subsets/multiple-scatter-plot/test/metadata.jsonl

{"file_name":"images/test_atomic-layer-deposition_experimental-usecase_1_figure_12.jpg","caption":"Figure 12. Refractive index,  $n$ , of the  $\\mathrm{Al}_2\\mathrm{O}_3$  ALD films versus growth temperature from 33 to  $177^{\\circ}C$  and comparisons to results from other  $\\mathrm{Al}_2\\mathrm{O}_3$  ALD studies at higher temperatures.9,12-15,21","id":"test/atomic-layer-deposition/experimental-usecase/1/figure_12","sample_id":"atomic-layer-deposition/experimental-usecase/1/figure_12","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the relationship between growth temperature and refractive index, with data points from this work  for lower temperatures (below 200 °C) and other studies for higher temperatures (~150°C - 500°C). This study shows stronger positive correlation between temperature and refractive index.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Growth Temperature (°C) | Refractive Index, n | Study |\\n|---|---|---|\\n| 31 | 1.511 | This work |\\n| 56 | 1.532 | This work |\\n| 78 | 1.541 | This work |\\n| 101 | 1.570 | This work |\\n| 123 | 1.582 | This work |\\n| 148 | 1.611 | Other studies |\\n| 148 | 1.631 | Other studies |\\n| 175 | 1.602 | This work |\\n| 176 | 1.651 | Other studies |\\n| 198 | 1.641 | Other studies |\\n| 248 | 1.641 | Other studies |\\n| 299 | 1.671 | Other studies |\\n| 398 | 1.682 | Other studies |\\n| 497 | 1.672 | Other studies |\\n| 499 | 1.651 | Other studies |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"This study.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The maximum refractive index achieved in this work is approximately 1.60, occuring at 175°C.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Among the temperatures studied in this work, the highest growth temperature (~175 °C) appears to produce the optically densest films. This is supported by the maximum refractive index measured in this study. A higher refractive index indicates improved greater optical density.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":595,"height":403}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_12.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_12.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/Groner et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"1","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":595,"height":403,"image_format":"jpeg","image_sha256":"64063e72c567306ff9aebbf0d0986d83ce9743cf3d191eaca3e07600c7dfd96f","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/test_atomic-layer-deposition_experimental-usecase_1_figure_4.jpg","caption":"Figure 4.  $\\mathrm{Al}_2\\mathrm{O}_3$  mass gain per cycle versus reactant exposure time at  $58^{\\circ}C$ . Purge times for both TMA and water were 30 s each. The exposure time of the reactant not being investigated was held constant at  $1\\mathrm{s}$ .","id":"test/atomic-layer-deposition/experimental-usecase/1/figure_4","sample_id":"atomic-layer-deposition/experimental-usecase/1/figure_4","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the relationship between exposure time and mass gain per cycle for two different substances at 58°C. The cycle sequence for water was (1/30/x/30), fro TMA it was (y/30/1/30).\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Exposure Time (s) | Mass Gain per Cycle (ng/cm²) | Reagent (cycle conditions) |\\n|---|---|---|\\n| 0,00 | 0,69 | Water (1,30,x,30) |\\n| 0,18 | 22,38 | Water (1,30,x,30) |\\n| 0,47 | 26,08 | Water (1,30,x,30) |\\n| 0,96 | 28,73 | Water (1,30,x,30) |\\n| 1,97 | 31,15 | Water (1,30,x,30) |\\n| 3,97 | 32,88 | Water (1,30,x,30) |\\n| 7,98 | 35,15 | Water (1,30,x,30) |\\n| 0,00 | 0,16 | TMA (y,30,1,30) |\\n| 0,18 | 26,08 | TMA (y,30,1,30) |\\n| 0,47 | 27,07 | TMA (y,30,1,30) |\\n| 0,96 | 27,07 | TMA (y,30,1,30) |\\n| 1,97 | 27,82 | TMA (y,30,1,30) |\\n| 3,97 | 29,18 | TMA (y,30,1,30) |\\n| 7,98 | 30,92 | TMA (y,30,1,30) |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The graph shows soft saturation. Hard saturation would be a perfectly horizontal line. The small upward slope here indicates the reaction is either not perfectly complete or purge times are insufficient.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"TMA.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Water produces higher mass gain (data point at 1 s pulse time).\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"1 second.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":511,"height":400}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_4.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/Groner et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"1","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":511,"height":400,"image_format":"jpeg","image_sha256":"94bf66aac13a14f0e6ef8eadcaa657abf6c1c7c5abbd6f6f8a35f64ad70f93ff","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/test_atomic-layer-deposition_experimental-usecase_1_figure_5.jpg","caption":"Figure 5.  $\\mathrm{Al}_2\\mathrm{O}_3$  mass gain per cycle versus purge time at 58  $^\\circ \\mathrm{C}$ . Exposure times for both TMA and water were 1 s each. The purge time of the reactant not being investigated was held constant at  $30~\\mathrm{s}$ .","id":"test/atomic-layer-deposition/experimental-usecase/1/figure_5","sample_id":"atomic-layer-deposition/experimental-usecase/1/figure_5","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the relationship between purge time (in seconds) and mass gain per cycle (in ng/cm²) at 58°C. The data points represent measurements for water (1,30,1,x) and TMA (1,y,1,30).\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Purge Time (s) | Mass Gain per Cycle (ng/cm²) | Reagent (cycle conditions) |\\n|---|---|---|\\n| 2,02 | 31,20 | Water (1,30,1,x) |\\n| 4,78 | 28,65 | Water (1,30,1,x) |\\n| 9,90 | 27,82 | Water (1,30,1,x) |\\n| 19,93 | 27,35 | Water (1,30,1,x) |\\n| 29,97 | 27,07 | Water (1,30,1,x) |\\n| 40,07 | 27,02 | Water (1,30,1,x) |\\n| 0,27 | 29,01 | TMA (1,y,1,30) |\\n| 1,95 | 27,35 | TMA (1,y,1,30) |\\n| 9,90 | 27,35 | TMA (1,y,1,30) |\\n| 20,00 | 27,52 | TMA (1,y,1,30) |\\n| 29,97 | 27,79 | TMA (1,y,1,30) |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At short purge times (approx. below 10 s).\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The evidence of CVD growth is the higher mass gain per cycle observed at short purge times. In ideal ALD, mass gain should be independent of purge time once surface reactions are saturated. The decrease in mass gain with increasing purge time indicates gas-phase reactions at short purges, which are characteristic of CVD.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Approximately 10 seconds.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"No, 10 seconds is likely not sufficient for purging after reaction with water. The plot shows a higher mass gain per cycle at around 10 seconds, indicating residual water remains and contributes to CVD‑like growth. A longer purge time is needed for the mass gain to reach a stable, ALD‑like plateau.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":528,"height":394}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_5.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/Groner et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"1","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":528,"height":394,"image_format":"jpeg","image_sha256":"dcfb4490f903d9bb591161b0b611fd5aa61d932ded2399673594000184bfd8c6","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/test_atomic-layer-deposition_experimental-usecase_1_figure_6.jpg","caption":"Figure 6. Ellipsometry data for  $\\mathrm{Al}_2\\mathrm{O}_3$  ALD films grown using 300 reaction cycles on Si(100) substrates. The refractive index, growth rate, and density (calculated using ellipsometry and QCM data) are plotted versus growth temperatures from 33 to  $177^{\\circ}C$ .","id":"test/atomic-layer-deposition/experimental-usecase/1/figure_6","sample_id":"atomic-layer-deposition/experimental-usecase/1/figure_6","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows how growth temperature affects the density, refractive index, and growth rate of the film. As temperature increases, density and refractive index generally rise, while the growth rate exhibits a peak at intermediate temperatures.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Growth Temperature (°C) | Density (g/cm³) | Refractive Index n | Growth Rate (Å/cycle) |\\n|---|---|---|---|\\n| 32,73 | 2,48 | 1,52 | 1,11 |\\n| 57,47 | 2,50 | 1,53 | 1,20 |\\n| 79,64 | 2,65 | 1,55 | 1,27 |\\n| 101,55 | 2,68 | 1,58 | 1,34 |\\n| 124,23 | 2,95 | 1,60 | 1,34 |\\n| 176,55 | 3,01 | 1,61 | 1,26 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"They are possitively correlated (the refractive index increases with increasing density).\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Yes, there is evidence of an ALD temperature window. The growth rate remains relatively constant over an intermediate temperature. This  indicates self‑limiting ALD behavior without strong temperature‑driven variations.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Higher temperatures lead to more dense films.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At 100°C - 125 °C.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":586,"height":367}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_6.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/Groner et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"1","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":586,"height":367,"image_format":"jpeg","image_sha256":"526dfb04b4e16db8234f1099a12146840fdf4bd4b058a27d9037a3204ad5fe88","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/test_atomic-layer-deposition_experimental-usecase_1_figure_9.jpg","caption":"Figure 9. Density of  $\\mathrm{Al}_2\\mathrm{O}_3$  ALD films versus growth temperature from 33 to  $177^{\\circ}\\mathrm{C}$  for the various methods.","id":"test/atomic-layer-deposition/experimental-usecase/1/figure_9","sample_id":"atomic-layer-deposition/experimental-usecase/1/figure_9","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows that the density of Al₂O₃ ALD films increases with growth temperature from 33 to 177 °C. Multiple measurement techniques (ellipsometry + QCM, profilometry + QCM, AFM + QCM, XRR, RBS/FRes + ellipsometry, and the Lorentz–Lorenz method) yield consistent upward trends with some variations between methods.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Growth Temperature (°C) | Density (g/cm³) | Measurement |\\n|---|---|---|\\n| 32.42 | 2.49 | Ellipsometry + QCM |\\n| 57.52 | 2.50 | Ellipsometry + QCM |\\n| 79.48 | 2.65 | Ellipsometry + QCM |\\n| 101.70 | 2.68 | Ellipsometry + QCM |\\n| 124.97 | 2.94 | Ellipsometry + QCM |\\n| 176.47 | 2.99 | Ellipsometry + QCM |\\n| 32.42 | 2.53 | Profilometry + QCM |\\n| 57.25 | 2.67 | Profilometry + QCM |\\n| 79.22 | 2.59 | Profilometry + QCM |\\n| 101.44 | 2.69 | Profilometry + QCM |\\n| 125.49 | 3.08 | Profilometry + QCM |\\n| 176.73 | 3.01 | Profilometry + QCM |\\n| 32.16 | 2.43 | AFM + QCM |\\n| 57.52 | 2.75 | AFM + QCM |\\n| 79.48 | 2.61 | AFM + QCM |\\n| 101.44 | 2.49 | AFM + QCM |\\n| 124.97 | 2.83 | AFM + QCM |\\n| 176.99 | 2.99 | AFM + QCM |\\n| 32.16 | 2.42 | XRR |\\n| 56.99 | 2.58 | XRR |\\n| 79.48 | 2.70 | XRR |\\n| 101.44 | 2.66 | XRR |\\n| 124.44 | 2.74 | XRR |\\n| 177.25 | 3.07 | XRR |\\n| 32.42 | 2.69 | RBS/FReS+Ellipsomet |\\n| 57.25 | 2.66 | RBS/FReS+Ellipsomet |\\n| 79.22 | 2.74 | RBS/FReS+Ellipsomet |\\n| 101.44 | 2.86 | RBS/FReS+Ellipsomet |\\n| 124.71 | 2.90 | RBS/FReS+Ellipsomet |\\n| 177.25 | 3.12 | RBS/FReS+Ellipsomet |\\n| 32.16 | 2.63 | Lorentz-Lorentz |\\n| 57.52 | 2.71 | Lorentz-Lorentz |\\n| 79.22 | 2.75 | Lorentz-Lorentz |\\n| 101.70 | 2.88 | Lorentz-Lorentz |\\n| 124.71 | 2.92 | Lorentz-Lorentz |\\n| 176.73 | 2.96 | Lorentz-Lorentz |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The ellipsometry + QCM, profilometry + QCM, and XRR techniques appear to agree most closely with each other.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"They show a positive correlation.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"RBS/FRes + ellipsometryand the Lorentz–Lorenz method.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The variability between measurements is larger at 100°C .\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":564,"height":408}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_9.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_9.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/Groner et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"1","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":564,"height":408,"image_format":"jpeg","image_sha256":"3a460c225e510b42586e062fb3b66d77891659d5d66bf521b63cffe2f8df3ea5","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/test_atomic-layer-deposition_experimental-usecase_10_fig_12.jpg","caption":"FIG. 12. (Color online) Thickness evolution of  $\\mathrm{Pt}$  and  $\\mathrm{PtO_2}$  films deposited on an  $\\mathrm{Al}_2\\mathrm{O}_3$  substrate (Ref. 164). The precursor was  $\\mathrm{Pt(Cp^{Me})Me_3}$  and  $\\mathrm{O}_2$  gas or an  $\\mathrm{O}_2$  plasma were used as the oxidants. After 150 cycles, the plasma-assisted ALD process was stopped and film growth was continued using thermal ALD. The plasma exposure time for was  $0.5\\mathrm{s}$  for  $\\mathrm{Pt}$  and  $5\\mathrm{s}$  for  $\\mathrm{PtO_2}$ . From H.C.M. Knoops et al., Electrochem. Solid-State Lett. 12, G35 (2009). Reproduced with permission of ECS-The Electrochemical Society.","id":"test/atomic-layer-deposition/experimental-usecase/10/fig_12","sample_id":"atomic-layer-deposition/experimental-usecase/10/fig_12","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows the relationship between film thickness and the number of ALD cycles for different deposition methods. Both Plasma ALD Pt and Plasma ALD PtO₂ display nearly linear growth with increasing cycles, while Thermal ALD Pt shows negligible thickness change, indicating limited film deposition. Plasma-enhanced processes thus exhibit higher growth rates and efficiency\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of ALD Cycles | Plasma ALD Pt (nm) | Thermal ALD Pt (nm) | Plasma ALD PtO₂ (nm) |\\n|----------------------|--------------------|---------------------|----------------------|\\n| 0                    | 0                  | 0                   | 0                    |\\n| 50                   | 2                  | 0                   | 1.8                  |\\n| 100                  | 4                  | 0                   | 3.5                  |\\n| 150                  | 6                  | 0                   | 5.5                  |\\n| 200                  | 8                  | 0                   | 7.5                  |\\n| 250                  | 10                 | 0                   | 9.5                  |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Plasma ALD Pt and Plasma ALD PtO₂ show steady linear increases in thickness with the number of cycles, while Thermal ALD Pt shows minimal or no growth, indicating poor deposition efficiency under the given conditions.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Film thickness (nm).\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Plasma ALD Pt, Thermal ALD Pt, Plasma ALD PtO₂\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":683,"height":533}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/10/images/fig_12.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/10/images/fig_12.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/10/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/10/H. B. Profijt et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"10","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":683,"height":533,"image_format":"jpeg","image_sha256":"f8b2a824400e957968673796907766e6172ec2443dcd3bea9c793a8d9bcbf3db","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/test_atomic-layer-deposition_experimental-usecase_15_fig_4.jpg","caption":"FIG. 4. Loss tangent of BST film grown on  $\\mathrm{MgO}$ ,  $\\mathrm{TiO_2 / Si}$ , and  $\\mathrm{Ta}_2\\mathrm{O}_5 / \\mathrm{Si}$  substrates.","id":"test/atomic-layer-deposition/experimental-usecase/15/fig_4","sample_id":"atomic-layer-deposition/experimental-usecase/15/fig_4","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Chart plots loss tangent against applied voltage for BST/MgO, BST/TiO2/Si and BST/Ta2O5/Si samples.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Applied Voltage (V) | Loss Tangent  of BST/MgO | Loss Tangent  of BST/TiO2/Si | Loss Tangent  of BST/Ta2O5/Si |\\n|-------------------|--------------|-----------|----------|\\n| 0                 | 0.0       | 0.05            | 0.12          |\\n| 2                 | 0.0       | 0.1             | 0.18          |\\n| 4                 | 0.0       | 0.18            | 0.22          |\\n| 6                 | 0.0       | 0.2             | 0.38          |\\n| 8                 | 0.0       | 0.3             | 0.45          |\\n| 10                | 0.0       | 0.35            | 0.6           |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The loss tangent increases with increase in applied voltage.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"After an applied voltage of 2 V, increasing the voltage the loss tangent of  BST/Ta2O5/Si is double that of BST/TiO2/Si\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"BST films grown on MgO showed  the lowest loss tangent\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The loss tangent of BST grown on TiO2 /Si and Ta2O5 /Si is high, presumably due to the lossy Si substrate. The figure of merit (FOM), ratio of\\ntunability to dielectric loss, is very important for microwave tunable device applications. At an applied voltage of 5 V, the\\nFOM of BST/TiO2 and BST/Ta2O5 were 5.22 and 1.57, respectively. The BST films grown on TiO2 and Ta2O5 buffered\\nSi substrates showed relatively lower FOM values as compared to the FOM (11.87) of BST/MgO due to the lossy Si substrate in the former case\"}]}]","bbox":[{"panel_id":"a","x":1,"y":0,"width":533,"height":531}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/15/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/15/images/fig_4.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/15/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/15/Il-Doo Kim et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"15","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":539,"height":531,"image_format":"jpeg","image_sha256":"b23c14ec128765c1e3fe1a27d88a3082c16f1241c53a28e568071ccafae437fb","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/test_atomic-layer-deposition_experimental-usecase_21_fig4.jpg","caption":"Fig.4 Phase shift of the incident light  $(\\Delta)$  from in situ SE measurements of the reaction of  $\\mathrm{CO_2}$  and  $\\mathsf{H}_2\\mathsf{O}$  with  $\\sim 50$  nm plasma ALD films grown at 250, 275 and  $300^{\\circ}C$  First,  $\\mathrm{CO_2}$  was dosed into the reaction chamber 20 times with pulses of 2 s, next  $\\mathsf{H}_2\\mathsf{O}$  was pulsed 20 times for  $50~\\mathrm{ms}$  and afterwards  $\\mathrm{CO_2}$  was again pulsed 20 times. SE data was recorded after every pulse and  $\\Delta$  is plotted at  $589~\\mathrm{nm}$","id":"test/atomic-layer-deposition/experimental-usecase/21/fig4","sample_id":"atomic-layer-deposition/experimental-usecase/21/fig4","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Phase shift of the incident light (D) from in situ SE measurements of the reaction of CO2 and H2O with ~50 nm plasma ALD films grown at 250, 275 and 300 °C\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Measurement Number | Δ (°) at 250 °C | Δ (°) at 275 °C | Δ (°) at 300 °C |\\n|---------------------|-----------------|-----------------|-----------------|\\n| 0                  | 0               | 0               | 0               |\\n| 10                 | 1               | 5               | 2               |\\n| 20                 | 1               | 10              | 3               |\\n| 30                 | 1               | 13              | 6               |\\n| 40                 | 1               | 15              | 8               |\\n| 50                 | 1               | 18              | 10              |\\n| 60                 | 1               | 20              | 12              |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"First, CO2 was dosed into the reaction chamber 20 times with pulses of 2 s, next H2O was pulsed 20 times for 50 ms and afterwards CO2 was again pulsed 20 times\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"1.  