{"file_name":"images/test_atomic-layer-deposition_experimental-usecase_1_figure_2.jpg","caption":"Figure 2. QCM measurements for  $\\mathrm{Al}_2\\mathrm{O}_3$  ALD at  $58^{\\circ}C$  showing the linear growth of the  $\\mathrm{Al}_2\\mathrm{O}_3$  ALD film over many reaction cycles. The average  $\\mathrm{Al}_2\\mathrm{O}_3$  mass gain per ALD cycle is  $30~\\mathrm{ng / cm^2}$ .","id":"test/atomic-layer-deposition/experimental-usecase/1/figure_2","sample_id":"atomic-layer-deposition/experimental-usecase/1/figure_2","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart shows a linear increase in mass over time for the cycle sequence (1/ 20/ 2/ 30s) at 58 °C, starting at 0 ng/cm² at 0 seconds and reaching approximately 720 ng/cm² at 1000 seconds.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Time (s) | Mass (ng/cm²) |\\n|---|---|\\n| 0 | 0,00 |\\n| 7 | 21,78 |\\n| 15 | 34,85 |\\n| 42 | 31,95 |\\n| 46 | 53,73 |\\n| 56 | 53,73 |\\n| 58 | 65,35 |\\n| 86 | 61,00 |\\n| 89 | 81,33 |\\n| 98 | 82,78 |\\n| 100 | 97,30 |\\n| 133 | 92,95 |\\n| 132 | 119,09 |\\n| 140 | 117,63 |\\n| 144 | 127,80 |\\n| 175 | 121,99 |\\n| 175 | 148,13 |\\n| 182 | 145,23 |\\n| 187 | 155,39 |\\n| 215 | 152,49 |\\n| 218 | 174,27 |\\n| 226 | 177,18 |\\n| 228 | 191,70 |\\n| 258 | 184,44 |\\n| 260 | 201,87 |\\n| 273 | 206,22 |\\n| 273 | 219,29 |\\n| 301 | 209,13 |\\n| 302 | 230,91 |\\n| 313 | 233,82 |\\n| 315 | 246,89 |\\n| 345 | 238,17 |\\n| 346 | 264,32 |\\n| 355 | 265,77 |\\n| 359 | 278,84 |\\n| 387 | 270,12 |\\n| 389 | 293,36 |\\n| 398 | 294,81 |\\n| 405 | 306,43 |\\n| 433 | 302,07 |\\n| 431 | 322,41 |\\n| 445 | 328,22 |\\n| 445 | 339,83 |\\n| 475 | 331,12 |\\n| 476 | 355,81 |\\n| 486 | 357,26 |\\n| 487 | 367,43 |\\n| 519 | 363,07 |\\n| 519 | 381,95 |\\n| 529 | 384,85 |\\n| 531 | 396,47 |\\n| 559 | 389,21 |\\n| 562 | 416,80 |\\n| 573 | 421,16 |\\n| 573 | 421,16 |\\n| 604 | 419,71 |\\n| 604 | 445,85 |\\n| 614 | 445,85 |\\n| 616 | 456,02 |\\n| 647 | 453,11 |\\n| 648 | 474,90 |\\n| 659 | 474,90 |\\n| 660 | 486,51 |\\n| 691 | 482,16 |\\n| 694 | 502,49 |\\n| 705 | 517,01 |\\n| 735 | 517,01 |\\n| 735 | 537,34 |\\n| 746 | 537,34 |\\n| 748 | 547,51 |\\n| 778 | 544,61 |\\n| 775 | 564,94 |\\n| 792 | 576,56 |\\n| 820 | 573,65 |\\n| 822 | 593,98 |\\n| 834 | 604,15 |\\n| 861 | 601,24 |\\n| 866 | 627,39 |\\n| 879 | 639,00 |\\n| 908 | 636,10 |\\n| 908 | 652,07 |\\n| 920 | 665,15 |\\n| 949 | 663,69 |\\n| 951 | 684,02 |\\n| 962 | 697,10 |\\n| 992 | 700,00 |\\n| 998 | 717,43 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The process is net deposition.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The film growth is linear.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The process shows stepwise growth.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The rate of deposition is constant. This is visually evident from the graph, where the mass increases along a straight  line. If the rate of deposition would increase, the trendline would be curving upward.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":589,"height":364}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_2.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":"line chart","caption_source":"content.json:img_caption"},"width":589,"height":364,"image_format":"jpeg","image_sha256":"bddee448af87923579afe43b35dc09cba825bdc0cbb20541d9c74dd4a256acbc","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_3.jpg","caption":"Figure 3. Detailed view of the QCM measurements at  $58^{\\circ}C$ . The shaded areas show the time periods during which the TMA and water dosing valves were open. The mass increases coincide with the TMA and water exposures.","id":"test/atomic-layer-deposition/experimental-usecase/1/figure_3","sample_id":"atomic-layer-deposition/experimental-usecase/1/figure_3","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart illustrates the mass (in ng/cm²) over time (in seconds), showing a stepwise increase during the TMA exposure and water exposure phases, followed by distinct plateaus.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Time (s) | Mass (ng/cm²) |\\n|---|---|\\n| 2 | 0,19 |\\n| 4 | 24,62 |\\n| 12 | 25,00 |\\n| 14 | 35,53 |\\n| 45 | 31,58 |\\n| 46 | 55,45 |\\n| 56 | 56,39 |\\n| 57 | 66,92 |\\n| 87 | 62,78 |\\n| 89 | 85,34 |\\n| 98 | 87,59 |\\n| 100 | 98,12 |\\n| 129 | 93,80 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"TMA.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Water dosing is longer than TMA dosing.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The TMA and Water steps are follower by purging steps. During purging, the unreacted precursors are being removed from the reaction chamber. This results in no deminishing mass changes, which results in plateaus in the graph.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Approximately 65 ng/cm².\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":667,"height":448}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/1/images/figure_3.