id\":\"c\",\"text\":\"| Max Relative frequency (%) | 0<abs(±θ_Kerr)<100 mdeg | 100< abs(±θ_Kerr)<1000 mdeg | abs(±θ_Kerr)>1000 mdeg |\\n|---|---|---|---|\\n| 15 | 55% | 33% | 12% |\"},{\"panel_id\":\"d\",\"text\":\"| Max Relative frequency (%) | 0<abs(±θ_Kerr)<100 mdeg | 100< abs(±θ_Kerr)<1000 mdeg | abs(±θ_Kerr)>1000 mdeg |\\n|---|---|---|---|\\n| 10| 25% | 55% | 20% |\"},{\"panel_id\":\"e\",\"text\":\"| Max Relative frequency (%) | 0<abs(±θ_Kerr)<100 mdeg | 100< abs(±θ_Kerr)<1000 mdeg | abs(±θ_Kerr)>1000 mdeg |\\n|---|---|---|---|\\n| 10 | 45% | 22% | 33% |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"All TM-doped layers  exhibit very large θKerr values, which are symmetrically distributed with respect to 0. Depending on the dopant, similar trends to those observed for the coercivity appeared from the θKerr distributions. The fraction with\\n| Kerr| > 100 mdeg increases with the dopant in the order Ni, Co, and Fe. The fraction of very large | Kerr| > 1000 mdeg\\nincreases in the same order, that is, 12% for ZnO/Ni; 20% for ZnO/Co, and 33% for ZnO/Fe.\"}]},{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The broadly varied MO Kerr parameters are dependent on localized Fe, Co, and Ni concentration variations, implying randomized inhomogeneous distribution of the dopants in the crystal structure. On the other hand, considering the alternating layer by layer (ZnO TMO) deposition process of the ALD technique, we have to point out the significant MOKEs detected in ZnO-based MO cavity/dielectric (with low optical losses) nano-laminates (Fabry−Perot type). The optical enhancement (several orders of magnitude) of MO Kerr polarization rotation in such structures is related to internal interference effects and the addition of multiple reflections in the nanolayered architecture\\u0001dependent on the layer arrangement, thickness, and the reflective properties of the substrate material\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The undoped ZnO layers (a and b) reveal very  weak magenetoptic (MO) activity with the majority of θKerr values less than 20\\nmdeg\"}]}]","bbox":[{"panel_id":"a","x":30,"y":0,"width":931,"height":182},{"panel_id":"b","x":29,"y":180,"width":944,"height":186},{"panel_id":"c","x":46,"y":365,"width":921,"height":174},{"panel_id":"d","x":47,"y":543,"width":922,"height":179},{"panel_id":"e","x":44,"y":721,"width":928,"height":230}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/66/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/66/images/figure_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/66/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/66/Paskaleva et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"66","first_classification_panel_id":"a","first_classification_label":"bar chart","caption_source":"content.json:image_caption"},"width":980,"height":956,"image_format":"jpeg","image_sha256":"fa4a768a88fe6ca2a7e695bd112877f16173711b67664a222441340906545663","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_7.jpg","caption":"Figure 7: Density functional theory model schematics used to calculate metal co-segregation on the B-site with a cation exchange and O vacancy: (a) the co-segregation energy in eV of the deposited metal; (b) the exchange energy comparison of the different deposited transition metals; and (c) DFT calculation representations for the cation exchange and alloy formation [117].","id":"train/atomic-layer-deposition/simulation-usecase/17/figure_7","sample_id":"atomic-layer-deposition/simulation-usecase/17/figure_7","subset":"bar-chart","split":"train","classification":[{"panel_id":"a","label":"bar chart"},{"panel_id":"b","label":"bar chart"},{"panel_id":"c","label":"process flow diagram"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The bar chart represents the Density functional theory model schematics used to calculate metal co-segregation on the B-site with a cation exchange and O vacancy where the co-segregation energy is represented in eV of the deposited metal\"},{\"panel_id\":\"b\",\"text\":\"The bar chart represents  the exchange energy comparison of the different deposited transition metals showing two different regions 'Exchangeable region' and 'Non-exchangeable region'.