LiOH/Li2O at 275 °C\\n, 2. Li2O at 300 °C\\n, 3. Li2CO3 at 250 °C\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"For plasma depositions at 250 °C, delta was unaffected by CO2 and H2O exposure, pointing out the stability of the film. However, the film deposited at a slightly higher temperature of 275 °C  strongly reacted with CO2, whereas for the film deposited at 300 °C, delta only slightly changes during the frst pulse. Both LiOH and Li2O could react with CO2 to form Li2CO3. the 300 °C film consisting of Li2O is reacting with the H2O forming LiOH. Next, CO2 is pulsed again, and the layer is again reacting to form Li2CO3. Unlike the film deposited at 300 °C, the as-deposited 275 °C film was very reactive towards the first CO2 pulses and therefore (partly) consisted of LiOH. For all other fims, which include plasma ALD films deposited at 50–250 °C and thermal ALD films deposited at 50–300 °C, the change in Delta was negligibly small, indicating that almost no reaction with H2O or CO2 occurred.\"}]}]","bbox":[{"panel_id":"a","x":2,"y":3,"width":552,"height":526}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/21/images/fig4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/21/images/fig4.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/21/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/21/N. Hornsveld et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"21","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":558,"height":531,"image_format":"jpeg","image_sha256":"7f778537308962d6097e460f66c2be82c4573463d58b8c38750637ed2c2fe640","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/test_atomic-layer-deposition_experimental-usecase_21_fig_10.jpg","caption":"Fig. 10 (a) Impedance spectroscopy of  $50 \\mathrm{nm} \\mathrm{Li_2CO_3}$  films prepared by thermal (blue) and plasma-assisted (red) ALD at  $150^{\\circ}\\mathrm{C}$  and (b) equivalent circuit models used to fit the data.","id":"test/atomic-layer-deposition/experimental-usecase/21/fig_10","sample_id":"atomic-layer-deposition/experimental-usecase/21/fig_10","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"},{"panel_id":"b","label":"apparatus diagram"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows a impedance spectroscopy of 50 nm Li2CO3 films prepared by thermal and plasma-assisted ALD.\"},{\"panel_id\":\"b\",\"text\":\"The figure depicts a equivalent circuits used to fit data with resistors (R_s, R_1) and capacitors (CPE_1, CPE_war).\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Condition | Z_Re (kΩ) Range | Z_Im (kΩ) Peak |\\n|-----------|------------------|---------------|\\n| Plasma    | 0 – 15          | ~5            |\\n| Thermal   | 0 – 30          | ~15           |\"},{\"panel_id\":\"b\",\"text\":\"| Model | Description                                      |\\n|-------|--------------------------------------------------|\\n| I     | Rs – (R1 II CPE1) – CPEwar                      |\\n| II    | Rs – (R1 II CPE1) – (R2 II CPE2) – CPEwar       |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The Li-ion conductivity of plasma and thermal ALD Li2CO3 layers was investigated using impedance spectroscopy. The measurements were performed at open circuit potential (OCP) on 50 nm Li2CO3 flms deposited at 150 °C on a TiN current collector.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Both the measured  thermal and plasma ALD layers show a behavior characteristic of a solid electrolyte: an intercept with the x-axis at high frequency,\\na semi-circle at medium frequency and a 45° inclined response at low frequency. Such behavior is typically associated with the ionic conductivity through a solid material. The high frequency intercept with the x-axis is generally attributed to the resistance of the cables, contacts and liquid electrolyte, the\\nsemi-circle originates from the ionic conductivity through the solid material and its capacitive response.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"When equivalent circuit model I was tried to fit the response of the plasma deposited Li2CO3 layer, no adequate fit could be obtained. For this reason, model II was used. Thus thermal ALD process uses model I and plasma ALD process uses model II to fit the data.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"1. For model I used to fit thermal ALD, a low chi2 value of 5 *10-4 is obtained.\\n2. For model II fitting plasma ALD, a low chi2 value of 2 *10-4 is obtained.\"}]}]","bbox":[{"panel_id":"a","x":4,"y":5,"width":483,"height":518},{"panel_id":"b","x":6,"y":532,"width":476,"height":387}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/21/images/fig_10.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/21/images/fig_10.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/21/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/21/N. Hornsveld et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"21","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":486,"height":919,"image_format":"jpeg","image_sha256":"93cdbe96210709458acf437a4420463fd75f0780d1c79728735d7cdc365e42c4","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/test_atomic-layer-deposition_experimental-usecase_21_fig_5.jpg","caption":"Fig. 5 In situ SE thickness measurement for plasma-assisted ALD of  $\\mathrm{Li}_{2}\\mathrm{CO}_{3}$  at  $300^{\\circ}\\mathrm{C}$  using different  $\\mathrm{O}_{2}$  plasma exposure times.","id":"test/atomic-layer-deposition/experimental-usecase/21/fig_5","sample_id":"atomic-layer-deposition/experimental-usecase/21/fig_5","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows thickness versus number of cycles  of plasma assisted ALD of Li2CO3 for three different plasma exposure times.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of Cycles | Thickness (nm) - 3 s plasma | Thickness (nm) - 5 s plasma | Thickness (nm) - 20 s plasma |\\n|-------------------|-----------------------------|-----------------------------|------------------------------|\\n| 0                | 0                           | 0                           | 0                            |\\n| 50               | 5                           | 8                           | 10                           |\\n| 100              | 10                          | 15                          | 18                           |\\n| 150              | 15                          | 22                          | 26                           |\\n| 200              | 20                          | 28                          | 32                           |\\n| 250              | 25                          | 33                          | 38                           |\\n| 300              | 30                          | 38                          | 44                           |\\n| 350              | 35                          | 43                          | 48                           |\\n| 400              | 40                          | 47                          | 50                           |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"From the thickness evolution with number of ALD cycles for various plasma exposure times, it can be observed, the bulk growth per cycle for different plasma exposure times is similar\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"For longer plasma exposure times, the growth per cycle requires a lower number of ALD cycles to become linear, compared to shorter plasma exposure times. The formation of LiOH/Li2O can thus be further stimulated by prolonging the plasma exposure time.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"300 °C\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":9,"y":4,"width":558,"height":445}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/21/images/fig_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/21/images/fig_5.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/21/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/21/N. Hornsveld et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"21","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":566,"height":447,"image_format":"jpeg","image_sha256":"00da872fadf822eaeda8b2bb0929b141ac0141118abeb6003c46ba9a46865a0a","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/test_atomic-layer-deposition_experimental-usecase_23_figure_6.jpg","caption":"Figure 6.  $\\mathrm{Ga}_2\\mathrm{O}_3$  growth rate at 250 and  $350^{\\circ}\\mathrm{C}$  as a function of (a) TMGa and (b)  $\\mathrm{O}_3$  exposures.","id":"test/atomic-layer-deposition/experimental-usecase/23/figure_6","sample_id":"atomic-layer-deposition/experimental-usecase/23/figure_6","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"},{"panel_id":"b","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the growth rate of a material as a function of TMGa exposure time at two different temperatures, 250°C and 350°C.\"},{\"panel_id\":\"b\",\"text\":\"The scatter plot illustrates the growth rate of a material as a function of ozone exposure time at two different temperatures, 250°C and 350°C.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| TMGa exposure (sec) | Growth rate (Å/cycle) 250 °C | Growth rate (Å/cycle) 350 °C |\\n|---|---|---|\\n| 0.5 | 0.4 | 0.5 |\\n| 1.0 | 0.4 | 0.55 |\\n| 2.0 | 0.4 | 0.55 |\"},{\"panel_id\":\"b\",\"text\":\"| Ozone exposure (sec) | Growth rate (Å/cycle) 250 °C | Growth rate (Å/cycle) 350 °C |\\n|---|---|---|\\n| 1 | 0.15 | 0.5 |\\n| 3 | 0.3 | 0.5 |\\n| 5 |  | 0.5 |\\n| 6 | 0.4 | 0.55 |\\n| 12 | 0.5 |  |\\n| 18 | 0.5 |  |\\n| 24 | 0.55 |  |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.5 seconds.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Roughly 0.4 and 0.5 Å/cycle, respectively.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The saturation of Ga₂O₃ growth with respect to ozone exposure occurs much faster at 350 °C than at 250 °C. At 350 °C, a saturated growth rate of 0.52 Å/cycle is reached with ozone exposures as short as 3 seconds, whereas at 250 °C, exposure times of at least 12 seconds are required to reach the same growth rate. This temperature dependence indicates that the reaction between ozone and the TMGa-terminated surface is the rate-limiting step in the ALD process, as higher temperatures facilitate more rapid completion of this surface reaction.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At least 12 seconds.\"}]}]","bbox":[{"panel_id":"a","x":5,"y":2,"width":471,"height":370},{"panel_id":"b","x":4,"y":365,"width":471,"height":363}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/23/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/23/images/figure_6.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/23/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/23/David J. Comstock et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"23","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":478,"height":731,"image_format":"jpeg","image_sha256":"162527fabdafbf4b36007a44c50e36e6d47cacfed600dbf597e69220b96d3966","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/test_atomic-layer-deposition_experimental-usecase_23_figure_7.jpg","caption":"Figure 7.  $\\mathrm{Ga}_{2}\\mathrm{O}_{3}$  growth rate as a function of the deposition temperature for fixed TMGa exposures of  $0.5\\mathrm{~s~}$  and variable ozone exposures.","id":"test/atomic-layer-deposition/experimental-usecase/23/figure_7","sample_id":"atomic-layer-deposition/experimental-usecase/23/figure_7","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the relationship between temperature and growth rate for ozone exposure times of 6 seconds, 24 seconds, and 72 seconds.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Temperature (°C) | Growth rate (Å/cycle) |\\n|---|---|\\n| 200 | 0.2 |\\n| 225 | 0.3 |\\n| 250 | 0.4 |\\n| 275 | 0.5 |\\n| 300 | 0.5 |\\n| 325 | 0.5 |\\n| 350 | 0.5 |\\n| 375 | 0.5 |\\n| 400 | 0.5 |\\n| 450 | 0.4 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.2 Å/cycle.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.4 Å/cycle.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.5 Å/cycle.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"First, the growth rate increases as a function of temperature. It plateaus at 0.5 Å/cycle at 275 °C. Above 400 °C, the growth rate starts to decrease. This means that the temperature window for this ALD process is between 275 and 400 °C.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":475,"height":475}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/23/images/figure_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/23/images/figure_7.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/23/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/23/David J. Comstock et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"23","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":475,"height":475,"image_format":"jpeg","image_sha256":"c139831dc1d020091cd6a5e805eef6dc4a65809b997daa91da186f880ec85276","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/test_atomic-layer-deposition_experimental-usecase_3_figure_5.jpg","caption":"Figure 5. AFM images of a smooth film (top) and a rough film (bottom). The area is  $1\\mu \\mathrm{m}\\times 1\\mu \\mathrm{m}$ . The height scale is from 0 to  $20 \\mathrm{nm}$ . The total film thickness is  $50 \\mathrm{nm}$ . Figure 6. Film roughness of hafnium oxide thin films as a function of deposition temperature and thickness.","id":"test/atomic-layer-deposition/experimental-usecase/3/figure_5","sample_id":"atomic-layer-deposition/experimental-usecase/3/figure_5","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the relationship between deposition temperature and film thickness, with white points indicating films with RMS roughness <1% and black points indicating films with RMS roughness >5%.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Deposition Temperature (°C) | Film Thickness (nm) | RMS Roughness |\\n|---|---|---|\\n| 50  | 120,4 | <1% |\\n| 100 | 70,4  | <1%|\\n| 125 | 50,3  | <1% |\\n| 150 | 34,9  | <1% |\\n| 150 | 80,3  | >5%  |\\n| 150 | 100,8 | >5% |\\n| 150 | 120,8 | >5% |\\n| 200 | 24,5  | <1%  |\\n| 200 | 50,1  | >5% |\\n| 200 | 120,8 | >5% |\\n| 250 | 9,7   | <1%  |\\n| 250 | 40,1  | >5% |\\n| 300 | 10,1  | <1%  |\\n| 300 | 35,1  | >5% |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"It represents a specific thickness, where the film switches from being smooth to rough, for a given temperature.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The downside is that it is impossible to produce smooth, thick films at these temperatures.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Approximately 100 °C or lower.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At approximately 100 °C .\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":544,"height":384}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/images/figure_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/images/figure_5.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/Hausmann et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"3","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":544,"height":384,"image_format":"jpeg","image_sha256":"3c56dab54e8374cc25d651912d3eb5503a946f066b40ee9662b6cf17482d3ca4","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/test_atomic-layer-deposition_experimental-usecase_3_figure_7.jpg","caption":"Figure 7. Thickness per cycle as a function of nitrogen purging for the reaction of tetrakis(dimethylamido)zirconium with water at a deposition temperature of  $100^{\\circ}\\mathrm{C}$ .","id":"test/atomic-layer-deposition/experimental-usecase/3/figure_7","sample_id":"atomic-layer-deposition/experimental-usecase/3/figure_7","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The  chart illustrates the relationship between deposition temperature and minimum purge time for two reagents: metal amide and water. The required purge times generally decrease with increasing temperature. Water generally requires longer purging times than metal amide.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Deposition Temperature (°C) | Water: Minimum Purge Time (sec) | Metal amide: Minimum Purge Time (sec) |\\n|---|---|---|\\n| 50 | 600 | 300 |\\n| 100 | 300 | 122 |\\n| 150 | 93 | 31 |\\n| 200 | 15 | 7 |\\n| 250 | 7 | 7 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Water.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The curves converge at higher temperatures. This trend occurs because increasing thermal energy significantly accelerates the desorption rate of physisorbed molecules. Consequently, at high temperatures, the difference in physisorption strength between water and the metal amide becomes negligible, and both can be purged almost instantly.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Water is physisorbed more strongly (requires longer purging).\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"200 °C (or 200 °C - 250 °C).\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":542,"height":441}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/images/figure_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/images/figure_7.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/Hausmann et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"3","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":542,"height":441,"image_format":"jpeg","image_sha256":"bb1f3ed9bbe532dd6856cf900be227b27e457757475b9e51fac3a1b4e0514be1","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/test_atomic-layer-deposition_experimental-usecase_39_fig_10.jpg","caption":"FIG. 10. (Color online) Integrated intensity of the time-resolved OES measurements of  $\\mathrm{CO^{*}}$ $\\mathrm{H^{*}}$  and  $\\mathrm{O^{*}}$  during the plasma step of the ALD cycle as a function of the number of successive  $\\mathrm{Al(CH_3)_3}$  exposures used in the cycle.","id":"test/atomic-layer-deposition/experimental-usecase/39/fig_10","sample_id":"atomic-layer-deposition/experimental-usecase/39/fig_10","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows the integrated intensity of OES signals at different wavelengths (777 nm - O*, 656 nm - H*, 519 nm - CO*) as a function of the number of Al(CH₃)₃ exposures per cycle.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of Al(CH3)3 exposures per cycle | Intensity (a.u.) 777 nm – O* | Intensity (a.u.) 656 nm – H* | Intensity (a.u.) 519 nm – CO* |\\n|---|---|---|---|\\n| 0  | 55 | 0  | 0  |\\n| 1  | 50 | 22 | 35 |\\n| 2  | 47 | 42 | 58 |\\n| 3  | 42 | 50 | 67 |\\n| 4  | 41 | 57 | 73 |\\n| 6  | 41 | 62 | 79 |\\n| 10 | 40 | 63 | 81 |\\n| 15 | 39 | 66 | 84 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"40.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"80.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"60.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The H* and CO* signals increase rapidly with increasing precursor dose at low numbers of exposures, reflecting increasing formation of reaction products, and both reach a plateau at approximately six exposures per cycle, indicating saturation of the Al(CH₃)₃ surface reactions. In contrast, the O* signal decreases with increasing precursor dose in the non-saturated regime and levels off at higher doses, consistent with increased oxygen consumption as more surface sites are occupied.