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":"line chart","caption_source":"content.json:img_caption"},"width":667,"height":448,"image_format":"jpeg","image_sha256":"60a256c00607eeaedc9e2135b9f152c158242a203768a0171e101d12c0b43013","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_10.jpg","caption":"FIG. 10. (Color online) Resistivity (at room temperature) of TiN films obtained at  $100 - 400^{\\circ}\\mathrm{C}$ . The films were deposited by plasma-assisted ALD using  $\\mathrm{TiCl}_4$  in combination with an  $\\mathrm{H}_2 / \\mathrm{N}_2$  plasma. The resistivity was determined by in situ spectroscopic ellipsometry (thin films,  $\\sim 10 \\mathrm{nm}$ ) and four-point probe measurements (thicker films,  $>45 \\mathrm{nm}$ ) (Ref. 237). From S.B.S. Heil et al., J. Electrochem. Soc. 153, G956 (2006). Reproduced by permission of ECS—The Electrochemical Society.","id":"test/atomic-layer-deposition/experimental-usecase/10/fig_10","sample_id":"atomic-layer-deposition/experimental-usecase/10/fig_10","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure illustrates how resistivity decreases as deposition temperature increases. Measurements from spectroscopic ellipsometry (solid line) and 4-point probe (open circles) both show a consistent downward trend, indicating improved conductivity at higher deposition temperatures. Error bars represent experimental uncertainty for each temperature point.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Deposition Temperature (°C) | Resistivity (µΩ·cm) | Measurement Method          |\\n|-----------------------------|----------------------|----------------------------|\\n| 100                         | 300                  | Spectroscopic ellipsometry |\\n| 200                         | 150                  | Spectroscopic ellipsometry |\\n| 300                         | 120                  | Spectroscopic ellipsometry |\\n| 400                         | 80                   | Spectroscopic ellipsometry |\\n| 100                         | 200                  | 4-point probe              |\\n| 400                         | 70                   | 4-point probe              |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Resistivity (µΩ·cm).\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Spectroscopic ellipsometry, 4-point probe\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Resistivity decreases steadily as deposition temperature rises, showing that higher temperatures improve the film’s conductivity due to better atomic ordering or reduced defects.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":686,"height":528}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/10/images/fig_10.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/10/images/fig_10.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":"line chart","caption_source":"content.json:img_caption"},"width":686,"height":528,"image_format":"jpeg","image_sha256":"e6c4b938d9f668ab070f2a3108cd731c67d6bc93d91b5488f8c684628ad7b133","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_11.jpg","caption":"FIG. 11. (Color online) Resistivity (at room temperature) of TaN films as a function of  $\\mathrm{H}_2$  plasma exposure time (Ref. 223). The data were obtained by in situ spectroscopic ellipsometry and four-point probe measurements. Reprinted with permission from E. Langereis et al., J. Appl. Phys. 102, 083517 (2007). Copyright 2007, American Institute of Physics.","id":"test/atomic-layer-deposition/experimental-usecase/10/fig_11","sample_id":"atomic-layer-deposition/experimental-usecase/10/fig_11","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure illustrates how resistivity decreases sharply with increasing H₂ plasma exposure time. Both spectroscopic ellipsometry (triangles) and 4-point probe (circles) measurements show consistent results, with the resistivity dropping rapidly in the first few seconds before stabilizing at longer exposure times. The data demonstrate that extended H₂ plasma treatment improves conductivity by reducing film defects.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| H₂ Plasma Exposure Time (s) | Resistivity (µΩ·cm) | Measurement Method         |\\n|-----------------------------|---------------------|----------------------------|\\n| 0                           | 10000               | 4-point probe              |\\n| 1                           | 1000                | 4-point probe              |\\n| 5                           | 300                 | Spectroscopic ellipsometry |\\n| 10                          | 200                 | Spectroscopic ellipsometry |\\n| 20                          | 120                 | Spectroscopic ellipsometry |\\n| 30                          | 100                 | Spectroscopic ellipsometry |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Resistivity (µΩ·cm)\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Resistivity decreases rapidly with increasing plasma exposure time, dropping several orders of magnitude within the first few seconds and then stabilizing, indicating improved conductivity due to surface modification.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Spectroscopic ellipsometry, 4-point probe\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":681,"height":530}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/10/images/fig_11.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/10/images/fig_11.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":"line chart","caption_source":"content.json:img_caption"},"width":681,"height":530,"image_format":"jpeg","image_sha256":"61db7ddfb4b08c316b7440a839817ea54f81f044a6a1e9cd4f40b1a26fd69fd8","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_20_fig_4.jpg","caption":"Fig. 4 Lithium content of the films, as measured by TOF-ERDA, as a function of number of subsequent lithium sub-cycles  $n$  in the pulsing scheme of  $400 \\times (1 \\times \\mathrm{TiO_2} + 3 \\times \\mathrm{La_2O_3} + n \\times \\mathrm{Li_2O})$ .","id":"test/atomic-layer-deposition/experimental-usecase/20/fig_4","sample_id":"atomic-layer-deposition/experimental-usecase/20/fig_4","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart illustrates a non-linear  increase in lithium concentration (in at.