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Deposited metal | Cosegregation energy (eV) |\\n|---|---|\\n| Ti | -0.12 |\\n| Cr | -1.05 |\\n| Fe | -1.45 |\\n| Mn | -1.89 |\\n| Co | -2.92 |\\n| Ni | -3.32 |\\n| Cu | -4.32 |\"},{\"panel_id\":\"b\",\"text\":\"| Deposited metal | Exchange energy (eV) |\\n|---|---|\\n| Ti | -3.2 |\\n| Cr | -2.27 |\\n| Fe | -1.87 |\\n| Mn | -1.43 |\\n| Co | -0.4 |\\n| Ni | 0 |\\n| Cu | 1.0 |\"},{\"panel_id\":\"c\",\"text\":\"\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"This implies it is thermodynamically unfavorable for Ni to spontaneously exsolve from this specific host oxide on its own. To form Ni nanoparticles, you cannot rely on self-exsolution; you must use an external driving force, such as the ALD-deposited Fe layer shown in figure (a), to provide the chemical potential gradient needed to pull Ni to the surface.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The  negative sign indicates that the co-segregation or exchange process is exothermic and spontaneous for all these metals. This suggests that once the metal is deposited and thermal energy is applied, the system will naturally move toward the exchanged state without requiring additional energy input to drive the reaction.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The ALD-deposited Fe method is superior. Traditional bulk Fe doping would homogeneously alter the composition and, thus, the bulk electronic properties. ALD deposits Fe only on the surface, creating a localized driving force for Ni to exsolve from the immediate sub-surface without changing the bulk's intrinsic properties.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The driving force becomes more favorable (more negative energy) as you move from Ti to Fe, then becomes less favorable after Fe through Cu. Fe is the most favorable. This trend is likely governed by the metal's binding energy or oxidation state stability within the specific host oxide lattice, with Fe achieving the optimal ionic size and charge match to minimize the system's total energy upon exchange.\"}]}]","bbox":[{"panel_id":"c","x":6,"y":575,"width":1321,"height":480},{"panel_id":"b","x":670,"y":12,"width":660,"height":531},{"panel_id":"a","x":1,"y":5,"width":645,"height":534}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/17/images/figure_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/17/images/figure_7.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":"bar chart","caption_source":"content.json:img_caption"},"width":1336,"height":1058,"image_format":"jpeg","image_sha256":"4cedaa42bed377210333893c2fe3d3c9065990c4f3e2e96838c91697bfe0717f","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_9_figure_11.jpg","caption":"Figure 11. Evolution of coverage rates  $\\mathrm{HfO_2}$  coverage) and crystallinity with respect to ALD cycles number. The coverage is cumulative. The legend is the same as in Figure 10.","id":"train/atomic-layer-deposition/simulation-usecase/9/figure_11","sample_id":"atomic-layer-deposition/simulation-usecase/9/figure_11","subset":"bar-chart","split":"train","classification":[{"panel_id":"a","label":"bar chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows how surface coverage and crystallinity of HfO₂ change with the number of ALD cycles. Coverage rises quickly in the first three cycles and then levels off near about 73%, showing that the surface becomes saturated early. Crystallinity also increases fast at first, reaching about 60% by the second cycle, and then slowly stabilizes around 62–65%. This means that surface filling happens faster than crystal ordering, which continues more gradually over later cycles.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| ALD cycle | Coverage (%) | Crystallinity (%) |\\n|-----------|--------------|-------------------|\\n| 1 | ~44 | ~20 |\\n| 2 | ~65 | ~60 |\\n| 3 | ~70 | ~64 |\\n| 4 | ~72 | ~65 |\\n| 5 | ~73 | ~64 |\\n| 6 | ~73 | ~63 |\\n| 7 | ~73 | ~62 |\\n| 8 | ~73 | ~62 |\\n| 9 | ~73 | ~62 |\\n| 10 | ~73 | ~62 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The rapid increase is driven by the high density of available surface hydroxyl sites during the initial ALD cycles, which promotes efficient precursor chemisorption. These reactive sites allow nearly all incoming precursor molecules to participate in self-limiting ligand-exchange reactions, rapidly increasing the surface-occupied fraction. Concurrently, the initial Hf–O–Si interfacial bonding promotes short-range structural ordering, enabling crystallinity to rise quickly once several monolayers have formed. As the surface becomes progressively saturated, the number of unreacted sites decreases, slowing additional growth and leading to the observed plateau.