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":672,"height":525}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/39/images/fig_10.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/39/images/fig_10.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/39/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/39/Heil et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"39","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":672,"height":525,"image_format":"jpeg","image_sha256":"b53a858cf471defa1aab46af5684676866cf05fea233ad718c35565c7eace0cf","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/test_atomic-layer-deposition_experimental-usecase_51_fig_4.jpg","caption":"Fig. 4  $\\mathrm{MoO}_x$  thickness measured by spectral ellipsometry data plotted versus the number of ALD cycles","id":"test/atomic-layer-deposition/experimental-usecase/51/fig_4","sample_id":"atomic-layer-deposition/experimental-usecase/51/fig_4","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart shows the relationship between deposition cycles and thickness for two different durations of oxygen exposure.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| deposition cycles | thickness (nm) 1 second O2 | thickness (nm) 2 second O2 |\\n|---|---|---|\\n| 0 | 0 | 0 |\\n| 10 | 1 | 1 |\\n| 20 | 1.5 | 1.5 |\\n| 30 | 2 | 2 |\\n| 40 | 3 | 3 |\\n| 50 | 4 | 4 |\\n| 60 | 4.5 | 4.5 |\\n| 70 | 5.5 | 5.5 |\\n| 80 | 5 | 5.5 |\\n| 90 | 6 | 6 |\\n| 100 | 6.5 | 7 |\\n| 110 | 7.5 | 7.5 |\\n| 120 | 8 | 8 |\\n| 130 | 9 | 9 |\\n| 140 | 10 | 10 |\\n| 150 | 10 | 10 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The effect of 1 s (green) oxygen plasma exposure vs. 2 s (blue) exposure.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"150.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"About 10 nm for both samples.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The only difference between the two dataset is the duration of the oxygen plasma exposure, which was either 1 or 2 seconds. After 150 cycles, both samples have reached a thickness of about 10 nm. The trendlines are almost identical and seemingly run parallel, indicating that there is no significant effect on the deposition rate when going from 1 to 2 seconds of oxygen plasma exposure time. This tells us that 1 second is sufficient.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":645,"height":536}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/51/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/51/images/fig_4.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/51/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/51/Ziegler et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"51","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":645,"height":536,"image_format":"jpeg","image_sha256":"06cfd5eeb1dd35f2d32002d3786af7f19749a1bab2a0669599483d44dad43625","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/test_atomic-layer-deposition_experimental-usecase_51_fig_8.jpg","caption":"Fig. 8 Solar cell results plotted versus  $\\mathrm{MoO}_x$  processes for 10- and 20-nm-thick oxide layers","id":"test/atomic-layer-deposition/experimental-usecase/51/fig_8","sample_id":"atomic-layer-deposition/experimental-usecase/51/fig_8","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"},{"panel_id":"b","label":"multiple scatter plot"},{"panel_id":"c","label":"multiple scatter plot"},{"panel_id":"d","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Scatter plot for the current density (J_SC, in mA cm^-2) showing the performance of the 10 nm and 20 nm MoO layers, and the three different plasma exposures, including their mean values with standard deviation.\"},{\"panel_id\":\"b\",\"text\":\"Scatter plot for the open-circuit voltage (V_OC, in mV) showing the performance of the 10 nm and 20 nm MoO layers, and the three different plasma exposures, including their mean values with standard deviation.\"},{\"panel_id\":\"c\",\"text\":\"Scatter plot for the fill factor (FF) showing the performance of the 10 nm and 20 nm MoO layers, and the three different plasma exposures, including their mean values with standard deviation.\"},{\"panel_id\":\"d\",\"text\":\"Scatter plot for the power conversion efficiency (η) showing the performance of the 10 nm and 20 nm MoO layers, and the three different plasma exposures, including their mean values with standard deviation.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Sample | 10 nm mean | 10 nm best | 20 nm mean | 20 nm best |\\n|---|---|---|---|---|\\n| 2 second O2 | 30.1 | 30.3 | 28.6 | 28.8 | \\n| 2 second O2 & Ar | 30.2 | 30.6 | 29.3 | 29.5 |\\n| 1 second O2 | 30.7 | 30.6 | 28.5 | 28.8 |\"},{\"panel_id\":\"b\",\"text\":\"| Sample | 10 nm mean | 10 nm best | 20 nm mean | 20 nm best |\\n|---|---|---|---|---|\\n| 2 second O2 | 630 | 640 | 600 | 620 | \\n| 2 second O2 & Ar | 620 | 630 | 640 | 650 |\\n| 1 second O2 | 575 | 570 | 560 | 565 |\"},{\"panel_id\":\"c\",\"text\":\"| Sample | 10 nm mean | 10 nm best | 20 nm mean | 20 nm best |\\n|---|---|---|---|---|\\n| 2 second O2 | 53 | 55 | 50 | 51 | \\n| 2 second O2 & Ar | 51 | 54 | 51 | 55 |\\n| 1 second O2 | 40 | 41 | 35 | 37 |\"},{\"panel_id\":\"d\",\"text\":\"| Sample | 10 nm mean | 10 nm best | 20 nm mean | 20 nm best |\\n|---|---|---|---|---|\\n| 2 second O2 | 10 | 10.5 | 8.5 | 9 | \\n| 2 second O2 & Ar | 9.5 | 10 | 9.5 | 10 |\\n| 1 second O2 | 7 | 7 | 5.5 | 6 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"On average, the 10 nm thick layers have a higher current density and a higher open-circuit voltage. Consequently, the 10 nm thick layers also have a higher fill factor (FF) and a higher power conversion efficiency (η). This means that, on average, the 10 nm thick layers outperform the 20 nm thick layers.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The 2-second O₂ plasma process.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"About 650 mV.\"}]},{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"About 10.5 %, which was obtained for 10 nm thick layer, which had a 2 second O2 plasma exposure.\"}]}]","bbox":[{"panel_id":"a","x":4,"y":8,"width":515,"height":392},{"panel_id":"b","x":522,"y":10,"width":486,"height":393},{"panel_id":"c","x":4,"y":361,"width":532,"height":547},{"panel_id":"d","x":480,"y":368,"width":528,"height":541}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/51/images/fig_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/51/images/fig_8.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/51/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/51/Ziegler et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"51","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":1011,"height":912,"image_format":"jpeg","image_sha256":"0d0a49cfc73ad967726180e5cdd013d76fbee8494cc3ba1c73b162fa70d1d2dd","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/test_atomic-layer-deposition_experimental-usecase_61_figure_3.jpg","caption":"Figure 3. QMS ion current at  $m / z$  ratios 16 and 17 for  $\\mathrm{H}_2 / \\mathrm{N}_2$  plasmas as a function of  $\\mathrm{H}_2$  fraction in the  $\\mathrm{H}_2 / \\mathrm{N}_2$  mixture. The  $\\mathrm{H}_2 / (\\mathrm{H}_2 + \\mathrm{N}_2)$  mixing ratios on the horizontal axis were determined using the ion currents at  $m / z$  ratios 2 and 14, corresponding to  $\\mathrm{H}_2^+$  and  $\\mathrm{N}^+$  (see the Supporting Information), before igniting the plasma. The total chamber pressure was kept constant at  $75\\mathrm{mTorr}$ .","id":"test/atomic-layer-deposition/experimental-usecase/61/figure_3","sample_id":"atomic-layer-deposition/experimental-usecase/61/figure_3","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This figure displays the ion current intensity as a function of the gas phase ratio H₂/(N₂+H₂) for two distinct mass-to-charge ratios: m/z = 16 (likely NH₂⁺ or CH₄⁺) and m/z = 17 (likely NH₃⁺). For both ion species, the ion current increases as the H₂ fraction increases, reaching a peak near a ratio of 0.6. Beyond this point, the current begins to decline. This suggests that intermediate hydrogen enrichment in the plasma favours the generation of these species, while overly hydrogen-rich environments suppress their formation.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| H₂/(N₂+H₂) | Ion Current @ m/z = 16 (10⁻⁹ A) | Ion Current @ m/z = 17 (10⁻⁹ A) |\\n|------------|-------------------------------|-------------------------------|\\n| 0.0        | 1.0                           | 0.8                           |\\n| 0.2        | 2.0                           | 1.6                           |\\n| 0.4        | 3.0                           | 2.2                           |\\n| 0.6        | 4.0                           | 2.8                           |\\n| 0.8        | 3.5                           | 2.5                           |\\n| 1.0        | 2.0                           | 1.5                           |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"As the H₂ fraction increases, the ion currents for both m/z = 16 and m/z = 17 initially increase, suggesting enhanced formation of these species under hydrogen-rich conditions. The peak ion current is observed around a gas ratio of 0.6, beyond which the current declines. This implies an optimal range of hydrogen content for generating maximum ion signal intensity, likely due to competing formation and quenching reactions in the plasma phase.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Optimal H₂ fraction around 0.6 maximizes ion production, excess hydrogen suppresses ion generation, plasma chemistry is tunable via gas composition\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Approximately 0.6.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":508,"height":430}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/61/images/figure_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/61/images/figure_3.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/61/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/61/Vos et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"61","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":508,"height":430,"image_format":"jpeg","image_sha256":"4e7464397f384041fd2c3d86116db5b13fa93f6c90ba94b7cb43881c981e036d","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/test_atomic-layer-deposition_experimental-usecase_65_fig_2.jpg","caption":"FIG. 2. Material properties as a function of plasma gas residence time  $\\tau$  for  $\\mathrm{SiN}_x$  films deposited at  $200^{\\circ}\\mathrm{C}$  and  $400^{\\circ}\\mathrm{C}$  using  $10\\mathrm{s}$ $\\mathbf{N}_2$  plasma exposure at various total gas flows and plasma gas pressures. The films had a thickness of  $20\\mathrm{nm}$  or higher. (a) Refractive index at  $2\\mathrm{eV}$  for  $\\mathrm{SiN}_x$  films deposited at  $200^{\\circ}\\mathrm{C}$ . (b) Oxygen and carbon impurity content and (c) wet-etch rates in BHF for  $\\mathrm{SiN}_x$  films deposited at  $400^{\\circ}\\mathrm{C}$ . For the buffered etch, a stock solution was used of a 7:1 volume ratio of  $40\\%$ $\\mathrm{NH_4F}$  in water and  $49\\%$  HF in water. The various gas flows are indicated in the legend and the data points are labeled with the associated gas pressures. The lines serve as a guide to the eye.","id":"test/atomic-layer-deposition/experimental-usecase/65/fig_2","sample_id":"atomic-layer-deposition/experimental-usecase/65/fig_2","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"},{"panel_id":"b","label":"multiple scatter plot"},{"panel_id":"c","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The refractive index decreases with increasing plasma gas residence time at 200°C.\"},{\"panel_id\":\"b\",\"text\":\"The impurity content increases with increasing plasma gas residence time at 400°C.\"},{\"panel_id\":\"c\",\"text\":\"The wet-etch rate increases with increasing plasma gas residence time at 400°C.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Plasma gas residence time (s) | Refractive Index |\\n|----------------|------------------|\\n| 0.1            | 1.9              |\\n| 0.4            | 1.85             |\\n| 0.6            | 1.77             |\\n| 0.8            | 1.75             |\\n| 1.1            | 1.73             |\"},{\"panel_id\":\"b\",\"text\":\"| Plasma gas residence time (s)  | Impurity Content oxygen (at. %) |Impurity Content carbon (at. %) |\\n|------------------|-------------------------|-------------------------|\\n| 0.1              | 2.5                      |0.5|\\n| 0.3              | 3.5                      |0.5|\\n| 0.4              | 4                        |1.5|\\n| 0.9              | 5                        |1.5|\"},{\"panel_id\":\"c\",\"text\":\"| Plasma gas residence time (s)  | Wet-etch Rate (nm/min) |\\n|------------------|------------------------|\\n| 0.1              | 0.5                    |\\n| 0.3              | 1                      |\\n| 0.4              | 1                      |\\n| 0.9              | 2                      |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"If the refractive is a measure for the density, then films deposited with long gas residence times have a lower density and are therefore less etch resistant.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The plasma pressure, the gas flows, and the plasma gas residence time.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"1 nm/min.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The argon gas flow and therefore the argon ions in the plasma can have an influence, where the ions can remove more impurities.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":1,"width":609,"height":274},{"panel_id":"b","x":3,"y":277,"width":605,"height":271},{"panel_id":"c","x":2,"y":545,"width":605,"height":304}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/65/images/fig_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/65/images/fig_2.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/65/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/65/Knoops et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"65","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":611,"height":850,"image_format":"jpeg","image_sha256":"efe30e630380839c20b97612ca6a4432e89e35b1e60036a7b5eef31fc0654de9","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/test_atomic-layer-deposition_experimental-usecase_67_FIG2_a.jpg","caption":"","id":"test/atomic-layer-deposition/experimental-usecase/67/FIG2_a","sample_id":"atomic-layer-deposition/experimental-usecase/67/FIG2_a","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This line chart presents the variation of Growth Per Cycle (GPC) with increasing plasma pulse time for two magnetic field strengths (850 mT and 1600 mT) under in-situ and ex-situ measurement conditions. GPC initially decreases rapidly with increasing pulse duration and plateaus near 0.05 nm/cy after about 50 seconds. Higher magnetic field strength (1600 mT) consistently results in slightly higher GPC than 850 mT across all time points, regardless of the measurement method.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Plasma Pulse (s) | GPC (nm/cy) |\\n|---|---|\\n| 0 | 0.11 |\\n| 10 | 0.105 |\\n| 20 | 0.07 |\\n| 30 | 0.06 |\\n| 40 | 0.055 |\\n| 50 | 0.05 |\\n| 60 | 0.05 |\\n| 80 | 0.05 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"As plasma pulse time increases, surface reactions proceed more fully, leading to more complete removal or modification of surface species per cycle. This enhanced efficiency saturates after a certain exposure duration, leading to a plateau in GPC. Longer pulses also reduce the chance of unreacted precursors or incomplete surface activation, which initially contributed to higher GPC in shorter pulses.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"1600 mT\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Marks etch saturation point, No GPC gain beyond 50 s, Optimizes cycle time for throughput and precision\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":670,"height":545}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG2_a.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG2_a.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/P. M. Litwin et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"67","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"not_found"},"width":670,"height":545,"image_format":"jpeg","image_sha256":"581781d562d4d341a28deeb03512aaae5f8b1ddba01bf3b933043a120751e040","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/test_atomic-layer-deposition_experimental-usecase_67_FIG2_d.jpg","caption":"","id":"test/atomic-layer-deposition/experimental-usecase/67/FIG2_d","sample_id":"atomic-layer-deposition/experimental-usecase/67/FIG2_d","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"d","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"d\",\"text\":\"This Raman spectra chart illustrates the vibrational modes of materials exposed to varying oxygen flow fractions through inductively coupled plasma (ICP). The three overlaid spectra correspond to O₂ flow fractions of 0.06, 0.29, and 0.53, with key Raman-active modes labeled along the spectral axis. Peaks corresponding to B₂g and A_g modes, as well as Al₂O₃, are clearly visible and consistent across all conditions, indicating that the overall structure remains stable across varying plasma environments. Minor differences in intensity at select peaks suggest possible changes in crystallinity or defect density influenced by oxygen content during deposition or treatment\"}]","data_extraction":"[{\"panel_id\":\"d\",\"text\":\"| Raman Shift (cm⁻¹) | Normalized Intensity (a.u.) |\\n|--------------------|-----------------------------|\\n| 100                | 0.05                        |\\n| 200                | 1.00                        |\\n| 300                | 0.60                        |\\n| 400                | 0.40                        |\\n| 500                | 0.55                        |\\n| 600                | 0.60                        |\\n| 700                | 0.50                        |\\n| 800                | 0.45                        |\\n| 900                | 0.50                        |\\n| 1000               | 0.60                        |\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Increasing the O₂ flow fraction from 0.06 to 0.53 does not significantly alter the positions of the Raman peaks, indicating that the chemical structure and vibrational modes remain largely consistent. However, subtle variations in peak intensities are observed. These changes suggest minor modifications in the material’s crystallinity or defect states, likely due to different oxidation conditions during treatment. For example, a higher flow fraction might reduce the presence of carbon-related defects or enhance oxide formation, subtly affecting Raman-active modes without drastically shifting peak positions\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"B₂g, A_g, and Al₂O₃ modes\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"O₂ flow fraction through ICP 0.06, 0.29, 0.53\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"}]}]","bbox":[{"panel_id":"d","x":0,"y":0,"width":728,"height":556}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG2_d.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG2_d.