%) as the number of LiO<sup>t</sup>Bu + H<sub>2</sub>O sub-cycles increases.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of LiO<sup>t</sup>Bu + H<sub>2</sub>O sub-cycles | Lithium concentration (at.%) |\\n|---|---|\\n| 0 | 0 |\\n| 1 | 19.67 |\\n| 3 | 21.06 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"After the  first sub-cycle.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"No, it is not efficient. Based on the plot, the lithium concentration saturates after the first cycle. Adding more consecutive lithium pulses wastes time and precursor without significantly changing the film composition.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The curve would show a linear, steady increase in concentration.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":477,"height":383}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/20/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/20/images/fig_4.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/20/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/20/Titta Aaltonen et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"20","first_classification_panel_id":"a","first_classification_label":"line chart","caption_source":"content.json:image_caption"},"width":477,"height":383,"image_format":"jpeg","image_sha256":"3f714365945734a41e1dc156e090b3eb98edf4078bbcc6add17a0fa5936b721e","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_1.jpg","caption":"Figure 1. (a) QCM data recorded during 40 cycles of  $\\mathrm{Ga}_2\\mathrm{O}_3$  ALD using TMGa-  $\\mathrm{O_3}$  at  $350^{\\circ}C$  with the timing sequence  $6 - 10 - 6 - 10$  b Expanded view of QCM data during 3 cycles of  $\\mathrm{Ga}_2\\mathrm{O}_3$  ALD where the individual TMGa and  $\\mathrm{O_3}$  exposures are indicated by the lower traces.","id":"test/atomic-layer-deposition/experimental-usecase/23/figure_1","sample_id":"atomic-layer-deposition/experimental-usecase/23/figure_1","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"},{"panel_id":"b","label":"process timing diagram"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"A line chart showing the mass increase over time with a constant rate of 31 ng/cm² per cycle.\"},{\"panel_id\":\"b\",\"text\":\"A line chart showing the mass change over time, focusing on the time range of 240 to 340 seconds, with Ozone and TMGa exposures marked.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Time (s) | Mass (ng/cm²) |\\n|---|---|\\n| 0 | 0 |\\n| 200 | 200 |\\n| 400 | 400 |\\n| 600 | 600 |\\n| 800 | 800 |\\n| 1000 | 1000 |\\n| 1200 | 1200 |\"},{\"panel_id\":\"b\",\"text\":\"| Time (s) | Mass (ng/cm²) |\\n|---|---|\\n| 260 | 260 |\\n| 280 | 255 |\\n| 300 | 270 |\\n| 320 | 320 |\\n| 340 | 315 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"An increase of 31 ng/cm^2 per cycle.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"350 °C.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Roughly 40 ng/cm^2.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Roughly 10 ng/cm^2.\"}]}]","bbox":[{"panel_id":"a","x":6,"y":8,"width":473,"height":471},{"panel_id":"b","x":485,"y":6,"width":481,"height":474}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/23/images/figure_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/23/images/figure_1.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":"line chart","caption_source":"content.json:image_caption"},"width":967,"height":480,"image_format":"jpeg","image_sha256":"2adb3f7989356fca4a3bdd7dba34016165e28c049e70be4a4f47586a4649fd7f","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_10.jpg","caption":"Figure 10. Optical transmittance measurements from  $800\\mathrm{\\AA}Ga_{2}O_{3}$  film deposited onto both sides of a fused  $\\mathrm{SiO}_2$  substrate.a) Transmittance of  $\\mathrm{Ga}_2\\mathrm{O}_3$  referenced to a fused  $\\mathrm{SiO}_2$  substrate background, and (b) determination of the optical bandgap of 4.95  $\\mathrm{eV}$  by fitting the optical absorption data.","id":"test/atomic-layer-deposition/experimental-usecase/23/figure_10","sample_id":"atomic-layer-deposition/experimental-usecase/23/figure_10","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"},{"panel_id":"b","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows the transmittance of Ga2O3 referenced to a fused SiO2 substrate background.\"},{\"panel_id\":\"b\",\"text\":\"Determination of the optical bandgap of 4.95 eV by fitting the optical absorption data.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Wavelength (nm) | Transmittance |\\n|---|---|\\n| 250 | 0 |\\n| 300 | 100 |\\n| 500 | 75 | \\n| 1000 | 85 |\\n| 1500 | 90 |\\n| 2000 | 90 |\"},{\"panel_id\":\"b\",\"text\":\"| Energy (eV) | (αhν)^2 (ev/cm)^2 |\\n|---|---|\\n| 4 | 0 |\\n| 4.5 | 0 |\\n| 5 | 1 |\\n| 5.5 | 3 |\\n| 6 | 5 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The films are highly transparent in the wavelength range of 300-2500 nm.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 300 nm.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 80%.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"4.95 eV.