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"-Coverage increases very fast\\n-Crystallinity rises strongly\\n-Many surface sites become occupied\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"-It shows self-limiting growth\\n, -It improves thickness control\\n, -It helps achieve uniform films\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Cycle 2.\"}]}]","bbox":[{"panel_id":"a","x":2,"y":5,"width":668,"height":497}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/9/images/figure_11.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/9/images/figure_11.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/9/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/9/A. Dkhissi et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"9","first_classification_panel_id":"a","first_classification_label":"bar chart","caption_source":"content.json:img_caption"},"width":673,"height":503,"image_format":"jpeg","image_sha256":"e2c8fbaa74c04281cd1583063c4ec63d728f8ef945384adf47e2b3cae07821b3","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_9.jpg","caption":"Figure 9. Etch rate dependence on the phase of  $\\mathrm{MoS}_2$  material. Crystalline  $\\mathrm{MoS}_2$  etches at a much slower rate compared to amorphous films. Crystalline  $\\mathrm{MoS}_2$  films etch at the rate of  $\\sim 0.2\\mathrm{\\AA}/$  cycle, while amorphous films were observed to etch at the rate of 0.5 A/cycle.","id":"train/atomic-layer-etching/experimental-usecase/22/figure_9","sample_id":"atomic-layer-etching/experimental-usecase/22/figure_9","subset":"bar-chart","split":"train","classification":[{"panel_id":"a","label":"bar chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The bar chart compares the etch rates of materials in crystalline and amorphous phases. The etch rate for the amorphous phase is significantly higher (0.5 Å/cycle) than that of the crystalline phase (0.2 Å/cycle), indicating that amorphous materials are more susceptible to etching under the same process conditions\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Phase | Etch Rate (Å/cyc) |\\n|---|---|\\n| Crystalline | 0.2 |\\n| Amorphous | 0.5 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.2 Å/cycle\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Crystalline: 0.2 Å/cycle, Amorphous: 0.5 Å/cycle\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The chart shows that the amorphous phase has a higher etch rate than the crystalline phase, with values of 0.5 Å/cycle and 0.2 Å/cycle, respectively. This suggests that amorphous materials are more readily etched under identical processing conditions. The increased etchability may be due to the disordered atomic structure of amorphous phases, which typically lack the strong, directional bonding present in crystalline materials. This structural difference likely makes them more vulnerable to removal during etching cycles.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":427,"height":353}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/images/figure_9.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/images/figure_9.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":"bar chart","caption_source":"content.json:img_caption"},"width":427,"height":353,"image_format":"jpeg","image_sha256":"77d0b9e514c5c067223793b080dd7464fb5af6685bdc03e539d47bff9a898a4b","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_2.jpg","caption":"Fig. 2. Results of EPC synergy test at  $80\\mathrm{V}$  bias.","id":"train/atomic-layer-etching/experimental-usecase/30/fig_2","sample_id":"atomic-layer-etching/experimental-usecase/30/fig_2","subset":"bar-chart","split":"train","classification":[{"panel_id":"a","label":"bar chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The bar chart shows an EPC synergy test at 80 V bias by comparing EPC for a full ALE cycle, chlorination only, and Ar sputter only. EPC is 0.37 nm for ALE, about 0.00 nm for chlorination only, and 0.05 nm for Ar sputter only.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Category | EPC [nm] |\\n|----------|----------|\\n| ALE      | 0.37     |\\n| Chlorination only | 0.00     |\\n| Ar sputter only | 0.05     |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"They represent the measured etch per cycle when running the full ALE cycle, only the chlorination step, or only the Ar ion removal step, respectively.