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/P. M. Litwin et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"67","first_classification_panel_id":"d","first_classification_label":"multiple scatter plot","caption_source":"not_found"},"width":728,"height":556,"image_format":"jpeg","image_sha256":"ce5658d7fa0ea8d030f7abb86a48378a4d26172d99c24396871ae9193f080ede","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/test_atomic-layer-deposition_experimental-usecase_67_FIG3_a.jpg","caption":"","id":"test/atomic-layer-deposition/experimental-usecase/67/FIG3_a","sample_id":"atomic-layer-deposition/experimental-usecase/67/FIG3_a","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This figure presents Raman spectra showing how the vibrational modes of a sample evolve with increasing inductively coupled plasma (ICP) power, ranging from 50 W to 950 W. The x-axis tracks Raman shift (in cm⁻¹), while the y-axis shows normalised intensity. Characteristic vibrational peaks are labeled with their symmetry assignments (e.g., B₂g, Ag), indicating specific phonon modes. Al₂O₃-associated features are also noted at key positions. As the ICP power increases, some peaks show a gradual reduction in intensity, suggesting potential structural or compositional changes under higher plasma energy. The consistent location of the phonon modes indicates the preserved crystal structure, while intensity differences reveal subtle variations due to processing.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Raman Shift (cm⁻¹) | Normalized Intensity @ 50 W | 200 W | 400 W | 600 W | 950 W |\\n|--------------------|-----------------------------|--------|--------|--------|--------|\\n| 100                | 0.05                        | 0.05   | 0.06   | 0.07   | 0.08   |\\n| 150                | 0.10                        | 0.10   | 0.11   | 0.12   | 0.12   |\\n| 200 (B₂g)          | 1.00                        | 0.98   | 0.96   | 0.93   | 0.90   |\\n| 250 (Ag)           | 0.55                        | 0.53   | 0.50   | 0.48   | 0.45   |\\n| 300 (B₂g)          | 0.70                        | 0.68   | 0.66   | 0.63   | 0.60   |\\n| 350 (Ag)           | 0.35                        | 0.34   | 0.33   | 0.31   | 0.30   |\\n| 400                | 0.20                        | 0.20   | 0.20   | 0.20   | 0.20   |\\n| 500 (Al₂O₃)        | 0.50                        | 0.50   | 0.52   | 0.55   | 0.58   |\\n| 600 (Ag)           | 0.30                        | 0.31   | 0.33   | 0.34   | 0.36   |\\n| 750 (Al₂O₃)        | 0.10                        | 0.10   | 0.12   | 0.14   | 0.16   |\\n| 950                | 0.08                        | 0.08   | 0.09   | 0.10   | 0.11   |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"As ICP power increases, the overall intensity of the Raman peaks, particularly those associated with B₂g and Ag modes, tends to decrease slightly. This trend may reflect increased disorder or reduced crystallinity introduced by higher plasma exposure. However, the peak positions remain consistent, suggesting that the fundamental bonding structure is preserved. The Al₂O₃-related peaks also exhibit a modest increase in prominence at higher ICP powers, potentially indicating enhanced surface contributions or phase-specific signals.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"B₂g\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"50 W, 200 W, 400 W, 600 W, 950 W\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":644,"height":517}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG3_a.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG3_a.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/P. M. Litwin et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"67","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"not_found"},"width":644,"height":517,"image_format":"jpeg","image_sha256":"098c8c172ca1bb618ba7586e7b7c28a77e141940b380ec51ea2d70b65d7fac70","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/test_atomic-layer-deposition_experimental-usecase_67_FIG5_b.jpg","caption":"FIG. 5. (a) Raman spectra of  $\\mathrm{VO}_x$  deposited with varying  $\\mathrm{O}_2$  flow fraction. (b) GPC as a function of  $\\mathrm{O}_2$  flow fraction. The  $y$  axis plotted on the same scale as Fig. 2(a) to highlight the small magnitude of change across the  $\\mathrm{O}_2$  flow fraction range investigated.","id":"test/atomic-layer-deposition/experimental-usecase/67/FIG5_b","sample_id":"atomic-layer-deposition/experimental-usecase/67/FIG5_b","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"b","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"b\",\"text\":\"The figure presents Growth Per Cycle (GPC) measurements as a function of the O₂:Ar flow ratio for both in-situ and ex-situ growth rate (GR) methods. Data points for both measurement types show nearly identical values across all tested flow ratios, indicating a high level of agreement between the two methods. Moreover, the GPC remains steady at approximately 0.06 nm/cycle, suggesting that variations in O₂:Ar ratio within the tested range have minimal effect on the deposition rate.\"}]","data_extraction":"[{\"panel_id\":\"b\",\"text\":\"| O₂:Ar Flow Ratio | In-situ GR (nm/cy) | Ex-situ GR (nm/cy) |\\n|------------------|--------------------|---------------------|\\n| 0.05             | 0.062              | 0.063               |\\n| 0.10             | 0.060              | 0.061               |\\n| 0.30             | 0.057              | 0.058               |\\n| 0.50             | 0.054              | 0.055               |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Comparing in-situ and ex-situ growth rate measurements helps validate the reliability and accuracy of growth monitoring techniques. In-situ methods provide real-time feedback, which is valuable for process control, while ex-situ measurements confirm final film thickness and uniformity. In this figure, the nearly identical GPC values from both approaches suggest that the in-situ method accurately reflects the actual film growth, thus confirming its suitability for real-time diagnostics and optimisation.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.06 nm/cycle\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Stable growth under gas variation, Reduced flow sensitivity, Robust precursor chemistry, Scalable reproducible processing\"}]}]","bbox":[{"panel_id":"b","x":0,"y":0,"width":630,"height":503}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG5_b.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG5_b.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/P. M. Litwin et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"67","first_classification_panel_id":"b","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":630,"height":503,"image_format":"jpeg","image_sha256":"b3bd728e40e235df434a6181dffd20bed3fb396523e1c1a9e644438c6901b208","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/test_atomic-layer-deposition_experimental-usecase_67_FIG7_d.jpg","caption":"","id":"test/atomic-layer-deposition/experimental-usecase/67/FIG7_d","sample_id":"atomic-layer-deposition/experimental-usecase/67/FIG7_d","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"d","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"d\",\"text\":\"This scatter plot illustrates the atomic percentages of carbon and nitrogen present on a surface following plasma exposure under different chamber pressures, ranging from 1000 to 5000 mTorr. The trends indicate relatively stable nitrogen incorporation between 1000–4000 mTorr, followed by a sharp increase at 5000 mTorr. Carbon follows a similar pattern but with more pronounced variability and a dramatic rise in atomic percentage at the highest pressure. These shifts align with a transition from inductively to capacitively coupled plasma modes, as indicated on the chart, implying that plasma coupling regime influences surface residue accumulation.\"}]","data_extraction":"[{\"panel_id\":\"d\",\"text\":\"| Pressure (mTorr) | Atomic % C | Atomic % N |\\n|---|---|---|\\n| 1000 | 1 | 0.5 |\\n| 2000 | 2 | 1 |\\n| 3000 | 3 | 1 |\\n| 4000 | 2 | 1 |\\n| 5000 | 11 | 6 |\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"As chamber pressure increases from 1000 to 4000 mTorr, the atomic percentages of both carbon and nitrogen remain relatively low and stable, with minor fluctuations. However, at 5000 mTorr, there is a sharp increase in both elements, with carbon rising to 11% and nitrogen to 6%. This suggests a threshold effect, where a change in plasma conditions, specifically a shift to capacitively coupled operation, leads to greater surface contamination or retention of carbonaceous and nitrogenous species. This behavior underscores the importance of controlling pressure to manage surface purity during plasma-based processing\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"5000 mTorr\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"CCP may leave more C/N residue, Lower pressure , ICP , Pressure control\"}]}]","bbox":[{"panel_id":"d","x":0,"y":0,"width":725,"height":550}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG7_d.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG7_d.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/P. M. Litwin et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"67","first_classification_panel_id":"d","first_classification_label":"multiple scatter plot","caption_source":"not_found"},"width":725,"height":559,"image_format":"jpeg","image_sha256":"16d3993ac549a8ab78809787aaf8f3d2fb6d12abeacfdaa0822852e0eded69da","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/test_atomic-layer-deposition_experimental-usecase_67_FIG8_a.jpg","caption":"FIG. 8. (a) Atomic O density and (b) ion flux and plasma potential as a function of chamber pressure during plasma operation. Some conditions measured in both inductively and capacitively coupled modes.","id":"test/atomic-layer-deposition/experimental-usecase/67/FIG8_a","sample_id":"atomic-layer-deposition/experimental-usecase/67/FIG8_a","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart shows how atomic oxygen density depends on pressure under two plasma conditions: inductively coupled and capacitively coupled. As pressure increases from 1000 to 5000 mTorr, atomic oxygen density rises significantly for the inductively coupled case, following an approximately logarithmic trend. Capacitive coupling data, shown only at higher pressures, yield slightly lower oxygen densities compared to the inductive case. This comparison underscores the enhanced dissociation and plasma density typically achieved in inductively coupled plasma sources.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Pressure (mTorr) | Atomic O Density (Inductive) (atoms/m³) | Atomic O Density (Capacitive) (atoms/m³) |\\n|------------------|-------------------------------------------|--------------------------------------------|\\n| 1000             | 5 × 10¹⁹                                 | –                                          |\\n| 2000             | 2 × 10²⁰                                 | –                                          |\\n| 3000             | 4 × 10²⁰                                 | –                                          |\\n| 4000             | 6 × 10²⁰                                 | 5 × 10²⁰                                   |\\n| 5000             | 7 × 10²⁰                                 | 6 × 10²⁰                                   |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Atomic oxygen density increases with pressure for both inductively and capacitively coupled systems, though the inductively coupled configuration achieves higher densities overall. Between 1000 and 5000 mTorr, inductively coupled plasma results in a steady logarithmic increase in oxygen density, peaking at approximately 7 × 10²⁰ atoms/m³. The capacitively coupled data, available only at 4000 and 5000 mTorr, show slightly lower densities, suggesting reduced dissociation efficiency. This trend indicates that inductively coupled systems are more effective at generating reactive oxygen species under similar pressure conditions\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"6 × 10²⁰ atoms/m³.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Pressure (mTorr), Plasma coupling type (indcutive vs. capacitive)\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":628,"height":522}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG8_a.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG8_a.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/P. M. Litwin et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"67","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":628,"height":522,"image_format":"jpeg","image_sha256":"b289008bf107b3b51d534ec182977e5a7520668ee1208b64c5bda89cb5de07d6","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/test_atomic-layer-deposition_experimental-usecase_67_FIG9.jpg","caption":"FIG. 9. Plot of the ion flux and plasma potential, following the arguments of Boris et al. (Ref. 2). The data represent the measured experimental data from this work for inductively coupled plasmas (unless otherwise noted). Thin dashed black lines represent the lines of constant energy flux density. The bold red dashed lines represent the approximated critical energy flux density necessary for the onset of crystallization.","id":"test/atomic-layer-deposition/experimental-usecase/67/FIG9","sample_id":"atomic-layer-deposition/experimental-usecase/67/FIG9","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot describes the relationship between plasma potential and ion flux across a range of pressures, using different symbols to represent each pressure condition. A red dashed boundary line demarcates a critical energy flux density threshold that separates regimes where crystalline versus amorphous films are likely to form. Data points above this line (higher flux and potential) are associated with crystalline film growth, while those below correspond to amorphous outcomes. The plot also labels “Capacitively coupled” in the lower right, indicating a trend at higher potentials and lower fluxes. This figure helps define the operational window for achieving specific film crystallinity via control of plasma conditions.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Pressure (mTorr) | Plasma Potential (V) | Ion Flux (cm⁻²·s⁻¹) |\\n|------------------|----------------------|----------------------|\\n| 830                    | ~18                             | ~5×10¹⁵              |\\n| 1600                  | ~28                                | ~2×10¹⁶           |\\n| 2400                  | ~45                                 | ~1.5×10¹⁶      |\\n| 3500                  | ~70                                 | ~5×10¹⁶          |\\n| 4300                  | ~100                               | ~8×10¹⁶          |\\n| 5000 (hexagon)   | ~250                   | ~2×10¹⁷              |\\n| 5000 (circle)         | ~900                        | ~4×10¹⁷          |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Crystalline film, Amorphous film, Critical energy flux density separates the two regimes\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"As plasma potential increases, the ion flux generally rises across all pressure conditions shown. This positive trend is nonlinear and captured more effectively using a log–log scale. Lower pressure markers (e.g., 830 mTorr) cluster at low plasma potentials and ion fluxes, while higher pressure conditions (e.g., 5000 mTorr) span into the high-potential, high-flux region. This suggests that both plasma potential and operating pressure jointly contribute to achieving higher ion flux densities.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Approximately 1 x 10^16 cm^-2 s^-1\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":658,"height":528}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG9.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG9.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/P. M. Litwin et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"67","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":658,"height":528,"image_format":"jpeg","image_sha256":"a151f9369215768ec1d9ee7e5f9fddeb2a7b3166ae574cd5bb71e01cd3ef54ed","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/test_atomic-layer-deposition_simulation-usecase_18_fig_16.jpg","caption":"Fig. 16. Comparison of simulated deposition rate with experiment [30] as a function of  $\\mathrm{NH_3}$  flow rate using the reactor shown in Fig. 15.","id":"test/atomic-layer-deposition/simulation-usecase/18/fig_16","sample_id":"atomic-layer-deposition/simulation-usecase/18/fig_16","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot represents the variation of deposition rate (microns/hr) of simulated GaN against experimental data from Thrush et al. for a Thomas Swan reactor as a function of ammonia flow. The deposition rate increases steadily from about 0.2 to 1.8 µm/hr with higher NH₃ flow, and the model matches this trend closely. The agreement shows that the reaction mechanism remains accurate even under the high-flow conditions typical of commercial reactors.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Ammonia Flow (slm) | Deposition Rate (microns/hr) | |\\n|---|---|---|\\n| 2 | 0.2 | Experiment |\\n| 2 | 0.2 | Simulation |\\n| 3 | 0.5 | Experiment |\\n| 3 | 0.4 | Simulation |\\n| 4 | 0.7 | Experiment |\\n| 4 | 0.6 | Simulation |\\n| 5 | 0.9 | Experiment |\\n| 5 | 0.8 | Simulation |\\n| 6 | 1.0 | Experiment |\\n| 6 | 1.0 | Simulation |\\n| 7 | 1.2 | Experiment |\\n| 7 | 1.2 | Simulation |\\n| 8 | 1.4 | Experiment |\\n| 8 | 1.3 | Simulation |\\n| 9 | 1.6 | Experiment |\\n| 9 | 1.5 | Simulation |\\n| 10 | 1.7 | Experiment |\\n| 10 | -- | Simulation |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The required flow rate is approximately 7 slm.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The Thomas Swan reactor (likely a vertical rotating-disc type) might hit a practical limit sooner. While both show increasing rates with NH₃ flow, high-flow conditions in commercial reactors can lead to premature gas-phase reactions (particle formation) or excessive precursor waste, imposing a practical and economic ceiling on useful ammonia flow long before the stagnation-flow reactor, which is designed for fundamental study, would.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The increase indicates the process is in a reaction-limited regime with respect to nitrogen. Higher ammonia flow increases the concentration of NH₃ and its decomposition products (like NH₂, NH) in the gas phase and at the surface. Since the surface has abundant Gallium (Ga) atoms, the rate-limiting step is the reaction of these nitrogen species with Ga, so their increased flux directly accelerates the deposition rate.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"YES, the close agreement between simulation and experiment across the operational range proves the model's predictive capability for this reactor and chemistry. Therefore, engineers can confidently use the simulation to interpolate or slightly extrapolate the required ammonia flow setting to hit the precise target growth rate of 1.5 µm/hr in a scaled or modified reactor design, saving time and cost on experimental trials.\"}]}]","bbox":[{"panel_id":"a","x":4,"y":5,"width":576,"height":391}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/18/images/fig_16.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/18/images/fig_16.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/18/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/18/Debasis Sengupta et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"18","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":581,"height":397,"image_format":"jpeg","image_sha256":"54d63a307f1c3887be10782a9dd31326a0f2d02106c1606f7797b0628daff4c5","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/test_atomic-layer-deposition_simulation-usecase_3_fig_5.jpg","caption":"FIG. 5. (Color online) Precursor distributions during the full cycle of flow simulations for: (a) TMA molar fraction for carrier gas flow rate of  $20~\\mathrm{sccm}$ ; (b) TMA molar fraction for carrier gas flow rate of  $200~\\mathrm{sccm}$ ; (c) water molar fraction for carrier gas flow rate of  $20~\\mathrm{sccm}$ ; and (d) water molar fraction for carrier gas flow rate of  $200~\\mathrm{sccm}$ . A is located in inlet area and C in the outlet area.","id":"test/atomic-layer-deposition/simulation-usecase/3/fig_5","sample_id":"atomic-layer-deposition/simulation-usecase/3/fig_5","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"},{"panel_id":"b","label":"multiple scatter plot"},{"panel_id":"c","label":"multiple scatter plot"},{"panel_id":"d","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"TMA molar fraction during a full ALD cycle at 20 sccm carrier gas flow. Inlet sample A responds fastest during the 0.015 s pulse, reaching saturation immediately. During the 10 s purge, TMA remains above 50% at most positions, indicating insufficient purging.