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":15,"width":469,"height":354},{"panel_id":"b","x":2,"y":358,"width":470,"height":365}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/23/images/figure_10.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/23/images/figure_10.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":"line chart","caption_source":"content.json:image_caption"},"width":475,"height":725,"image_format":"jpeg","image_sha256":"f55f2c3fb8c224242b1a54823caefa5e91efec4fc127e089e084f76713c19410","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_24_figure_9.jpg","caption":"Figure 9. Intensity of  $\\mathrm{MeOH}$  during background cycles and process cycles at a pressure of  $3\\times 10^{-3}$  mbar. MeOH is released during the precursor pulse and during the ozone pulse.","id":"test/atomic-layer-deposition/experimental-usecase/24/figure_9","sample_id":"atomic-layer-deposition/experimental-usecase/24/figure_9","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows time-resolved mass spectrometry of methanol (MeOH, m/e = 31) during an ALD process using a metal precursor and ozone. Methanol is released during both half-cycles: small peaks during the precursor pulse indicate chemisorption via reactions with surface –OH groups, while larger peaks during the ozone pulse reflect oxidation and removal of remaining methoxy ligands. This dual release confirms that ligand elimination occurs in both steps, with ozone being the dominant cleanup mechanism.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Time (min) | Intensity (a.u.) | Condition / Action        |\\n|-----------:|----------------:|---------------------------|\\n| 5          | 3.5             | Precursor Pulse (P)       |\\n| 20         | 2.5             | Precursor Pulse (P)       |\\n| 35         | 2.1             | Precursor Pulse (P)       |\\n| 50         | 1.9             | Precursor Pulse (P)       |\\n| 65         | 1.8             | Precursor Pulse (P)       |\\n| 85         | 8.8             | Ozone Pulse (O₃)          |\\n| 100        | 2.5             | Precursor Pulse (P)       |\\n| 115        | 9.4             | Ozone Pulse (O₃)          |\\n| 130        | 2.7             | Precursor Pulse (P)       |\\n| 145        | 9.7             | Ozone Pulse (O₃)          |\\n| 160        | 2.8             | Precursor Pulse (P)       |\\n| 175        | 9.6             | Ozone Pulse (O₃)          |\\n| 190        | 2.7             | Precursor Pulse (P)       |\\n| 205        | 9.5             | Ozone Pulse (O₃)          |\\n| 220        | 2.6             | Precursor Pulse (P)       |\\n| 235        | 8.0             | Ozone Pulse (O₃)          |\\n| 250        | 7.3             | Ozone Pulse (O₃)          |\\n| 265        | 6.8             | Ozone Pulse (O₃)          |\\n| 280        | 6.4             | Ozone Pulse (O₃)          |\\n| 295        | 6.0             | Ozone Pulse (O₃)          |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The precursor pulse produces a methanol signal because the metal precursor reacts with surface hydroxyl groups (–OH) remaining from the previous step. This ligand-exchange reaction (for example, M–OMe + surface–OH → surface–O–M + MeOH) releases methanol immediately upon chemisorption, without requiring the subsequent oxidant pulse.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"If all methanol were released during the precursor pulse, it would imply the precursor reacted completely with the surface (consuming all ligands) without needing a second step. This would effectively be a self-limiting adsorption but might lack the mechanism to regenerate -OH groups for the next cycle.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The relatively small peak during the precursor step suggests that only a fraction of the ligands (e.g., 1 out of 4) are lost during the initial anchoring to the surface. The much larger peak during the ozone step confirms that the majority of the ligands are still attached to the metal center and are essentially \\\"burned off\\\" or displaced only when the strong oxidant is introduced.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The peaks show a noticeable decay tail. If the purge time is too short, residual methanol will remain in the chamber. Since methanol can re-adsorb or react with the precursors in the gas phase (CVD-like reaction), this would lead to Particle Formation (dust) or non-uniform film thickness across the wafer. It necessitates extended purge times, which reduces manufacturing throughput .\"}]}]","bbox":[{"panel_id":"a","x":1,"y":1,"width":626,"height":479}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/24/images/figure_9.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/24/images/figure_9.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/24/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/24/Martin Rose et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"24","first_classification_panel_id":"a","first_classification_label":"line chart","caption_source":"content.json:image_caption"},"width":627,"height":481,"image_format":"jpeg","image_sha256":"72d792aafda0afbbfb5327374d624143aee5bbf386d7f9f7682ba3e708c1c87d","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_10.jpg","caption":"Figure 10. QCM mass changes during two ALD cycles using saturating doses of the tetrakis(dimethylamido)zirconium precursor and water at  $200^{\\circ}\\mathrm{C}$ .","id":"test/atomic-layer-deposition/experimental-usecase/3/figure_10","sample_id":"atomic-layer-deposition/experimental-usecase/3/figure_10","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart illustrates the change in mass over time. Addition of metal precursors produces rapid net mass gain, adding water produces net mass loss.