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The ALE EPC is 0.37 nm, which is much larger than either step alone. Chlorination only is negligible, and Ar sputter only is 0.05 nm. This means most of the etch occurs when chlorination and ion bombardment are combined, consistent with a synergistic reactive ion etching mechanism.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"About 10 to 15% of the ALE EPC.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Chlorination alone does not produce measurable net etching.\\n\\nAr ion bombardment alone removes only a small amount per cycle.\\n\\nThe combination enables efficient removal of the chlorinated surface layer, giving the large EPC seen for ALE.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":1,"width":595,"height":373}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/images/fig_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/images/fig_2.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":"bar chart","caption_source":"content.json:image_caption"},"width":600,"height":378,"image_format":"jpeg","image_sha256":"8507e5708ce7cb756a191282482b244543960ddbdff65f09fd791e3a89eb1065","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_33_fig_3.jpg","caption":"FIG. 3. Etching amount of TiN films. The sample temperature during radical exposure were  $-10$  or  $30^{\\circ}\\text{C}$ , and the radical exposure time was fixed at  $300 \\text{s}$ .","id":"train/atomic-layer-etching/experimental-usecase/33/fig_3","sample_id":"atomic-layer-etching/experimental-usecase/33/fig_3","subset":"bar-chart","split":"train","classification":[{"panel_id":"a","label":"bar chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The bar chart compares etching amounts at two different temperatures, showing a significant difference between -10°C and 30°C.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Temperature | Etching Amount (nm) |\\n|---|---|\\n|-10 °C | 0.3 nm |\\n| 30 °C | 20 nm |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It is likely that the temperature provides the energy to make the created species volatile.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The data represents continuous etching in the case of 20 degrees and possibly ALE for -10 degrees.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It is likely that there will even be more etching. The process looks like temperature dependent continuous etching so likely with temperature, the etching will increase.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"3 angstroms\"}]}]","bbox":[{"panel_id":"a","x":4,"y":2,"width":660,"height":469}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/33/images/fig_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/33/images/fig_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/33/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/33/Atomic layer etching of titanium nitride.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"33","first_classification_panel_id":"a","first_classification_label":"bar chart","caption_source":"content.json:image_caption"},"width":667,"height":472,"image_format":"jpeg","image_sha256":"f295b428fbcf6c02d3196ddd9a8db44096166630228f734fb88d8c479d67a62b","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_38_fig5.jpg","caption":"FIG.5. ALE synergy results for both a) AlGaN and (b) GaN.","id":"train/atomic-layer-etching/experimental-usecase/38/fig5","sample_id":"atomic-layer-etching/experimental-usecase/38/fig5","subset":"bar-chart","split":"train","classification":[{"panel_id":"a","label":"bar chart"},{"panel_id":"b","label":"bar chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The bar chart shows the etch rate per cycle for the individual half-reactions (Reaction A only and Reaction B only) compared with the full ALE cycle. Reaction A alone produces no measurable etch, Reaction B (AlGaN) alone causes only a small background etch (~0.07 Å/cycle), while the combined ALE process yields ~0.807 Å/cycle, giving a  synergy of ~91%.\"},{\"panel_id\":\"b\",\"text\":\"The bar chart shows the etch rate per cycle for the individual half-reactions (Reaction A only and Reaction B only) compared with the full ALE cycle. Reaction A alone produces no measurable etch, Reaction B (GaN) alone causes only a small background etch (~0.07 Å/cycle), while the combined ALE process yields ~2.639 Å/cycle, giving a  synergy of ~97%.