\"},{\"panel_id\":\"b\",\"text\":\"TMA molar fraction at 200 sccm carrier gas. Pulsing behavior is similar to (a), but purging is much more effective with all samples dropping below 20% concentration.\"},{\"panel_id\":\"c\",\"text\":\"Water molar fraction at 20 sccm carrier gas. Water fills the chamber faster than TMA. High residual concentrations (>80%) persist after purging at positions B, C, D, E.\"},{\"panel_id\":\"d\",\"text\":\"Water molar fraction at 200 sccm carrier gas. Purging efficiency is greatly improved, with concentrations dropping significantly during the purge step.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Flow Time (s) | A (TMA Molar Fraction) | B (TMA Molar Fraction) | C (TMA Molar Fraction) | D (TMA Molar Fraction) | E (TMA Molar Fraction) |\\n|---|---|---|---|---|---|\\n| 0.000 | 0.60 | 0.00 | 0.00 | 0.00 | 0.00 |\\n| 0.001 | 0.90 | 0.00 | 0.05 | 0.00 | 0.10 |\\n| 0.002 | 1.00 | 0.02 | 0.18 | 0.02 | 0.22 |\\n| 0.003 | 1.00 | 0.05 | 0.28 | 0.05 | 0.40 |\\n| 0.004 | 1.00 | 0.08 | 0.40 | 0.08 | 0.58 |\\n| 0.006 | 1.00 | 0.15 | 0.60 | 0.15 | 0.80 |\\n| 0.008 | 1.00 | 0.35 | 0.85 | 0.35 | 0.92 |\\n| 0.010 | 1.00 | 0.50 | 0.95 | 0.50 | 0.98 |\\n| 0.012 | 1.00 | 0.60 | 0.98 | 0.60 | 1.00 |\\n| 0.014 | 1.00 | 0.65 | 0.98 | 0.65 | 1.00 |\\n| 10.015 | 0.45 | 0.70 | 0.80 | 0.70 | 0.75 |\\n| 10.016 | 0.02 | 0.72 | 0.78 | 0.72 | 0.65 |\\n| 10.017 | 0.00 | 0.70 | 0.65 | 0.70 | 0.40 |\\n| 10.018 | 0.00 | 0.60 | 0.40 | 0.60 | 0.20 |\\n| 10.019 | 0.00 | 0.40 | 0.20 | 0.40 | 0.10 |\\n| 10.020 | 0.00 | 0.20 | 0.10 | 0.20 | 0.05 |\\n| 10.022 | 0.00 | 0.05 | 0.02 | 0.05 | 0.00 |\"},{\"panel_id\":\"b\",\"text\":\"| Flow Time (s) | A (TMA Molar Fraction) | B (TMA Molar Fraction) | C (TMA Molar Fraction) | D (TMA Molar Fraction) | E (TMA Molar Fraction) |\\n|---|---|---|---|---|---|\\n| 0.000 | 0.60 | 0.00 | 0.00 | 0.00 | 0.00 |\\n| 0.001 | 0.90 | 0.00 | 0.05 | 0.00 | 0.10 |\\n| 0.003 | 1.00 | 0.05 | 0.28 | 0.05 | 0.40 |\\n| 0.005 | 1.00 | 0.15 | 0.55 | 0.15 | 0.70 |\\n| 0.007 | 1.00 | 0.35 | 0.85 | 0.35 | 0.90 |\\n| 0.009 | 1.00 | 0.50 | 0.95 | 0.50 | 0.98 |\\n| 0.011 | 1.00 | 0.60 | 1.00 | 0.60 | 1.00 |\\n| 0.013 | 1.00 | 0.65 | 1.00 | 0.65 | 1.00 |\\n| 10.015 | 0.10 | 0.75 | 1.00 | 0.75 | 0.80 |\\n| 10.016 | 0.00 | 0.78 | 0.95 | 0.78 | 0.60 |\\n| 10.017 | 0.00 | 0.75 | 0.80 | 0.75 | 0.30 |\\n| 10.018 | 0.00 | 0.60 | 0.40 | 0.60 | 0.15 |\\n| 10.019 | 0.00 | 0.35 | 0.10 | 0.35 | 0.05 |\\n| 10.020 | 0.00 | 0.15 | 0.02 | 0.15 | 0.00 |\\n| 10.022 | 0.00 | 0.02 | 0.00 | 0.02 | 0.00 |\"},{\"panel_id\":\"c\",\"text\":\"| Flow Time (s) | A (Water Molar Fraction) | B (Water Molar Fraction) | C (Water Molar Fraction) | D (Water Molar Fraction) | E (Water Molar Fraction) |\\n|---|---|---|---|---|---|\\n| 10.015 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 |\\n| 10.016 | 0.85 | 0.20 | 0.10 | 0.20 | 0.25 |\\n| 10.017 | 0.95 | 0.40 | 0.40 | 0.40 | 0.55 |\\n| 10.018 | 1.00 | 0.60 | 0.58 | 0.55 | 0.80 |\\n| 10.019 | 1.00 | 0.80 | 0.78 | 0.70 | 0.90 |\\n| 10.021 | 1.00 | 0.95 | 0.90 | 0.90 | 0.98 |\\n| 10.023 | 1.00 | 0.98 | 1.00 | 0.98 | 1.00 |\\n| 20.030 | 0.95 | 1.00 | 1.00 | 1.00 | 1.00 |\\n| 20.032 | 0.80 | 1.00 | 1.00 | 1.00 | 1.00 |\\n| 20.034 | 0.68 | 1.00 | 1.00 | 1.00 | 1.00 |\\n| 20.036 | 0.55 | 1.00 | 1.00 | 1.00 | 1.00 |\\n| 20.038 | 0.45 | 1.00 | 1.00 | 1.00 | 1.00 |\\n| 20.040 | 0.35 | 1.00 | 1.00 | 1.00 | 1.00 |\"},{\"panel_id\":\"d\",\"text\":\"| Flow Time (s) | A (Water Molar Fraction) | B (Water Molar Fraction) | C (Water Molar Fraction) | D (Water Molar Fraction) | E (Water Molar Fraction) |\\n|---|---|---|---|---|---|\\n| 10.015 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 |\\n| 10.016 | 0.85 | 0.25 | 0.10 | 0.22 | 0.28 |\\n| 10.017 | 0.98 | 0.40 | 0.30 | 0.38 | 0.58 |\\n| 10.018 | 1.00 | 0.60 | 0.55 | 0.55 | 0.82 |\\n| 10.020 | 1.00 | 0.85 | 0.78 | 0.80 | 0.95 |\\n| 10.022 | 1.00 | 0.95 | 0.95 | 0.95 | 1.00 |\\n| 20.030 | 0.02 | 1.00 | 1.00 | 1.00 | 1.00 |\\n| 20.032 | 0.00 | 1.00 | 0.90 | 0.95 | 0.85 |\\n| 20.034 | 0.00 | 0.92 | 0.78 | 0.85 | 0.55 |\\n| 20.036 | 0.00 | 0.60 | 0.40 | 0.70 | 0.25 |\\n| 20.038 | 0.00 | 0.45 | 0.15 | 0.48 | 0.05 |\\n| 20.040 | 0.00 | 0.38 | 0.10 | 0.42 | 0.00 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The ALD half-cycle phases: pulsing is the brief 0.015 s precursor injection, purging is the ~10 s period when only carrier gas flows to remove excess precursor.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Sample A is closest to the precursor inlets, so it experiences the highest concentration first before precursors diffuse to other chamber regions.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Water vapor fills the chamber faster than TMA. Comparing panels (a) and (c), water reaches high molar fractions at all sample positions more quickly during the pulse. This is due to water's higher vapor pressure (23.68 Torr vs 9.43 Torr for TMA at 25°C) providing a stronger driving force for injection.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Carrier gas flow rate has little influence on the pulsing step since precursor injection is driven by vapor pressure. However, it strongly affects purging efficiency. At 20 sccm, significant precursor residue remains after purging (>50% for TMA, >80% for water at some positions), while 200 sccm effectively reduces concentrations below 20%.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":675,"height":570},{"panel_id":"b","x":658,"y":0,"width":638,"height":554},{"panel_id":"c","x":0,"y":564,"width":663,"height":551},{"panel_id":"d","x":659,"y":553,"width":634,"height":565}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/3/images/fig_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/3/images/fig_5.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/3/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/3/Dongqing Pan et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"3","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":1306,"height":1120,"image_format":"jpeg","image_sha256":"0abfd1562598df7a7d24f8295c85d37dd27de8fc822c74d20161afd507f70ad1","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/test_atomic-layer-deposition_simulation-usecase_34_figure_3.jpg","caption":"Figure 3. a)  $\\mathbb{D} / \\mathbb{D}^{\\prime}$  peak ratio of the graphene spectra for the different  $\\mathsf{H}_{2}$  and  $\\Omega_2$  plasma treatment times and b) broadening of the Raman G and 2D band as a function of the plasma treatment time, determined from the data in Figure 2.","id":"test/atomic-layer-deposition/simulation-usecase/34/figure_3","sample_id":"atomic-layer-deposition/simulation-usecase/34/figure_3","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"},{"panel_id":"b","label":"multiple line chart"},{"panel_id":"c","label":"multiple line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart displays the D/D' peak ratio for different exposure times of H₂ and O₂ plasmas.\"},{\"panel_id\":\"b\",\"text\":\"The chart displays the width of the Raman G band for different exposure times of H₂ and O₂ plasmas.\"},{\"panel_id\":\"c\",\"text\":\"The chart displays the width of the Raman 2D band for different exposure times of H₂ and O₂ plasmas.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Plasma time (min) | D/D' ratio H2 plasma | D/D' ratio O2 plasma |\\n|---|---|---|\\n| 1 |  | 10 |\\n| 2 | 12.5 | 10 |\\n| 3 | 9.5 | 6.5 |\\n| 4 | 11 |  |\\n| 5 | 10 | 5 |\\n| 6 | 10 |  |\\n| 7 |  |  |\\n| 8 | 9.5 |  |\"},{\"panel_id\":\"b\",\"text\":\"| Plasma Time (min) | Width G (cm⁻¹) H2 plasma | Width G (cm⁻¹) O2 plasma |\\n|---|---|---|\\n| 0 | 10 | 10 |\\n| 1 |  | 20 |\\n| 2 | 10 | 25 |\\n| 3 | 15 | 40 |\\n| 4 | 15 |  |\\n| 5 | 15 | 45 |\\n| 6 | 20 |  |\\n| 7 |  |  |\\n| 8 | 25 |  |\"},{\"panel_id\":\"c\",\"text\":\"| Plasma Time (min) | Width 2D (cm⁻¹) H2 plasma | Width 2D (cm⁻¹) O2 plasma |\\n|---|---|---|\\n| 0 | 35 | 35 |\\n| 1 |  | 50 |\\n| 2 | 35 | 50 |\\n| 3 | 35 | 70 |\\n| 4 | 35 |  |\\n| 5 | 40 | 110 |\\n| 6 | 40 |  |\\n| 7 |  |  |\\n| 8 | 45 |  |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Exposure to H2 plasma predominantly leads to the functionalisation of graphene, where the D/D' ratio remains more or less constant around 10 as a function of plasma treatment time. This indicates that the relative amount of functionalization compared to defects stays constant even though the graphene is slowly being etched by the H2 plasma.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"About 6.5.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The changes are minimal compared to O2 plasma and significant broadening only occurs 4 minutes of H2 plasma exposure.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The effect of H2 plasma exposure is minimal, similar to the Raman G band. However, here significant broadening only occurs after 7 minutes of H2 plasma exposure.\"}]}]","bbox":[{"panel_id":"a","x":1,"y":2,"width":692,"height":523},{"panel_id":"b","x":712,"y":6,"width":660,"height":242},{"panel_id":"c","x":711,"y":260,"width":662,"height":265}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/34/images/figure_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/34/images/figure_3.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/34/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/34/René H. J. Vervuurt et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"34","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":1375,"height":525,"image_format":"jpeg","image_sha256":"2735e5b9c043003c4ac8f14ffd7655d81f539166ac7c55b9b3fc1d5c3e6a80af","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/test_atomic-layer-deposition_simulation-usecase_34_figure_8.jpg","caption":"Figure 8. Contact resistance determined from the CTLM measurements as a function of the  $\\mathrm{H}_{2}$  or  $\\mathrm{O}_{2}$  plasma treatment time. A comparison is made between the devices synthesized by AS-ALD and plasma functionalization and EBL and e-beam evaporation (PVD).","id":"test/atomic-layer-deposition/simulation-usecase/34/figure_8","sample_id":"atomic-layer-deposition/simulation-usecase/34/figure_8","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the relationship between plasma exposure time and contact resistance for samples treated with O₂ Plasma ALD, H₂ Plasma ALD, and H₂ Plasma PVD.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Plasma Exposure Time (min) | Contact Resistance (Ωµm) O2 Plasma ALD | Contact Resistance (Ωµm) H2 Plasma ALD | Contact Resistance (Ωµm) H2 Plasma PVD |\\n|---|---|---|---|\\n| 0 |  |  | 11500 |\\n| 5 | 4000 | 1000 | 4000 |\\n| 7 |  | 500 |  |\\n| 10 | 1500 | 1500 |  |\\n| 15 | 1500 | 500 |  |\\n| 20 |  | 500 |  |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"11.5 ± 0.4 kΩ µm.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The contact resistance is relatively constant around 0.5 - 1 kΩ µm.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 4 kΩ µm.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Generally, all samples show improved contact resistance when exposed to plasma. The surface groups created by the plasma treatment enhance the transport of charge carriers from the metal to the graphene through an increase of the number of chemical bonds between the metal and the graphene. The additional improvement observed for the ALD created contacts could be due to the improved wetting interaction of the deposited Pt with the underlying graphene or an improved Pt morphology due to the use of ALD. Additionally, as opposed to Pt PVD, for the creation of the Pt ALD contacts, a Ti adhesion layer is no longer required. This could also have a positive effect on the metal-graphene interaction, due to the different work function and interaction strength of Ti compared to Pt.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":666,"height":495}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/34/images/figure_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/34/images/figure_8.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/34/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/34/René H. J. Vervuurt et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"34","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":666,"height":495,"image_format":"jpeg","image_sha256":"e60e22b820f702308b3907e5ba670a7adb3f55f76701d9de144e1afeba87489d","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/test_atomic-layer-etching_experimental-usecase_1_figure_8.jpg","caption":"Figure 8. Graphs showing comparison of number of publications in ALD versus ALEt. Plot constructed from a SCOPUS database search for the terms \"Atomic Layer Etching\" and \"Atomic Layer Depository\", retrieved on December  $3^{\\mathrm{rd}}$  2014.","id":"test/atomic-layer-etching/experimental-usecase/1/figure_8","sample_id":"atomic-layer-etching/experimental-usecase/1/figure_8","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":0,"y":0,"width":675,"height":572}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/1/images/figure_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/1/images/figure_8.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/1/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/1/Atomic Layer Etching An Industry Perspective.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"1","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":675,"height":572,"image_format":"jpeg","image_sha256":"3a8a7c9065721d27426a72ad989c007c4f0829a4ac5c51914553b89dab443bf9","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/test_atomic-layer-etching_experimental-usecase_20_figure_8.jpg","caption":"Figure 8. AlN film thickness vs number of  $\\mathrm{XeF}_2 / \\mathrm{BCl}_3$  cycles on single-crystal AlN film for various temperatures. Temperature-dependent etch rates vary from  $0.19\\mathrm{\\AA}$  cycle at  $212^{\\circ}C$  to  $0.93\\mathrm{\\AA}/$  cycle at  $298^{\\circ}C$","id":"test/atomic-layer-etching/experimental-usecase/20/figure_8","sample_id":"atomic-layer-etching/experimental-usecase/20/figure_8","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows the thickness loss of AlN as a function of XeF₂/BCl₃ cycles at various temperatures. For all cases, the thickness decreases as the number of cycles increases.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| XeF₂/BCl₃ Cycles | 212°C | 230°C | 255°C | 263°C | 280°C | 298°C |\\n| --- | --- | --- | --- | --- | --- | --- |\\n| 0 | 0 | 0 | 0 | 0 | 0 | 0 |\\n| 10 | 0 | -5 | -6 | -5 | -6 | -7 |\\n| 20 | -2 | -10 | -12 | -12 | -13 | -17 |\\n| 30 | -4 | -16 | -16 | -19 | -21 | -27 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"There is a linear relationship  between these two variables.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The AlN thickness decreases as the number of cycles increases for all temperatures.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At 212 °C.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"It can be concluded that the AlN etch rate can be controlled by the XeF2 and BCl3 exposures.\"}]}]","bbox":[{"panel_id":"a","x":1,"y":2,"width":656,"height":567}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/20/images/figure_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/20/images/figure_8.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/20/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/20/Thermal Atomic Layer Etching of Aluminum Nitride.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"20","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":658,"height":570,"image_format":"jpeg","image_sha256":"f23a0549a70c0db1d7bc934ab3dd6d5d75cd6d2a42cf8a1acc61f8abbb2eb6c8","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/test_atomic-layer-etching_experimental-usecase_21_36d51786c634f31a43a7df7d794ccd77ac03e8f8ed581d6cf1e61c63269b5f7f.jpg","caption":"Fig. 5a shows the GIXRD results for the initial CoO thin film. The as","id":"test/atomic-layer-etching/experimental-usecase/21/36d51786c634f31a43a7df7d794ccd77ac03e8f8ed581d6cf1e61c63269b5f7f","sample_id":"atomic-layer-etching/experimental-usecase/21/36d51786c634f31a43a7df7d794ccd77ac03e8f8ed581d6cf1e61c63269b5f7f","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows the change in CoOx thickness over cycles at two temperatures, 200 °C and 250 °C. The rate of change are 0.09 Å/cycle and 0.43 Å/cycle for 200 °C and 250 °C, respectively.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of Cycles | CoOx Thickness Change (Å) 200 °C | CoOx Thickness Change (Å) 250 °C |\\n| --- | --- | --- |\\n| 0 | 0 | 0 |\\n| 5 | 0 | -2.0 |\\n| 10 | -1.0 | -4.0 |\\n| 15 | -1.8 | -6.0 |\\n| 20 | -2.0 | -8.0 |\\n| 25 | -2.5 | -10.0 |\\n| 30 | -2.8 | -12.5 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The change in the CoOx thickness is evaluated as a function of the temperature and the number of cycles.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The thickness of the CoOx decreases as the number of cycles increases for both temperatures, 200 °C and 250 °C.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"As can be seen in the figure, a higher temperature (i. e. 250 °C) produces greater changes in the CoOx thickness, reaching an etch per cycle value of 0.43 Å/cycle.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At this temperature, it is possible to achieve a higher etch rate of 0.43 Å/cycle compared to 0.09 Å/cycle at 200 °C.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":5,"width":651,"height":563}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/21/images/36d51786c634f31a43a7df7d794ccd77ac03e8f8ed581d6cf1e61c63269b5f7f.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/21/images/36d51786c634f31a43a7df7d794ccd77ac03e8f8ed581d6cf1e61c63269b5f7f.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/21/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/21/Thermal atomic layer etching of CoO using acetylacetone and ozone.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"21","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":656,"height":570,"image_format":"jpeg","image_sha256":"83baf17b43258fd97e797d283a73834e8bf9a436ad86410818e9de0e0d9677f2","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/test_atomic-layer-etching_experimental-usecase_21_fig_2.jpg","caption":"Fig. 2. CoO film thickness change versus number of Hacac doses at 250 and  $270^{\\circ}C$  Dashed lines show the linear fits.","id":"test/atomic-layer-etching/experimental-usecase/21/fig_2","sample_id":"atomic-layer-etching/experimental-usecase/21/fig_2","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows the change in CoOx thickness over the number of Hacac doses at two different temperatures, 250°C and 275°C. The data points are represented by blue squares and red circles respectively, with black dashed lines indicating the trend and the etch rate per dose (0.08 Å/dose at 250°C and 0.09 Å/dose at 275°C).\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of Hacac Doses | CoOx Thickness Change (Å) 250 °C | CoOx Thickness Change (Å) 275 °C |\\n| --- | --- | --- |\\n| 0 | 0 | 0 |\\n| 20 | -1.5 | -6.4 |\\n| 40 | -4.0 | -8.4 |\\n| 60 | -4.8 | -9.2 |\\n| 80 | -8.0 | - |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Spectroscopic ellipsometry is used for thickness measurements in this case.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"As the number of Hacac doses increases, it can be seen that the CoO thickness decreases for both temperatures.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The highest value, with only a slight difference, is 0.09 Å/dose at 275 °C.