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Time (seconds) | Mass (micrograms) | Step |\\n|---|---|---|\\n| 0  |  0  | - |\\n| 4  |  0  | Start Metal doses  |\\n| 8  |  0.053  | Start Water doses  |\\n| 15 |  0 .037 | Start Metal doses  |\\n| 19  |  0.089  | Start Water doses  |\\n| 25  |  0.073  | -  |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The graph demonstrates that the surface reactions are fast. This is supported by the nearly vertical lines visible at both the \\\"Metal Doses\\\" and \\\"Water Doses\\\" points. This indicates that the mass change upon precursor introduction is occurring more rapidly than the instrument's sampling time.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The ratio is bigger than one.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The flat horizontal plateaus after each dose confirm the process is self-limiting.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The net growth is the difference in mass between the baseline before the metal dose and the new baseline established after the water dose is complete. After the first full cycle, the mass change is approx.  0.037 micrograms.  After the second full-cycle, it is appox. 0.036 micrograms.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":534,"height":417}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/images/figure_10.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/images/figure_10.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":"line chart","caption_source":"content.json:img_caption"},"width":534,"height":417,"image_format":"jpeg","image_sha256":"a794c7eb0862cafc9c68189bc40655c3762db6b66409eee1d4ab79af38403ec5","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_11.jpg","caption":"Figure 11. QCM mass changes during one ALD cycle using undersaturating doses of the tetrakis(dimethylamido)zirconium precursor and water at  $200^{\\circ}\\mathrm{C}$ .","id":"test/atomic-layer-deposition/experimental-usecase/3/figure_11","sample_id":"atomic-layer-deposition/experimental-usecase/3/figure_11","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart illustrates the mass change over time, upon adding 12 doses of metal precursor and one dose of water, with a significant increase at around 40 seconds followed by a decrease after water addition.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Time (seconds) | Mass (micrograms) |\\n|---|---|\\n| 4.280 | 0.000 |\\n| 5.010 | 0.002 |\\n| 7.247 | 0.001 |\\n| 7.636 | 0.002 |\\n| 10.895 | 0.001 |\\n| 11.381 | 0.002 |\\n| 14.251 | 0.001 |\\n| 14.737 | 0.002 |\\n| 18.045 | 0.001 |\\n| 18.482 | 0.002 |\\n| 21.547 | 0.001 |\\n| 21.887 | 0.002 |\\n| 25.146 | 0.001 |\\n| 25.438 | 0.002 |\\n| 28.405 | 0.000 |\\n| 28.940 | 0.002 |\\n| 31.858 | 0.000 |\\n| 32.539 | 0.002 |\\n| 35.700 | 0.000 |\\n| 35.992 | 0.001 |\\n| 39.008 | 0.001 |\\n| 40.807 | 0.053 |\\n| 42.753 | 0.052 |\\n| 46.109 | 0.053 |\\n| 49.903 | 0.038 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"After 11 metal doses.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"11th metal dose.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The surface reaction is self-limiting and the surface was already saturated by the previous dose.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The process is net deposition. Although the water dose causes a slight mass decrease, the gain from the metal dose is significantly larger. The final mass after the complete cycle is higher than the starting mass, resulting in a permanent net mass increase.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":536,"height":370}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/images/figure_11.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/images/figure_11.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":"line chart","caption_source":"content.json:img_caption"},"width":536,"height":370,"image_format":"jpeg","image_sha256":"0c19323630a8c6f9b56eccfc4f572e5be011f5bbd4da9bf89d5fe2b11de34b21","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_8.jpg","caption":"Figure 8. Purge times required to achieve  $0.096\\mathrm{-nm}$  per cycle thickness for zirconium oxide films  $\\mathrm{0.093 - nm}$  for the hafnium oxide) at low deposition temperatures.","id":"test/atomic-layer-deposition/experimental-usecase/3/figure_8","sample_id":"atomic-layer-deposition/experimental-usecase/3/figure_8","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows the relationship between purge time (in seconds) and thickness per cycle (in nanometers). As purge time increases, the thickness per cycle decreases.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Purge Time (seconds) | Thickness per Cycle (nm) |\\n|---|---|\\n| 5  | 0.17 |\\n| 10 | 0.15 |\\n| 30 | 0.13 |\\n| 45 | 0.11 |\\n| 120 | 0.10 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The thickness at very short purge times is higher than at very high purge times. The difference arises because short purge times fail to remove excess precursor molecules. This leads to non-self-limiting growth and thicker films.