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Etch rate (Å/cycle): Reaction A (240 cycles) | Etch rate (Å/cycle):ALE (100 Cycles) | Etch rate (Å/cycle):Reaction B (400 cycles) |\\n|---|---|---|\\n| 0.000 | 0.807 | 0.07 |\"},{\"panel_id\":\"b\",\"text\":\"| Etch rate (Å/cycle):Reaction A (240 cycles) | Etch rate (Å/cycle):ALE (72 Cycles) | Etch rate (Å/cycle):Reaction B (400 cycles) |\\n|---|---|---|\\n| 0.000 | 2.639 | 0.07 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Reaction B.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Synergy would decrease. By definition, synergy is calculated by subtracting the sum of the two single-step etch rates from the full ALE rate. Numerically, it would be: (2.639-0.14)/2.639 = 94.6%.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The GaN process is closer to ideal ALE, because its synergy is higher.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":2,"width":666,"height":530},{"panel_id":"b","x":3,"y":599,"width":667,"height":464}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/38/images/fig5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/38/images/fig5.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/38/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/38/High synergy atomic layer etching of AlGaNGaN.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"38","first_classification_panel_id":"a","first_classification_label":"bar chart","caption_source":"content.json:image_caption"},"width":675,"height":1072,"image_format":"jpeg","image_sha256":"4788da71c879c6fd78d9a5af749be0a8e27da4c642af6c5c0926301ab66b2fdb","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_3.jpg","caption":"FIG. 3. (a) Synergy test for the chosen ITO ALE parameters. For parameters, see Fig. 1. Synergy parameter, ALE EPC, and the individual reaction EPCs,  $\\alpha$ , for Reaction A and  $\\beta$  for Reaction B, are displayed in the graph. (b) ALE EPC (crosses) and etch depth (circles) as a function of the number of cycles. The lines are linear fits. The EPC line fit is nearly parallel. (c) Saturation curve for Reaction A. The  $\\mathsf{BCl}_3$  exposure time is varied. Other parameters are constant. (d) Saturation curve for Reaction B. The ALE EPC and Ar milling only EPC as a function of the Ar plasma exposure time are shown. Other parameters constant. Error bars in (b)–(d) show standard deviation from the mean, when multiple measurements are done with the same parameters.","id":"train/atomic-layer-etching/experimental-usecase/4/fig_3","sample_id":"atomic-layer-etching/experimental-usecase/4/fig_3","subset":"bar-chart","split":"train","classification":[{"panel_id":"a","label":"bar chart"},{"panel_id":"b","label":"multi-axis chart"},{"panel_id":"c","label":"line chart"},{"panel_id":"d","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"A bar chart comparing etch per cycle for different reactions.\"},{\"panel_id\":\"b\",\"text\":\"A multi-axis chart showing EPC and etch depth against the number of cycles.\"},{\"panel_id\":\"c\",\"text\":\"A line chart showing EPC change with BCl3 exposure time.\"},{\"panel_id\":\"d\",\"text\":\"A line chart showing EPC change with Ar plasma exposure time for two conditions.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Reaction | Etch per cycle (Å) |\\n|---|---|\\n| Reaction A | 0.15 |\\n| Reaction A+B | 1.10 |\\n| Reaction B | 0.20 |\"},{\"panel_id\":\"b\",\"text\":\"| Number of cycles | EPC (Å) | Etch depth (Å) |\\n|---|---|---|\\n| 50 | 0.90 | 50 |\\n| 100 | 1.10 | 100 |\\n| 150 | 1.30 | 200 |\\n| 200 | 1.00 | 210 |\"},{\"panel_id\":\"c\",\"text\":\"| BCl3 exposure (s) | EPC (Å) |\\n|---|---|\\n| 0 | 0.10 |\\n| 5 | 0.60 |\\n| 10 | 0.80 |\\n| 15 | 0.90 |\\n| 20 | 1.00 |\"},{\"panel_id\":\"d\",\"text\":\"| Ar plasma exposure (s) | EPC (Å) ALE | EPC (Å) Ar milling only |\\n|---|---|---|\\n| 0 | 0.10 | 0 |\\n| 5 | 0.30 |  |\\n| 10 | 0.90 | 0.10 |\\n| 15 | 1.10 |  |\\n| 20 | 1.90 | 0.25 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Synergy is a way to numerically represent how much the exposure of the individual precursors to the substrate etch the material compared to the combined use of the precursors. This is represented by the following formula: S = 1 - (α + β)/ALE x 100%, where α and β represent the EPC during exposure of the individual components and ALE the EPC during the dombined use. In this case, S = 1 - 0.2/1.1 = 82%.