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Increasing the temperature doesn't result in a significant difference in etch per dose; the film thickness is similar in both cases.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":1,"width":660,"height":561}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/21/images/fig_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/21/images/fig_2.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/21/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/21/Thermal atomic layer etching of CoO using acetylacetone and ozone.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"21","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":667,"height":566,"image_format":"jpeg","image_sha256":"4b4ecea448b12b58b199f246387288fd263dde088b7a4306fe979ac13058c766","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/test_atomic-layer-etching_experimental-usecase_25_figure_12.jpg","caption":"Figure 12.  $\\mathrm{WO}_3$  and W thickness versus number of half-cycles during W ALE under self-limiting conditions using  $\\mathrm{O}_2 / \\mathrm{O}_3,$ $\\mathrm{BCl}_3,$  and HF as reactants at  $207^{\\circ}C$  a  $\\mathrm{WO}_3$  thickness showing oscillation of  $\\mathrm{WO}_3$  thickness after  $\\mathrm{O}_2 / \\mathrm{O}_3$  exposure and  $\\mathrm{BCl}_3 / \\mathrm{HF}$  reaction.b)W thickness showing linear reduction versus number of half-cycles with an etch rate of  $2.44\\mathrm{\\AA}/$  cycle.","id":"test/atomic-layer-etching/experimental-usecase/25/figure_12","sample_id":"atomic-layer-etching/experimental-usecase/25/figure_12","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows the thickness of WO₃ after exposure to O₂/O₃ and BCl₃/HF as a function of the number of half-cycles.\"},{\"panel_id\":\"b\",\"text\":\"The figure illustrates the thickness of W as a function of the number of half-cycles, with an etch rate of 2.44 Å/cycle.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of Half-Cycles | WO₃ Thickness (Å) After O₂/O₃ | WO₃ Thickness (Å) After BCl₃/HF |\\n|---|---|---|\\n| 0 | 35 | 27 |\\n| 10 | 42 | 36 |\\n| 20 | 37 | 30 |\\n| 30 | 32 | 25 |\\n| 40 | 29 | 22 |\\n| 50 | 26 | 21 |\\n| 60 | 25 | 17 |\"},{\"panel_id\":\"b\",\"text\":\"| Number of Half-Cycles | W Thickness (Å) |\\n|---|---|\\n| 0 | 230 |\\n| 10 | 215 |\\n| 20 | 200 |\\n| 30 | 185 |\\n| 40 | 170 |\\n| 50 | 160 |\\n| 60 | 150 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It seems that in the initial stages of the process, there still is an increase in thickness of WO3, but it seems to settle towards a constant thickness after 50 half cycles\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"No, it does not. It could be that the initial thickness of W enables the modification step to have a larger effect, whilst the etching of W remains relatively constant.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"2800, 329, and 3150 mTorr s respectively\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"By determining the slope and multiplying this with 2, as the x-axis denotes half-cycles and not full cycles.\"}]}]","bbox":[{"panel_id":"a","x":4,"y":1,"width":665,"height":426},{"panel_id":"b","x":2,"y":431,"width":663,"height":464}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/figure_12.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/figure_12.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/WO3 and W Thermal Atomic Layer Etching Using Conversion-Fluorination.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"25","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":673,"height":900,"image_format":"jpeg","image_sha256":"810c10b1e6c38485015233458318ac1fced1ff4ea0cf6f11451c91edbb2c1be2","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/test_atomic-layer-etching_experimental-usecase_25_figure_13.jpg","caption":"Figure 13. Enlargement of  $\\mathrm{WO}_3$  thickness versus number of halfcycles showing increase and decrease of  $\\mathrm{WO}_3$  thickness after  $\\mathrm{O}_2 / \\mathrm{O}_3$  exposure and  $\\mathrm{BCl}_3 / \\mathrm{HF}$  reaction.","id":"test/atomic-layer-etching/experimental-usecase/25/figure_13","sample_id":"atomic-layer-etching/experimental-usecase/25/figure_13","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows the thickness of WO₃ after exposure to O₂/O₃ and BCl₃/HF at 207°C, measured in Ångströms, as a function of the number of half-cycles.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of Half-Cycles | WO₃ Thickness (Å) |\\n|---|---|\\n| 39 | 21.5 |\\n| 40 | 28.93 |\\n| 41 | 21.29 |\\n| 42 | 28.46 |\\n| 43 | 20.58 |\\n| 44 | 27.57 |\\n| 45 | 20.52 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The increase in thickness comes from the fact that the top layer is transformed into WO3, so added to the top.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"In this case the average of a half cycle is negative, indicating that there is more WO3 removed than added every ALE cycle. If the W film would be infinitely thick, this would result in no complete etch, so in the ideal case there is no slope.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"20.5 angstrom\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.33 angstrom per cycle\"}]}]","bbox":[{"panel_id":"a","x":3,"y":3,"width":665,"height":560}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/figure_13.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/figure_13.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/WO3 and W Thermal Atomic Layer Etching Using Conversion-Fluorination.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"25","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":672,"height":567,"image_format":"jpeg","image_sha256":"7f6324f7d779c8a9d76846d48ee80aa9d526ac49316d381947282df470575c25","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/test_atomic-layer-etching_experimental-usecase_25_figure_5.jpg","caption":"Figure 5.  $\\mathrm{B}_2\\mathrm{O}_3$  thickness versus number of HF exposures showing spontaneous etching of  $\\mathrm{B}_2\\mathrm{O}_3$  film at  $207^{\\circ}\\mathrm{C}$ . HF exposure was 100 mTorr s, and  $\\mathrm{B}_2\\mathrm{O}_3$  etch rate is  $\\sim 2\\mathrm{\\AA}$  per HF exposure.","id":"test/atomic-layer-etching/experimental-usecase/25/figure_5","sample_id":"atomic-layer-etching/experimental-usecase/25/figure_5","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart illustrates the thickness changes of B2O3 after HF exposures at 207°C. The blue squares represent measurements taken before the HF dose, while the red circles represent those taken after the HF dose.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of HF Exposures | Before HF Dose | After HF Dose |\\n|---|---|---|\\n| 1 | 540 | 538 |\\n| 2 | 538 | 536 |\\n| 3 | 536 | 534 |\\n| 4 | 534 | 532 |\\n| 5 | 532 | 530 |\\n| 6 | 530 | 530 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"2 angstrom.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Artificially grown, the B2O3 layer on top of the WO3 is not that thick as it is a result of the monolayer modification.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"53 nm.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":2,"width":665,"height":542}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/figure_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/figure_5.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/WO3 and W Thermal Atomic Layer Etching Using Conversion-Fluorination.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"25","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":672,"height":547,"image_format":"jpeg","image_sha256":"ac2cf95c675da486fd944e7a5d896360a8ab1eb623c367286518232a931861c6","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/test_atomic-layer-etching_experimental-usecase_25_figure_6.jpg","caption":"Figure 6.  $\\mathrm{WO}_3$  etch rate versus reactant exposure during  $\\mathrm{WO}_3$  ALE at  $207^{\\circ}\\mathrm{C}$ . (a)  $\\mathrm{BCl}_3$  exposure was varied with HF exposure held at 200 mTorr s. (b) HF exposure was varied with  $\\mathrm{BCl}_3$  exposure held at 327 mTorr s. Figure 7.  $\\mathrm{WO}_3$  thickness versus number of half-cycles during  $\\mathrm{WO}_3$  ALE at  $207^{\\circ}\\mathrm{C}$  under self-limiting conditions.","id":"test/atomic-layer-etching/experimental-usecase/25/figure_6","sample_id":"atomic-layer-etching/experimental-usecase/25/figure_6","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows the thickness of WO₃ after exposure to BCl₃ and HF doses at 207°C, plotted against the number of half-cycles.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of Half-Cycles | WO₃ Thickness (Å) |\\n|---|---|\\n| 5 | 97 |\\n| 10 | 84 |\\n| 15 | 75 |\\n| 20 | 65 |\\n| 25 | 55 |\\n| 30 | 48 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"This can be determined by calculating the slope and multiplying it with two, as the x-axis shows the number of half-cycles.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"207 degrees celcius.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"A technique that can be used to determine the thickness of thin films is ellipsometry.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Yes, the decrease in thickness is linear with respect to number of cycles.\"}]}]","bbox":[{"panel_id":"a","x":1,"y":3,"width":668,"height":540}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/figure_6.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/WO3 and W Thermal Atomic Layer Etching Using Conversion-Fluorination.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"25","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":672,"height":547,"image_format":"jpeg","image_sha256":"6e040408d19ed409a7800b52a9c70094617cf661b76962255add9be6cb1bc671","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/test_atomic-layer-etching_experimental-usecase_25_figure_8.jpg","caption":"Figure 8. Analysis of  $\\mathrm{WO}_3$  thickness change after  $\\mathrm{BCl}_3$  exposure and HF exposure during  $\\mathrm{WO}_3$  ALE at  $207^{\\circ}\\mathrm{C}$ .","id":"test/atomic-layer-etching/experimental-usecase/25/figure_8","sample_id":"atomic-layer-etching/experimental-usecase/25/figure_8","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot illustrates the thickness of WO₃ at various numbers of half-cycles, with specific arrows indicating etching, conversion, and removal processes.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of Half-Cycles | WO₃ Thickness (Å) |\\n|---|---|\\n| 9 | 88 |\\n| 10 | 86.5 |\\n| 11 | 84 |\\n| 12 | 82.5 |\\n| 13 | 80 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The full etch step is composed of the conversion of WO3 to B2O3 and the removal of this B2O3 top level.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"2.5 angstrom\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It should be, but the graph is misleading. It could be that HF is removed upon HF exposure, but also WO3 is removed according to the y-axis.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"around 4 angstrom per cycle\"}]}]","bbox":[{"panel_id":"a","x":3,"y":4,"width":670,"height":554}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/figure_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/figure_8.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/WO3 and W Thermal Atomic Layer Etching Using Conversion-Fluorination.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"25","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":675,"height":561,"image_format":"jpeg","image_sha256":"b8dfd236a9ff4cecf3a994801ae2bb9d5af8b752fab6ef4478cd00153bd43523","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/test_atomic-layer-etching_experimental-usecase_25_figure_9.jpg","caption":"Figure 9.  $\\mathrm{WO}_3$  thickness versus number of cycles for  $\\mathrm{WO}_3$  ALE at 128, 160, 196, and  $207^{\\circ}\\mathrm{C}$ .","id":"test/atomic-layer-etching/experimental-usecase/25/figure_9","sample_id":"atomic-layer-etching/experimental-usecase/25/figure_9","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows the thickness of WO₃ as a function of the number of cycles at different temperatures (128°C, 160°C, 196°C, and 207°C). The thickness decreases linearly with increasing cycle number for each temperature.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of Cycles | 128°C | 160°C | 196°C | 207°C |\\n|------------------|-------|-------|-------|-------|\\n| 0               | 140   | 140   | 140   | 140   |\\n| 5               | 138   | 130   | 128   | 120   |\\n| 10              | 136    | 120    | 116    | 100    |\\n| 15              | 132    | 110    | 102    | 80    |\\n| 20              | 130    | 100    | 90    | 60    |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The highest temperature will also result in the highest EPC, this can be concluded based on the slopes of the different lines.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"10 angstrom per cycle.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"This will be determined by the height of the EPC. This is the lowest for the process at 128 degrees. This process will then also result in the most control.\"}]}]","bbox":[{"panel_id":"a","x":2,"y":4,"width":665,"height":543}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/figure_9.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/figure_9.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/WO3 and W Thermal Atomic Layer Etching Using Conversion-Fluorination.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"25","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":670,"height":550,"image_format":"jpeg","image_sha256":"7c5f62693b78ca93a560d65ce09273b82e0351663bda5404e3dbee448e8f5a87","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/test_atomic-layer-etching_experimental-usecase_27_fig_1.jpg","caption":"Fig. 1. The etch depth of  $\\mathrm{Al}_2\\mathrm{O}_3$  per cycle (A/cycle) by Ar sputtering and ALET, measured as a function of 1st grid voltage from 50 to  $150\\mathrm{V}$  for the  $\\mathrm{Ar}^+$  ion beam source.  $\\mathrm{BCl}_3$  gas flow rate of  $100~\\mathrm{sccm}$  was used during the 1st adsorption step.","id":"test/atomic-layer-etching/experimental-usecase/27/fig_1","sample_id":"atomic-layer-etching/experimental-usecase/27/fig_1","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart illustrates the etch rate (in A/cycle) at various first grid voltages (in V) for both conditions: without BCl<sub>3</sub> adsorption and with BCl<sub>3</sub> adsorption.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| 1<sup>st</sup> Grid Voltage (V) | Etch Rate (A / cycle) without BCl<sub>3</sub> adsorption | Etch Rate (A / cycle) with BCl<sub>3</sub> adsorption |\\n|---|---|---|\\n| 50 | 0.0 | 0.0 |\\n| 75 | 0.0 | 0.25 |\\n| 100 | 0.0 | 1.0 |\\n| 125 | 0.5 | 1.25 |\\n| 150 | 1.0 | 1.75 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"1.0 Å/cycle.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0 Å/cycle.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"1.0 Å/cycle.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Without BCl3 adsorption, material is only being etched at grid voltages higher than 105 V, which corresponds to the Al2O3 substrate itself being etched. With BCl3 adsorption, an increase in etch rate is observed from 0 to 95 V. This means that etch rate is constant between 95 and 105 V. In this small region, only the chemisorbed layer (which is created during BCl3 exposure) is etched, meaning that the process meets the self-limiting reaction requirement between 95 and 105 V, so future experiments will be operated in this range.\"}]}]","bbox":[{"panel_id":"a","x":6,"y":4,"width":589,"height":444}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/27/images/fig_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/27/images/fig_1.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/27/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/27/Atomic layer etching of Al2O3 using BCl3Ar.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"27","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":597,"height":447,"image_format":"jpeg","image_sha256":"4d4c94e9fcc1328752117607baa164e5102ed96ed8bb3472866ae55e202748dc","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/test_atomic-layer-etching_experimental-usecase_31_fig_2.jpg","caption":"Fig. 2. Etch rates for  $\\mathrm{HfO_2}$ ,  $\\mathrm{ZrO_2}$ , and  $\\mathrm{Al_2O_3}$  films with respect to (a) substrate temperature, (b) RF bias power, (c) Hhfac pulse time, and (d) O radical pulse time.","id":"test/atomic-layer-etching/experimental-usecase/31/fig_2","sample_id":"atomic-layer-etching/experimental-usecase/31/fig_2","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"},{"panel_id":"b","label":"multiple scatter plot"},{"panel_id":"c","label":"multiple scatter plot"},{"panel_id":"d","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This multi-panel figure evaluates etch per cycle (EPC) for HfO₂, ZrO₂, and Al₂O₃ across four experimental variables relevant to ALE optimisation. Panel (a) shows that EPC increases with substrate temperature, particularly for HfO₂, which reaches >3 Å/cycle by 400 °C, while Al₂O₃ remains inert across the temperature range. Panel (b) reveals an \\\"ALE window\\\" for HfO₂ etching as RF bias power increases, with ZrO₂ showing moderate reactivity and Al₂O₃ remaining largely unaffected. Panels (c) and (d) display saturation-like behavior for Hhfac exposure and O radical exposure, respectively, again showing clear selectivity of the process for HfO₂ and ZrO₂ over Al₂O₃. These trends confirm process controllability and material selectivity critical for high-precision ALE.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Temperature (°C) | HfO₂ (Å/cycle) | ZrO₂ (Å/cycle) | Al₂O₃ (Å/cycle) |\\n|------------------|----------------|----------------|-----------------|\\n| 100              | 0.5            | 0.2            | 0.0             |\\n| 150              | 1.0            | 0.3            | 0.0             |\\n| 200              | 1.5            | 0.5            | 0.0             |\\n| 250              | 2.0            | 0.7            | 0.0             |\\n| 300              | 2.5            | 0.8            | 0.0             |\\n| 350              | 3.0            | 0.9            | 0.0             |\\n| 400              | 3.2            | 1.0            | 0.0             |\"},{\"panel_id\":\"b\",\"text\":\"| RF Bias Power (W) | HfO₂ (Å/cycle) | ZrO₂ (Å/cycle) | Al₂O₃ (Å/cycle) |\\n|-------------------|----------------|----------------|-----------------|\\n| 0                 | 0.5            | 0.2            | 0.0             |\\n| 50                | 1.0            | 0.5            | 0.0             |\\n| 100               | 2.0            | 0.8            | 0.0             |\\n| 150               | 3.0            | 1.0            | 0.0             |\\n| 200               | 3.5            | 1.2            | 0.0             |\\n| 250               | 4.0            | 1.5            | 0.0             |\"},{\"panel_id\":\"c\",\"text\":\"| Hhfac Exposure (sec) | HfO₂ (Å/cycle) | ZrO₂ (Å/cycle) | Al₂O₃ (Å/cycle) |\\n|----------------------|----------------|----------------|-----------------|\\n| 0                    | 0.0            | 0.0            | 0.0             |\\n| 2                    | 1.0            | 0.5            | 0.0             |\\n| 4                    | 2.0            | 0.8            | 0.0             |\\n| 6                    | 2.8            | 1.0            | 0.0             |\\n| 8                    | 3.0            | 1.2            | 0.0             |\\n| 10                   | 3.1            | 1.3            | 0.0             |\"},{\"panel_id\":\"d\",\"text\":\"| O Radical Exposure (sec) | HfO₂ (Å/cycle) | ZrO₂ (Å/cycle) | Al₂O₃ (Å/cycle) |\\n|--------------------------|----------------|----------------|-----------------|\\n| 0                        | 0.0            | 0.0            | 0.0             |\\n| 5                        | 1.5            | 0.7            | 0.0             |\\n| 10                       | 2.5            | 1.0            | 0.1             |\\n| 15                       | 3.0            | 1.2            | 0.1             |\\n| 20                       | 3.2            | 1.3            | 0.1             |\\n| 25                       | 3.3            | 1.4            | 0.1             |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"EPC increases significantly for HfO₂ and to a lesser extent for ZrO₂ as temperature, RF power, Hhfac exposure, and O radical exposure increase, indicating that both materials are etched efficiently. In contrast, Al₂O₃ remains nearly unaffected under all conditions, suggesting excellent selectivity. The trends indicate that the ALE process can be tuned to target high-k materials like HfO₂ and ZrO₂ while preserving low-k or etch-resistant materials like Al₂O₃, which is valuable for precise pattern transfer in multilayer device architectures\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"HfO₂.