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"High purge times are more effective in achieving controllable film growth.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At approximately 100 - 120 s.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The larger error bars at short times indicate higher variability. At 120 seconds, the process is self-limiting, leading to consistent results with small error bars. At short purge times, the mechanism depends on the amount of unremoved precursor and therefore varies from run to run.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":464,"height":342}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/images/figure_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/3/images/figure_8.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":"line chart","caption_source":"content.json:img_caption"},"width":464,"height":342,"image_format":"jpeg","image_sha256":"5cd47604cd422ec474db7ebed61ff179139cd756066cd62dff4345ca5115f7a1","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_4.jpg","caption":"FIG. 4. (Color online) QCM measurements during plasma-assisted ALD of  $\\mathrm{Al}_2\\mathrm{O}_3$  at a deposition temperature of  $70^{\\circ}\\mathrm{C}$ . The mass gain is shown for ten reaction cycles with the length of one ALD cycle being indicated. The precursor dose is smaller than necessary for saturation of the surface reactions.","id":"test/atomic-layer-deposition/experimental-usecase/39/fig_4","sample_id":"atomic-layer-deposition/experimental-usecase/39/fig_4","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart illustrates the change in QCM mass gain over time, showing a stepwise increase at regular intervals.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Time (s) | QCM: Mass gain (Hz) |\\n|---|---|\\n| 0 | 0 |\\n| 100 | 10 |\\n| 200 | 18 |\\n| 300 | 25 |\\n| 400 | 32 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"70 °C.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"10.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"2 seconds.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The figure illustrates how the mass of the growing film evolves during ten ALD cycles at a substrate temperature of 70 °C when a non-saturated aluminium precursor dose is used in each cycle. The quartz crystal microbalance data show a clear stepwise increase in mass, confirming that material is added incrementally during every ALD cycle. A transient negative mass change appears during each oxygen plasma step; however, this is attributed to a plasma-induced measurement artifact rather than actual material loss.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":675,"height":547}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/39/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/39/images/fig_4.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":"line chart","caption_source":"content.json:image_caption"},"width":675,"height":547,"image_format":"jpeg","image_sha256":"c06cb4e2ccc6afa284a64cc9861649dc515bca19a6be8988eaa965d5747746b6","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_4.jpg","caption":"Figure 4. Time-resolved QMS signals for  $m / z$  ratios 17  $\\mathrm{(NH_3^+)}$  28  $\\mathrm{(N_2^+)}$  27  $\\mathrm{(C_2H_3^+}$ $\\mathrm{HCN^{+}}$  39  $\\mathrm{(C_3H_3^+}$ $\\mathrm{HCN^{+}}$  and 66  $\\mathrm{(C_5H_6^+)}$  , collected during the plasma subcycle for the (a)  $\\mathrm{AB - NH_3}$  process, (b)  $\\mathrm{AB - H_2 / N_2}$  process, and (c)  $\\mathrm{ABC - N_2 - H_2}$  process. A normal ALD cycle and a reference cycle without the  $\\mathrm{CoCp_2}$  precursor dosing were measured, with plasma ignition for 11 s during both cycles (indicated with an arrow in the panels for  $m / z = 28$  ). The  $\\mathrm{H}_2 / (\\mathrm{H}_2 + \\mathrm{N}_2)$  mixing ratio of the  $\\mathrm{H}_2 / \\mathrm{N}_2$  plasma was  $\\sim 0.77$","id":"test/atomic-layer-deposition/experimental-usecase/61/figure_4","sample_id":"atomic-layer-deposition/experimental-usecase/61/figure_4","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"},{"panel_id":"b","label":"line chart"},{"panel_id":"c","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This figure presents a comparison of ion current signals at specific mass-to-charge ratios (m/z) for three types of processes used in cobalt deposition: AB-NH₃, AB-H₂/N₂, and ABC-N₂-H₂. Each subfigure shows the evolution of key ions such as NH₃⁺ (m/z = 17), HCN⁺ (m/z = 27), and larger hydrocarbon fragments (e.g., m/z = 39, 52, 66) during plasma exposure and precursor pulsing. Signals are compared between conditions with and without the cobalt precursor (CoCp₂), allowing for assessment of how precursor chemistry and plasma composition influence ion generation and reaction pathways. The ABC process appears to produce sharper and more defined signals, suggesting more controlled surface reactions\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Time (s) | m/z = 17 (NH₃⁺, AB-NH₃) | m/z = 27 (C₂H₃⁺) | m/z = 39 (C₃H₃⁺) |\\n|----------|--------------------------|------------------|------------------|\\n| 0        | 0.10                     | 0.00             | 0.00             |\\n| 5        | 0.15                     | 0.02             | 0.01             |\\n| 10       | 0.20                     | 0.05             | 0.04             |\\n| 15       | 0.18                     | 0.03             | 0.02             |\\n| 20       | 0.12                     | 0.01             | 0.01             |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The AB-NH₃ process exhibits sustained NH₃⁺ ion signals (m/z = 17) and weaker signals for hydrocarbon species. The AB-H₂/N₂ process shows moderate signals for NH₃⁺ and some fragmentation peaks (e.g., m/z = 27, HCN⁺), with lower overall intensity compared to the ABC process. In the ABC-N₂-H₂ process, ion current responses are sharper and more defined, especially for hydrocarbon fragments, indicating more abrupt gas-phase reactions and possibly better surface saturation. These differences reflect how plasma composition and step timing affect decomposition and desorption dynamics in each process.