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The EPC increases slightly over the number of cycles, but the authors consider it \\\"nearly constant\\\", so this process meets the ALE requirements.\"}]},{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"About 1.2 Å.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"About 0.6 Å.\"}]}]","bbox":[{"panel_id":"a","x":14,"y":14,"width":694,"height":455},{"panel_id":"b","x":730,"y":34,"width":679,"height":428},{"panel_id":"c","x":7,"y":508,"width":689,"height":425},{"panel_id":"d","x":722,"y":512,"width":683,"height":421}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/4/images/fig_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/4/images/fig_3.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":"bar chart","caption_source":"content.json:img_caption"},"width":1414,"height":936,"image_format":"jpeg","image_sha256":"14a7f4c87cfc768b28879519eb34cdac26286196af1c7e171f0d6192b613ba30","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_15_fig_10.jpg","caption":"FIG. 10. Distribution of etch products from MD simulations during the first  $15\\%$  [panel (a)] and final  $85\\%$  [panel (b)] of the  $(70\\mathrm{eV})\\mathrm{Ar}^+$  ion bombardment step. The cycle in this case corresponds to  $2255\\mathrm{Cl}_2$  molecule impacts and  $1000\\mathrm{Ar}^+$  ion impacts. The distributions of products are taken once the cyclic steady state is reached.","id":"train/atomic-layer-etching/simulation-usecase/15/fig_10","sample_id":"atomic-layer-etching/simulation-usecase/15/fig_10","subset":"bar-chart","split":"train","classification":[{"panel_id":"a","label":"bar chart"},{"panel_id":"b","label":"bar chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The histogram shows that early in the Ar⁺ bombardment, etching is chemically driven, dominated by highly chlorinated species (mainly SiCl₂) from a saturated surface layer, with minimal pure Si. In the later phase, the chlorinated layer is depleted and the process shifts to physical sputtering, evidenced by a sharp increase in pure Si and leaner SiCl products, while a large fraction of atomic Cl indicates significant inefficiency due to sputtered etchant. Early phase: Etching is chemically driven, dominated by highly chlorinated species (mainly SiCl₂) from a chlorine-saturated surface layer, with negligible pure Si.\"},{\"panel_id\":\"b\",\"text\":\"Late phase: The chlorinated layer is depleted, and etching shifts to physical sputtering, marked by increased pure Si and leaner SiCl products, with significant atomic Cl indicating sputtered etchant inefficiency.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Etch product | First 15% of Ar⁺ step (%) |\\n|-------------|---------------------------:|\\n| Si          | 3                          |\\n| Si<sub>2</sub> | 1                       |\\n| SiCl        | 11                         |\\n| SiCl<sub>2</sub> | 24                  |\\n| SiCl<sub>4</sub> | 3                   |\\n| Si<sub>2</sub>Cl | 4                   |\\n| Si<sub>2</sub>Cl<sub>2</sub> | 2      |\\n| Cl          | 54                         |\\n| Cl<sub>2</sub> | 2                    |\"},{\"panel_id\":\"b\",\"text\":\"| Etch product | Last 85% of Ar⁺ step (%) |\\n|-------------|--------------------------:|\\n| Si          | 22                        |\\n| Si<sub>2</sub> | 2                       |\\n| SiCl        | 24                        |\\n| SiCl<sub>2</sub> | 15                  |\\n| SiCl<sub>4</sub> | 1                   |\\n| Si<sub>2</sub>Cl | 2                   |\\n| Si<sub>2</sub>Cl<sub>2</sub> | 2      |\\n| Cl          | 36                        |\\n| Cl<sub>2</sub> | 2                    |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Because it caused Over-etching. The high 'Si' signal proves the process continued to sputter bulk material after the chemically modified layer was gone, breaking the self-limiting control.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It signals the onset of Physical Sputtering. The ions have finished removing the chemically modified layer and are now damaging/removing the unmodified bulk silicon substrate.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"It reveals a Gradient. The surface is Chlorine-rich (SiCl₂), but the deeper sub-surface layers are Chlorine-poor, yielding mostly SiCl as the etch progresses downwards.