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Temperature, RF bias power, Hhfac exposure time, O radical exposure time.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":41,"y":20,"width":449,"height":585},{"panel_id":"b","x":560,"y":20,"width":453,"height":578},{"panel_id":"c","x":41,"y":666,"width":449,"height":570},{"panel_id":"d","x":562,"y":666,"width":451,"height":568}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/31/images/fig_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/31/images/fig_2.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/31/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/31/Atomic layer etching of high-k oxide thin films.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"31","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":1014,"height":1239,"image_format":"jpeg","image_sha256":"0eb3a2d08e868e59bf4a10836b9efaa3040505e07ff478899d3d68d69efefc70","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/test_atomic-layer-etching_experimental-usecase_31_fig_6.jpg","caption":"Fig. 6. Thickness variation of the  $\\mathrm{HfO_2}$  film as a function of ALE cycles under repeated injections of Hhfac gas/  $\\mathrm{O_2}$  plasma and Hhfac plasma/  $\\mathrm{O_2}$  plasma at a substrate temperature of  $100^{\\circ}\\mathrm{C}$  CCP discharges at a frequency of  $13.56\\mathrm{MHz}$  are applied via an RF power generator.","id":"test/atomic-layer-etching/experimental-usecase/31/fig_6","sample_id":"atomic-layer-etching/experimental-usecase/31/fig_6","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure compares the effect of using Hhfac gas versus Hhfac plasma in combination with O₂ plasma on the etching behavior of HfO₂ at 100 °C. When Hhfac gas is used, there is no measurable reduction in HfO₂ thickness over 50 cycles, indicating zero etch rate. In contrast, using Hhfac plasma results in a clear linear reduction in film thickness, with an etch rate of 1.49 Å/cycle. This difference highlights the critical role of precursor activation (via plasma) in enabling efficient etching under mild thermal conditions.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of Process Cycles | HfO₂ Thickness (nm) – Hhfac gas + O₂ plasma | HfO₂ Thickness (nm) – Hhfac plasma + O₂ plasma |\\n|--------------------------|---------------------------------------------|------------------------------------------------|\\n| 0                        | 15.0                                        | 15.0                                           |\\n| 10                       | 15.0                                        | 13.5                                           |\\n| 20                       | 15.0                                        | 12.0                                           |\\n| 30                       | 15.0                                        | 10.5                                           |\\n| 40                       | 15.0                                        | 9.0                                            |\\n| 50                       | 15.0                                        | 7.5                                            |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The use of Hhfac plasma in combination with O₂ plasma leads to a significant reduction in HfO₂ film thickness, demonstrating an etch rate of 1.49 Å/cycle. In contrast, using Hhfac in gas form under the same conditions results in no measurable thickness change, corresponding to an etch rate of 0 Å/cycle. This comparison highlights that plasma activation of the precursor is essential to initiate etching at low temperatures, likely due to enhanced surface reactivity and fragmentation of the Hhfac ligand under plasma conditions.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"1.49 Å/cycle.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Hhfac gas + O₂ plasma, Hhfac plasma + O₂ plasma\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":589,"height":605}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/31/images/fig_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/31/images/fig_6.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/31/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/31/Atomic layer etching of high-k oxide thin films.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"31","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":589,"height":605,"image_format":"jpeg","image_sha256":"1fe7569fe960f2538553a71676e8e18a2ddbb9cdeac58848eaf20a05de487b47","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/test_atomic-layer-etching_experimental-usecase_37_fig_4.jpg","caption":"FIG. 4. The thickness of the  $\\mathrm{TiO_2}$  layer on Si-OH and Si-H surfaces for  $90\\mathrm{ms}$ ,  $180\\mathrm{ms}$ , and  $270\\mathrm{ms}$ $\\mathrm{TiCl_4 / H_2O}$  dosing times as measured by in situ ellipsometry.","id":"test/atomic-layer-etching/experimental-usecase/37/fig_4","sample_id":"atomic-layer-etching/experimental-usecase/37/fig_4","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows the thickness of SiOH and SiH layers as a function of cycles, with different exposure times of TiCl4, Ar and H2O.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Cycle | Thickness (nm) SiOH 0.09 / 45 / 0.09 / 45 s | Thickness (nm) SiOH 0.18 / 45 / 0.18 / 45 s | Thickness (nm) SiOH 0.27 / 45 / 0.27 / 45 s | Thickness (nm) SiH 0.09 / 45 / 0.09 / 45 s | Thickness (nm) SiH 0.18 / 45 / 0.18 / 45 s | Thickness (nm) SiH 0.27 / 45 / 0.27 / 45 s |\\n|---|---|---|---|---|---|---|\\n| 0 | 0 | 0 | 0 | 0 | 0 | 0 |\\n| 25 | 1 | 1 | 1 | 0.1 | 0.1 | 0.1 |\\n| 50 |  2 | 2 | 2 | 0.5 | 0.5 | 0.5 |\\n| 75 | 3.5 | 3.5 | 3.5 | 1 | 1 | 1 |\\n| 100 | 5 | 5 | 5 | 2.5 | 2.5 | 2.5 |\\n| 125 | 6.5 | 6.5 | 6.5 | 3.5 | 3.5 | 3.5 |\\n| 150 | 7.5 | 7.5 | 7.5 | 5.5 | 5.5 | 5.5 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"SiO2 and Si substrates, with Si-OH and Si-H surface groups, respectively.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Roughly 5 nm for all three experimental setups.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Roughly 2 nm for all three experimental setups.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"On Si-H substrates, all dosing conditions exhibit a nucleation delay of approximately 25 cycles, followed by an acceleration to a steady growth rate of about 0.05 nm per cycle. On Si–OH substrates, the three dosing conditions also produce nearly identical growth behaviour, with approximately 8 nm of TiO₂ deposited after 150 cycles, and the lowest dosing condition yielding slightly lower thickness consistent with near-saturation dosing. Applying the selectivity definition shows that a selectivity value of 0.9 corresponds to a TiO₂ thickness of about 1.5 nm on the growth surface for all dosing conditions. Overall, the figure demonstrates that near- and over-saturated H₂O dosing does not significantly alter the nucleation delay on Si–H surfaces nor the achievable selectivity window.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":672,"height":572}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/37/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/37/images/fig_4.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/37/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/37/Effect of reactant dosing on selectivity during area-selective.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"37","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":672,"height":572,"image_format":"jpeg","image_sha256":"e5bf2bdf456ac75e5d7bf80c22baf8e3d36e59f181f57317c7d66a7b68cd0e24","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/test_atomic-layer-etching_experimental-usecase_37_fig_5.jpg","caption":"FIG. 5. The thickness of the  $\\mathrm{TiO_2}$  layer on  $\\mathrm{Si - H}$  as a function of the number of super-cycles using  $\\mathrm{TiCl_4 / H_2O}$  dosing times of 90, 180, and  $270~\\mathrm{ms}$  as measured by in situ ellipsometry. Data are collected after every five ALD cycles and after each individual ALE cycle. The data for 180 and  $270~\\mathrm{ms}$  is offset on the  $y$  axis by  $1\\mathrm{nm}$  and  $2\\mathrm{nm}$ , respectively, to improve visualization.","id":"test/atomic-layer-etching/experimental-usecase/37/fig_5","sample_id":"atomic-layer-etching/experimental-usecase/37/fig_5","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart shows the thickness of the deposited TiO2 layer as a function of super-cycles (combined ALD/ALE cycles). Three data series are plotted with varying TiCl4 and H2O exposure times (0.09, 0.18 and 0.27 for both, respectively).\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Super-cycles | Thickness (nm) 0.09 / 45 / 0.09 / 45 s | Thickness (nm) 0.18 / 45 / 0.18 / 45 s | Thickness (nm) 0.27 / 45 / 0.27 / 45 s |\\n|---|---|---|---|\\n| 0 | 0 | 1 | 2 |\\n| 5 | 0.25 | 1.5 | 2.5 |\\n| 10 | 0.25 | 1.5 | 5.5 |\\n| 15 | 0.5 | 1.5 |  |\\n| 20 | 0.5 | 1.5 |  |\\n| 25 | 0.5 |  |  |\\n| 30 | 0.75 |  |  |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"About 0.25 nm.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"About 1.5 nm (0.5 nm when taking into account the offset).\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Roughly 5.5 nm (3.5 nm when taking into account the offset).\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"During the initial three to five super-cycles, all dosing conditions exhibit similar alternating deposition and etching behaviour, indicating comparable early-stage nucleation resistance on the non-growth surface. Beyond approximately five super-cycles, however, the longest dosing condition (270 ms for both TiCl₄ and H₂O) shows a marked increase in net film accumulation after the ALE steps, signifying more rapid unwanted nucleation when longer ALD doses are combined with ALE. This contrasts with pure ALD behaviour, where increased dose times did not significantly affect nucleation on Si–H, and is attributed to ALE-induced removal of passivating Si–H bonds that promotes formation of Si–O sites favourable for TiO₂ growth. In addition, the figure indicates that the amount of material removed per ALE cycle decreases over successive super-cycles for all dosing conditions, leaving a thin residual layer after etching; this residual thickness, on the order of ~0.2 nm after extended cycling, is interpreted as arising from surface oxidation and formation of non-etchable species rather than incomplete removal of TiO₂.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":669,"height":492}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/37/images/fig_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/37/images/fig_5.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/37/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/37/Effect of reactant dosing on selectivity during area-selective.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"37","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":669,"height":492,"image_format":"jpeg","image_sha256":"aec23ed3fbcf9197aba419647dbdbc8156590bc08218f2b708ffa7a9a03377a3","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/test_atomic-layer-etching_experimental-usecase_37_fig_7.jpg","caption":"FIG. 7. The thickness of the  $\\mathrm{TiO_2}$  layer on  $\\mathrm{Si - H}$  for  $\\mathrm{TiCl_4 / H_2O}$  90/270 ms and 270/90 ms dosing times as measured by in situ ellipsometry. Data are shown for 15 super-cycles, consisting of 30 ALD and 5 ALE cycles each, with data points taken every five cycles for ALD and every cycle for ALE. The data for 90/270 ms are offset on the y axis by 1 nm to improve visualization.","id":"test/atomic-layer-etching/experimental-usecase/37/fig_7","sample_id":"atomic-layer-etching/experimental-usecase/37/fig_7","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows the thickness of two TiO2 layers over super-cycles. Both layers are grown on Si-H and are grown at two different precursor exposure times (90/270 ms for TiCl4/H2O, and 270/90 ms for TiCl4/H2O, respectively).\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Super-Cycles | Thickness (nm) 0.09 / 45 / 0.27 / 45 s | Thickness (nm) 0.27 / 45 / 0.09 / 45 s |\\n|---|---|---|\\n| 0 | 1.0 | 0 |\\n| 3 | 1.25 | 0.25 |\\n| 6 | 1.5 | 0.25 |\\n| 9 | 1.5 | 0.25 |\\n| 12 | 2.0 | 0.5 |\\n| 15 | 2.5 | 0.5 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Si-H.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"30 ALD and 5 ALE cycles.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"They differ in exposure time of the precursors during ALD (either 90 ms TiCl4 and 270 ms H2O, or 270 ms TiCl4 and 90 ms H2O).\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"15.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":669,"height":555}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/37/images/fig_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/37/images/fig_7.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/37/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/37/Effect of reactant dosing on selectivity during area-selective.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"37","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":669,"height":555,"image_format":"jpeg","image_sha256":"3c1aca0c1d2b20af4731d9cb195327a02c78c5dd425d33800f2d86ccbaa53249","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/test_atomic-layer-etching_experimental-usecase_44_fig_2.jpg","caption":"FIG. 2. (Color online) Film composition vs sputter time obtained by XPS depth-profiling analysis. The initial AlN film on the Si(111) wafer had a thickness of  $\\sim 500\\mathrm{\\AA}$","id":"test/atomic-layer-etching/experimental-usecase/44/fig_2","sample_id":"atomic-layer-etching/experimental-usecase/44/fig_2","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The multiple line chart represents the film composition of (Al, C, N, O, Si) agains Sputtering time (s) obtained by XPS depth-profiling analysis.  This image indicates that there is a large concentration of oxygen on the surface that decays as a function of sputtering time.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Composition (at.%) | Sputtering Time (s) | Material |\\n|--------------------|---------------------|----------|\\n| 0    | 40 | Al |\\n| 0    | 10 | C  |\\n| 0    | 30 | N  |\\n| 0    | 17 | O  |\\n| 0    | 0  | Si |\\n| 1000 | 57 | Al |\\n| 1000 | 0  | C  |\\n| 1000 | 40 | N  |\\n| 1000 | 2  | O  |\\n| 1000 | 0  | Si |\\n| 2000 | 40 | Al |\\n| 2000 | 0  | C  |\\n| 2000 | 40 | N  |\\n| 2000 | 2  | O  |\\n| 2000 | 0  | Si |\\n| 3000 | 58 | Al |\\n| 3000 | 0  | C  |\\n| 3000 | 40 | N  |\\n| 3000 | 2  | O  |\\n| 3000 | 0  | Si |\\n| 4000 | 56 | Al |\\n| 4000 | 0  | C  |\\n| 4000 | 40 | N  |\\n| 4000 | 1  | O  |\\n| 4000 | 1  | Si |\\n| 5000 | 13 | Al |\\n| 5000 | 0  | C  |\\n| 5000 | 19 | N  |\\n| 5000 | 0  | O  |\\n| 5000 | 64 | Si |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Surface Contamination: High Carbon (C) and Oxygen (O).\\n\\nAlOₓNᵧ Layer: High O and Al, with some N.\\n\\nBulk AlN Film: High Al and N, low O, with Al > N (Al-rich).\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The Oxygen (O) concentration is high at the surface and decreases with depth, while Nitrogen (N) is low at the surface and increases inward. This reveals the competition between a surface oxidation route (forming AlOₓ) and the intended nitride formation route (forming AlN). The dominance of the oxidation route at the surface creates the AlOₓNᵧ layer, which must be accounted for as an etching barrier before accessing the target AlN film.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"It may lead to a slightly different etch rate compared to perfectly stoichiometric AlN, as the reaction with HF could form a different AlFₓ surface layer.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Surface carbon acts as a contamination barrier that can block the reactant pulses (HF, Sn(acac)₂) from reaching the underlying AlN. This would cause poor etch initiation, non-uniformity, and inconsistent cycle counts. Therefore, an in-situ or pre-etch cleaning step is essential to remove carbon and ensure reproducible ALE.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":3,"width":659,"height":550}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/44/images/fig_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/44/images/fig_2.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/44/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/44/Thermal atomic layer etching of crystalline aluminum nitride.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"44","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":664,"height":553,"image_format":"jpeg","image_sha256":"7ecae6a04743fe0996bb16dafede71e714458ed80909f7554fe434d0306e335b","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/test_atomic-layer-etching_experimental-usecase_47_FIG10.jpg","caption":"FIG. 10. Resistivities of unetched and etched molybdenum films as a function of film thickness. Resistivity values reported by Kim et al. (Ref. 32) are included for reference.","id":"test/atomic-layer-etching/experimental-usecase/47/FIG10","sample_id":"atomic-layer-etching/experimental-usecase/47/FIG10","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The image plots the resistivity  as a function of thickness for samples before and after etching.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Thickness (nm) | Resistivity (µΩ·cm) | Condition        |\\n|---------------|----------------------|------------------|\\n| ~5           | 33                   | Kim et al.      |\\n| ~10          | 25                   | After etching   |\\n| ~12          | 17                   | After etching   |\\n| ~15          | 14                   | After etching   |\\n| ~15          | 13                   | Kim et al.      |\\n| ~18          | 12                   | Before etching  |\\n| ~20          | 12                   | After etching   |\\n| ~25          | 11                   | After etching   |\\n| ~30          | 10                   | After etching   |\\n| ~35          | 9                    | After etching   |\\n| ~40          | 8                    | Before etching  |\\n| ~45          | 8                    | After etching   |\\n| ~50          | 7                    | Before etching  |\\n| ~55          | 7                    | Before etching  |\\n| ~60          | 7                    | Before etching  |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Sheet resistances of the films were measured with the fourpoint probe at nine points in a rectangular array along the film surface and the average values were used for calculating resistivities of the films by multiplying the average sheet resistances by the film thicknesses\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The etching itself did not seem to affect the resistivity, despite the grain boundaries becoming more distinct. The resistivities of the etched and as-deposited films follow the same trend down to about 20 nm. There were no as-deposited films with thicknesses smaller than 20 nm to serve as a reference, but values from a recent study of ALD molybdenum are included in the graph\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Bulk resistivity of molybdenum is reported to be 5.34 μΩ cm, whereas the films in the present study had at the lowest resistivity of about 8 μΩ cm. In a study by van der Zouw et al., ALD grown molybdenum films showed increased resistivity below 20 nm thickness, and a sharp increase when the thickness was approaching the mean free path of electrons in Mo (11.2 nm). The record low resistivity of 13 μΩ cm at 10 nm thickness claimed by Kim et al. is\\nonly slightly lower than seen here. Based on this, it seems that the increased resistivity is mostly caused by thin film effects rather than the etching.