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"AB-NH₃ shows a weak, broad signal, AB-H₂/N₂ yields a moderate peak, ABC-N₂-H₂ produces a strong, sharp peak\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The AB-NH₃ process.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":10,"y":9,"width":283,"height":741},{"panel_id":"b","x":324,"y":9,"width":256,"height":743},{"panel_id":"c","x":606,"y":9,"width":486,"height":746}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/61/images/figure_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/61/images/figure_4.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":"line chart","caption_source":"content.json:image_caption"},"width":1092,"height":755,"image_format":"jpeg","image_sha256":"165be5e57e8d6b94ba7f136165ff49c18a2f5870eaa50e4cd27d3478587d5e1d","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_6.jpg","caption":"Figure 6. Co content from XPS as a function of QMS ion current at  $m / z = 17$ . The QMS ion current is a measure for the  $\\mathrm{NH}_3$  production in the  $\\mathrm{H}_2 / \\mathrm{N}_2$  plasma and was varied by changing the  $\\mathrm{H}_2 / \\mathrm{N}_2$  mixing ratio of the source gas. The pressure of the  $\\mathrm{H}_2 / \\mathrm{N}_2$  gas mixture was kept constant at  $13\\mathrm{mTorr}$ . The Co content was determined using XPS on films obtained by performing  $800\\mathrm{ALD}$  using the various  $\\mathrm{H}_2 / \\mathrm{N}_2$  mixing ratios. XPS was carried out after sputtering with  $\\mathrm{Ar^{+}}$  ions for  $3\\mathrm{min}$ . The resistivity values of the Co films are indicated in the figure, in which the dashed line represents a linear fit through the data. It is noted that the decrease of the film resistivity as a function of  $\\mathrm{NH}_x$  concentration might be related to the changes in the film structure, aside from the increased film purity.","id":"test/atomic-layer-deposition/experimental-usecase/61/figure_6","sample_id":"atomic-layer-deposition/experimental-usecase/61/figure_6","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This figure shows how increasing ion current at m/z = 17 correlates with increasing cobalt content in atomic percent, with error bars indicating measurement uncertainty. As the ion current rises from 2 to 12 × 10⁻¹⁰ A, Co content increases from approximately 76.5% to 88%. Superimposed resistivity values show a dramatic drop from >10⁹ μΩ·cm to as low as 78 μΩ·cm, highlighting a strong correlation between plasma-generated NH₃ species (linked to m/z = 17) and improved film conductivity. The trend suggests that higher ion currents at this mass-to-charge ratio contribute to better cobalt incorporation and lower resistivity, revealing a potential control knob for tuning material performance in ALD or CVD processes.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Ion current m/z = 17 (10⁻¹⁰ A) | Co content (at.%) | Resistivity (μΩ·cm) |\\n|-------------------------------|--------------------|----------------------|\\n| 2                             | ~76.5              | >10⁹                |\\n| 4                             | ~78.0              | 3.6×10⁸             |\\n| 6                             | ~80.0              | 2.5×10³             |\\n| 8                             | ~82.0              | 1.5×10³             |\\n| 10                            | ~85.0              | 78                  |\\n| 12                            | ~88.0              | Not shown           |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"As the ion current at m/z = 17 increases, the cobalt content in atomic percent also increases, indicating a positive correlation between plasma-generated NH₃ species and cobalt incorporation. This trend suggests that higher concentrations of reactive species in the plasma enhance the reduction and integration of cobalt into the film during deposition processes.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Higher Co content lowers resistivity, ion current tunes conductivity, plasma composition control improves deposition\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At an ion current of 10 × 10⁻¹⁰ A.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":503,"height":424}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/61/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/61/images/figure_6.