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It implies Low Efficiency / Wasted Energy. Many ions are simply knocking off the attached Chlorine atoms (reversing the chemical dose) rather than removing the target Silicon atoms.\"}]}]","bbox":[{"panel_id":"b","x":689,"y":31,"width":669,"height":440},{"panel_id":"a","x":6,"y":25,"width":679,"height":445}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/15/images/fig_10.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/15/images/fig_10.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/15/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/15/Molecular dynamics study of silicon atomic layer etching by chlorine gas and argon ions.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"15","first_classification_panel_id":"a","first_classification_label":"bar chart","caption_source":"content.json:img_caption"},"width":1367,"height":475,"image_format":"jpeg","image_sha256":"61d8f5317c1b1f7eb276bc27f69c857cfa5dfe222be678063c557bef73db5eeb","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_15_fig_8.jpg","caption":"FIG. 8. Distribution of etch products from MD simulations during the (70 eV)  $\\mathrm{Ar}^+$  ion bombardment step. The cycle in this case corresponds to  $2255\\mathrm{Cl}_2$  molecule impacts and  $1000\\mathrm{Ar}^+$  ion impacts. The distribution of products is taken once the cyclic steady state is reached.","id":"train/atomic-layer-etching/simulation-usecase/15/fig_8","sample_id":"atomic-layer-etching/simulation-usecase/15/fig_8","subset":"bar-chart","split":"train","classification":[{"panel_id":"a","label":"bar chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The histogram shows that atomic chlorine (>40%) is the dominant ejected species, indicating significant sputtering of adsorbed chlorine without effective etching. Actual etching occurs via SiCl and SiCl₂ fragments (~35%), while the presence of pure Si (~16%) reveals unintended physical sputtering of the underlying silicon by 70 eV ions.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Etch product | Ar⁺ Bombardment step (%) |\\n|-------------|--------------------------:|\\n| Si          | ~18                      |\\n| Si<sub>2</sub> | ~2                     |\\n| SiCl        | ~22                       |\\n| SiCl<sub>2</sub> | ~18                |\\n| SiCl<sub>4</sub> | ~1                 |\\n| Si<sub>2</sub>Cl | ~2                 |\\n| Si<sub>2</sub>Cl<sub>2</sub> | ~1    |\\n| Cl          | ~42                       |\\n| Cl<sub>2</sub> | ~2                  |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It proves Physical Sputtering is occurring. The ions are energetic enough to remove the underlying bulk silicon, effectively breaking the 'self-limiting' rule of ideal ALE.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It represents Low Efficiency / Wasted Energy. Nearly half of the ion impacts are wasted on simply reversing the previous dosing step (knocking off Cl) rather than removing the target material.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It implies a kinetic/physical mechanism. The ions are violently knocking off incomplete fragments (SiCl) rather than allowing the chemical reaction to proceed fully to the volatile stable state (SiCl₄).\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It indicates Substrate Damage. Removing pure silicon means the process is uncontrollably eating into the functional crystal lattice, which would alter the electrical properties of the device.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":12,"width":675,"height":433}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/15/images/fig_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/15/images/fig_8.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/15/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/15/Molecular dynamics study of silicon atomic layer etching by chlorine gas and argon ions.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"15","first_classification_panel_id":"a","first_classification_label":"bar chart","caption_source":"content.json:img_caption"},"width":675,"height":445,"image_format":"jpeg","image_sha256":"959e8acda147ba82acdf359a9ae4ac9b19bdfb48055b8d61a83a7f0d13d45f3b","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"}