\"}]}]","bbox":[{"panel_id":"a","x":2,"y":2,"width":673,"height":576}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/47/images/FIG10.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/47/images/FIG10.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/47/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/47/Atomic layer etching of molybdenum with O2 and NbCl5.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"47","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":680,"height":584,"image_format":"jpeg","image_sha256":"2e1218134ffcb0a7ae5c56fc3b9e21e8b4dc9d41d26ac106d94014f09b8d9264","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/test_atomic-layer-etching_experimental-usecase_47_FIG5_b.jpg","caption":"FIG. 5. (a) EPC as a function of temperature using cycles of  $4\\mathrm{~s~}\\mathrm{NbCl}_5$  and  $3\\mathrm{~s~}\\mathrm{O}_2$  pulses followed by  $1\\mathrm{~s~}$  purges. (b) EPC as a function of the  $\\mathrm{O}_2$  pulse length at different temperatures using  $4\\mathrm{~s~}\\mathrm{NbCl}_5$  pulses and  $1\\mathrm{~s~}$  purges.","id":"test/atomic-layer-etching/experimental-usecase/47/FIG5_b","sample_id":"atomic-layer-etching/experimental-usecase/47/FIG5_b","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure plots EPC against oxygen pulse length for different temperatures.\"}]","data_extraction":"[{\"panel_id\":\"b\",\"text\":\"| O₂ pulse length (s) | 400 °C | 350 °C | 300 °C | 250 °C |\\n|---|---|---|---|---|\\n| 1 | 3.0 | 1.5 | 0.6 | 0.1 |\\n| 2 | 3.5 | 2.0 | 0.8 | 0.2 |\\n| 3 | 4.0 | 2.5 | 1.2 | 0.3 |\\n| 6 | 4.5 | 2.8 | 1.6 | 0.4 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The oxidation step was further studied at etching temperatures of 250–400 °C by varying the O2 pulse length\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The EPC does not fully saturate at any of the temperatures, but the growth of EPC slows down gradually. This is most likely due to the forming surface oxide layer acting as a diffusion barrier that inhibits further oxidation\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Oxidation of metals at moderate temperatures generally follows a parabolic rate law. The EPC values seen here are somewhat lower than what would be expected for the parabolic rate law, but at the higher end of the O2 pulse lengths, the EPC might be limited by the dose of NbCl5.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":4,"width":671,"height":571}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/47/images/FIG5_b.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/47/images/FIG5_b.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/47/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/47/Atomic layer etching of molybdenum with O2 and NbCl5.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"47","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:image_caption"},"width":672,"height":575,"image_format":"jpeg","image_sha256":"b5efc30202c887efeb949c94d8bcdb14b9f44869a21f8dcef52f15904626055b","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/test_atomic-layer-etching_simulation-usecase_12_figure_8.jpg","caption":"Figure 8. (a) Scatter plot showing the coverage of total HF adsorbed vs the Al-F coverage in the minima shown in Figure 7. The data points are colored according to the corresponding total binding energy per square nanometer (color plot online). The square markers along the diagonal dashed line indicate geometries where the HF coverage is equal to the Al-F coverage. Panels b and c show the change in binding energy per square nanometer with an increase in HF and Al-F coverage, respectively.","id":"test/atomic-layer-etching/simulation-usecase/12/figure_8","sample_id":"atomic-layer-etching/simulation-usecase/12/figure_8","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"},{"panel_id":"b","label":"multiple scatter plot"},{"panel_id":"c","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The subfigure compares total HF coverage with Al–F coverage across the adsorption minima from Figure 7. Most data points lie below the dashed diagonal, indicating that increasing HF adsorption does not translate one-to-one into Al–F bond formation, especially at higher coverages.\"},{\"panel_id\":\"b\",\"text\":\"The subfigure shows that the total binding energy per unit area becomes progressively more negative with increasing HF coverage, reflecting stronger overall adsorption as surface loading increases.\"},{\"panel_id\":\"c\",\"text\":\"The subfigure shows a similar trend with Al–F coverage: higher densities of Al–F bonds correlate with more negative binding energies, with the strongest stabilization occurring at high Al–F coverages.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Coverage HF (nm⁻²) | Coverage Al–F (nm⁻²) |\\n|-------------------|----------------------|\\n| ~1 | ~1 |\\n| ~3 | ~1 |\\n| ~5 | ~3 |\\n| ~7 | ~4 |\\n| ~9 | ~5 |\\n| ~10 | ~5 |\\n| ~14 | ~9 |\\n| ~15 | ~7 |\"},{\"panel_id\":\"b\",\"text\":\"| Coverage HF (nm⁻²) | E_bind (eV/nm²) |\\n|-------------------|----------------|\\n| ~1 | ~ -2 |\\n| ~3 | ~ -3 |\\n| ~5 | ~ -8 |\\n| ~7 | ~ -11 |\\n| ~9 | ~ -12 |\\n| ~10 | ~ -14 |\\n| ~13 | ~ -15 |\\n| ~13 | ~ -14 |\\n| ~15 | ~ -16 |\"},{\"panel_id\":\"c\",\"text\":\"| Coverage Al–F (nm⁻²) | E_bind (eV/nm²) |\\n|---------------------|----------------|\\n| ~1 | ~ -3 |\\n| ~3 | ~ -7 |\\n| ~3 | ~ -10 |\\n| ~5 | ~ -12--16 |\\n| ~7 | ~ -16 |\\n| ~10 | ~ -15 |\\n| ~14 | ~ -16 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Al–F coverage increases with HF coverage but remains consistently lower for most configurations. This indicates that a significant fraction of adsorbed HF remains molecular or participates in hydrogen-bonded networks rather than forming Al–F bonds.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 14–15 nm⁻².\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The binding energy becomes more negative.\"}]}]","bbox":[{"panel_id":"a","x":15,"y":26,"width":418,"height":388},{"panel_id":"b","x":439,"y":36,"width":400,"height":378},{"panel_id":"c","x":841,"y":26,"width":566,"height":385}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/12/images/figure_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/12/images/figure_8.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/12/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/12/Modeling the Chemical Mechanism of the Thermal Atomic Layer Etch of Aluminum Oxide.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"12","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":1409,"height":417,"image_format":"jpeg","image_sha256":"033f91c74bf30aadcaea84365b2fc8f4a2e8fa5165ef13f8d275f8db59deac1b","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/test_atomic-layer-etching_simulation-usecase_14_fig_3.jpg","caption":"Fig. 3. Total etched depths after each cycle of the SiN PE-ALE processes with  $\\mathrm{Ar^{+}}$ $\\mathrm{Kr^{+}}$  and  $\\mathrm{Xe^{+}}$  ion irradiations, evaluated from the MD simulations of Fig. 2. The EPC averaged over the five cycles are 0.41, 0.17, and  $0.08\\mathrm{nm}$  for the PE-ALE processes with  $\\mathrm{Ar^{+}}$ $\\mathrm{Kr^{+}}$  and  $\\mathrm{Xe^{+}}$  ion irradiations, respectively.","id":"test/atomic-layer-etching/simulation-usecase/14/fig_3","sample_id":"atomic-layer-etching/simulation-usecase/14/fig_3","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure presents total etched depth as a function of processing cycles for three noble gas ions: argon (Ar), krypton (Kr), and xenon (Xe). Ar demonstrates a steep and consistent increase in etch depth across five cycles, indicating its superior etching efficacy. Kr follows a moderate linear trend, while Xe shows the least depth gain and plateaus after Cycle 3. This suggests that Ar is most effective at enhancing etch performance in this context, while Xe may saturate or be less reactive under the same conditions\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Cycle     | Ar (nm) | Kr (nm) | Xe (nm) |\\n|-----------|---------|---------|---------|\\n| Initial   | 0.0     | 0.0     | 0.0     |\\n| Cycle 1   | 0.1     | 0.1     | 0.1     |\\n| Cycle 2   | 0.3     | 0.2     | 0.2     |\\n| Cycle 3   | 0.6     | 0.4     | 0.3     |\\n| Cycle 4   | 1.2     | 0.5     | 0.4     |\\n| Cycle 5   | 2.1     | 0.8     | 0.4     |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The figure evaluates the total etched depth resulting from processing with noble gases Ar, Kr, and Xe over five cycles.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Argon (Ar): highest , Krypton (Kr): moderate , Xenon (Xe): minimal increase\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The steadily increasing etched depth for Ar across cycles suggests it is the most reactive or energetic etchant in this setup, enabling efficient material removal. Kr demonstrates moderate etching capacity with diminishing returns after Cycle 4, possibly due to saturation or lower energy transfer. Xe's depth stabilises early, indicating limited reactivity or surface saturation.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":675,"height":553}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/14/images/fig_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/14/images/fig_3.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/14/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/14/Molecular dynamics simulations of silicon nitride.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"14","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":673,"height":553,"image_format":"jpeg","image_sha256":"887430939500f945b57d2ff1f899a19046dfa97d69312133d47f8d76162827a4","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/test_atomic-layer-etching_simulation-usecase_21_figure_6.jpg","caption":"Figure 6. Uniformity across a blanket silicon wafer before and after the ALE process. Reproduced with permission, copyright 2013.","id":"test/atomic-layer-etching/simulation-usecase/21/figure_6","sample_id":"atomic-layer-etching/simulation-usecase/21/figure_6","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The plot compares silicon thickness across the wafer before and after ALE. Thickness decreases uniformly from 87 nm to 40 nm, while the 3σ variation remains ±1.5 nm in both cases. The preservation of the same statistical spread indicates that material removal occurs evenly across the wafer. This demonstrates that ALE maintains wafer-scale uniformity despite substantial thickness reduction\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Condition   | Si thickness (nm) | Variation (3σ, nm) |\\n|-------------|------------------|--------------------|\\n| Before etch | 87               | ±1.5               |\\n| After ALE   | 40               | ±1.5               |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Uniformity is preserved because each ALE cycle consists of a self-limiting surface modification step followed by controlled, ion-activated removal. The saturation of the modification reaction ensures that all wafer regions experience the same chemically altered layer thickness, independent of local exposure history. Subsequent ion activation removes only this modified layer uniformly, suppressing spatial variations that typically arise in continuous plasma etching.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The thickness decreases from 87 nm before etching to 40 nm after ALE.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"-Etch response is largely insensitive to local thickness variations.\\n-Each region undergoes nearly identical cycle-by-cycle material removal.\\n-Topography-dependent microloading effects are minimized by self-limiting surface chemistry.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Uniform thickness ensures predictable alignment and interaction between stacked layers in complex device architectures. Variations in underlying film thickness can propagate into overlay errors or interface defects. ALE’s ability to preserve uniformity supports reliable multi-layer integration with tight dimensional control.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":4,"width":545,"height":385}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/21/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/21/images/figure_6.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/21/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/21/Review Paper -- Atomic Layer Etching Rethinking the Art of Etch.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"21","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":550,"height":391,"image_format":"jpeg","image_sha256":"3193f04e0c852e9582bb7dfe56f0e83c3e7161f782ebc70b782927a370191d53","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/test_atomic-layer-etching_simulation-usecase_25_figure_11.jpg","caption":"Figure 11. (a) Scatter plot for  $\\mathrm{Hf - F}$  coverage versus total HF coverage for the surface coverage values in Table 3. Note that some HF coverages resulted in partial or complete  $\\mathrm{Hf - F}$  coverage for different configurations, for example, the coverage of  $1.0\\mathrm{HF / nm^2}$ . The square data points are where the adsorbed HF coverage equals  $\\mathrm{Hf - F}$  coverage. The circular data points are where partially dissociated HF molecules are present. Plots (b) and (c) show the change in binding energy per square nanometer with an increase in HF and  $\\mathrm{Hf - F}$  coverage, respectively.","id":"test/atomic-layer-etching/simulation-usecase/25/figure_11","sample_id":"atomic-layer-etching/simulation-usecase/25/figure_11","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"},{"panel_id":"b","label":"scatter plot"},{"panel_id":"c","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Scatter plot comparing dissociated Hf–F coverage with total HF coverage. Square points lie on the dashed 1:1 line, indicating configurations where HF fully dissociates. Circular points fall below this line, showing partial dissociation at higher coverages and the onset of saturation behavior.\"},{\"panel_id\":\"b\",\"text\":\"Scatter plot showing that Ebind per unit area becomes increasingly negative as total HF coverage increases. The trend suggests stronger net binding with higher surface coverage, with some spread reflecting different adsorption configurations.\"},{\"panel_id\":\"c\",\"text\":\"Scatter plot showing that Ebind per unit area also decreases with increasing Hf–F coverage. Compared to (b), this representation highlights how the degree of dissociation directly correlates with surface stabilization.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Data Series | Total Coverage [HF/nm²] | Hf-F Coverage [Hf-F/nm²] |\\n|----------------|----------------------------------|-----------------------------------|\\n| Squares (on line) | 1.0 | 1.0 |\\n| Squares (on line) | 2.5 | 2.5 |\\n| Circles (deviation) | 2.0 | 1.5 |\\n| Circles (deviation) | 4.0 | 3.0 |\\n| Circles (deviation) | 6.0 | 5.0 |\\n| Circles (deviation) | 8.0 | 6.5 |\\n| Circles (saturation) | 15.0 | 6.0 |\\n| Circles (saturation) | 17.0 | 7.0 |\"},{\"panel_id\":\"b\",\"text\":\"| Coverage [HF/nm²] | Ebind [eV/nm²] | \\n|---|---| \\n| 1.0 | ~ -1.0 | \\n| 2.0 | ~ -3.0 | \\n| 4.0 | ~ -4.5 | \\n| 8.0 | ~ -8.0 | \\n| 8.0 | ~ -9.5 | \\n| 15.0 | ~ -12.0 | \\n| 17.0 | ~ -13.0 |\"},{\"panel_id\":\"c\",\"text\":\"| Coverage [Hf-F/nm²] | Ebind [eV/nm²] |\\n|---|---|\\n| 0.5 | ~ -1.0 |\\n| 2.0 | ~ -3.0 |\\n| 3.0 | ~ -4.5 |\\n| 5.8 | ~ -8.0 |\\n| 6.0 | ~ -12.0 |\\n| 7.0 | ~ -13.0 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Yes. The square points fall on the dashed 1:1 line, indicating complete HF dissociation.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The square points follow the 1:1 line, indicating configurations where all adsorbed HF dissociates to form Hf–F. In contrast, the circular points deviate below this line, especially at higher total HF coverages. This suggests that as coverage increases, not all HF molecules dissociate, likely due to site saturation or geometric constraints on the surface. The coexistence of fully and partially dissociated configurations indicates coverage-dependent limits on dissociation.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Hf–F coverage (per nm²).\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Ebind decreases (becomes more negative) with increasing coverage.\"}]}]","bbox":[{"panel_id":"a","x":4,"y":38,"width":274,"height":315},{"panel_id":"b","x":290,"y":37,"width":274,"height":316},{"panel_id":"c","x":579,"y":31,"width":265,"height":322}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/25/images/figure_11.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/25/images/figure_11.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/25/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/25/Self-Limiting Temperature Window for Thermal Atomic Layer Etching.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"25","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":847,"height":353,"image_format":"jpeg","image_sha256":"e86dda43ca9e5b059dfa4babe8832060d1de99f943e6f48a6aadd26129650566","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/test_atomic-layer-etching_simulation-usecase_25_figure_12.jpg","caption":"Figure 12. (a) Scatter plot for  $\\mathrm{Zr - F}$  coverage versus total HF coverage for the surface coverage values in Table 4. Note that some HF coverages resulted in partial or complete  $\\mathrm{Zr - F}$  coverage for different configurations, for example, the coverage of  $1.0\\mathrm{HF / nm^2}$ . The square data points are where the adsorbed HF coverage equals  $\\mathrm{Zr - F}$  coverage. The circular data points are where partially dissociated HF molecules are present. Plots (b) and (c) show the change in binding energy per square nanometer with an increase in HF and  $\\mathrm{Zr - F}$  coverage, respectively.","id":"test/atomic-layer-etching/simulation-usecase/25/figure_12","sample_id":"atomic-layer-etching/simulation-usecase/25/figure_12","subset":"multiple-scatter-plot","split":"test","classification":[{"panel_id":"a","label":"multiple scatter plot"},{"panel_id":"b","label":"scatter plot"},{"panel_id":"c","label":"scatter plot"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":6,"y":31,"width":278,"height":317},{"panel_id":"b","x":289,"y":30,"width":280,"height":316},{"panel_id":"c","x":573,"y":31,"width":273,"height":315}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/25/images/figure_12.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/25/images/figure_12.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/25/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/25/Self-Limiting Temperature Window for Thermal Atomic Layer Etching.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"25","first_classification_panel_id":"a","first_classification_label":"multiple scatter plot","caption_source":"content.json:img_caption"},"width":847,"height":347,"image_format":"jpeg","image_sha256":"c7e1490cdb1915ea35a621e0610ef9150d504e23781ee51515945264ed759e1a","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"}