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":"line chart","caption_source":"content.json:image_caption"},"width":503,"height":425,"image_format":"jpeg","image_sha256":"4d14147ccdc56bcd4d4615d997afcdf9cfc5aae9aaf3e8cc6a6c1197706d7758","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_62_fig_7.jpg","caption":"FIG. 7. (Color online) (a) Time-resolved emission intensity recorded at  $777~\\mathrm{nm}$  related to atomic oxygen during plasma-assisted ALD of  $\\mathrm{Al}_2\\mathrm{O}_3$  Failures of plasma ignition can be observed for the fourth and sixth ALD cycles. (b) Time-resolved emission intensities recorded during plasma-assisted ALD of TiN in which a  $\\mathrm{H}_2 - \\mathrm{N}_2$  plasma is employed. The dashed line indicates the point at which the  $\\mathrm{H}_2$  gas flow fails.","id":"test/atomic-layer-deposition/experimental-usecase/62/fig_7","sample_id":"atomic-layer-deposition/experimental-usecase/62/fig_7","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"},{"panel_id":"b","label":"multiple line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"A line chart displaying intensity over time with consistent peaks at regular intervals.\"},{\"panel_id\":\"b\",\"text\":\"A multiple line chart showing intensity of two plasma species over time with distinct peaks.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Time (s) | Intensity (a.u.) |\\n|----------|------------------|\\n| 0        | 0             |\\n| 10       | ~175             |\\n| 20       | 0             |\\n| 30       | 0             |\\n| 40       | ~175             |\\n| 50       | ~175             |\\n| 60       | ~175             |\\n| 70       | ~175             |\"},{\"panel_id\":\"b\",\"text\":\"| Time (s) | H (656 nm) | N₂ (358 nm) |\\n|-----------|------------|-------------|\\n| 3000      | ~30000     | ~5000       |\\n| 3100      | ~30000     | ~5000       |\\n| 3200      | ~30000     | ~5000       |\\n| 3300      | ~30000     | ~8000       |\\n| 3400      | ~20000     | ~35000       |\\n| 3500      | ~12000     | ~40000       |\\n| 3600      | ~12000     | ~40000       |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"At these times the plasma is not on, this means that these cycles are not added up to the total amount of cycles.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The plasma takes 5s.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 14 s.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"b\"}]}]","bbox":[{"panel_id":"a","x":0,"y":1,"width":608,"height":494},{"panel_id":"b","x":0,"y":479,"width":619,"height":511}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/62/images/fig_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/62/images/fig_7.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/62/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/62/Mackus et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"62","first_classification_panel_id":"a","first_classification_label":"line chart","caption_source":"content.json:image_caption"},"width":614,"height":988,"image_format":"jpeg","image_sha256":"d3ef151ceb8928057ab3589b28c5dc36c3bf2fd6f6e3d55ae6e0911f5a3247db","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_62_fig_8.jpg","caption":"FIG. 8. (Color online) (a) Evolution of the emission intensity at  $656~\\mathrm{nm}$  related to atomic hydrogen recorded during plasma-assisted ALD of TiN without reactor wall conditioning prior to deposition. (b) Integrated emission intensities per cycle at  $656~\\mathrm{nm}$  as a function of cycle number for plasma-assisted ALD of TiN in a reactor in which the reactor wall was conditioned by various procedures.","id":"test/atomic-layer-deposition/experimental-usecase/62/fig_8","sample_id":"atomic-layer-deposition/experimental-usecase/62/fig_8","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"},{"panel_id":"b","label":"multiple line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"A line chart displaying intensity over time, showing a decreasing trend.\"},{\"panel_id\":\"b\",\"text\":\"A line chart showing the integrated 656 nm signal per cycle across different conditions.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Time (s) | Intensity (a.u.) |\\n|---|---|\\n| 0 | 0 |\\n| 50 | 0 |\\n| 100 | 4x10^4 |\\n| 150 | 4x10^4 |\\n| 200 | 4x10^4 |\\n| 250 | 6x10^4 |\\n| 300 | 0 |\"},{\"panel_id\":\"b\",\"text\":\"| Cycle number | Integrated 656 nm signal per cycle No cond.|Integrated 656 nm signal per cycle H2-N2 plasma|Integrated 656 nm signal per cycle H2 plasma|Integrated 656 nm signal per cycle TiCl4|\\n|---|---|---|---|---|\\n| 0 | 9x10^5 |5x10^5 |7x10^5 |6x10^5 |\\n| 20 | 8x10^5 |5x10^5 |5x10^5 |5x10^5 |\\n| 40 | 6x10^5 |5x10^5 |5x10^5 |4.5x10^5 |\\n| 60 | 6x10^5 |5x10^5 |5x10^5 |4.5x10^5 |\\n| 80 | 6x10^5 |5x10^5 |5x10^5 |4.5x10^5 |\\n| 100 | 5x10^5 |5x10^5 |5x10^5 |4.5x10^5 |\\n| 120 | 5x10^5 |5x10^5 |5x10^5 |4.5x10^5 |\\n| 140 | 5x10^5 |5x10^5 |5x10^5 |4.5x10^5 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"There does seem to be a startinf effect. The initial plasma pulses show higher intensities than the last pulses.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The best process is the one where the effect is only present in very little number of cycles. This would be the orange line.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"After 140 cycles the decrease seems to saturate. This indicates that there are species in the plasma present that are not related to the ALD chemistry of the process.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The saturation level is 500000.\"}]}]","bbox":[{"panel_id":"a","x":1,"y":1,"width":631,"height":505},{"panel_id":"b","x":1,"y":512,"width":629,"height":503}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/62/images/fig_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/62/images/fig_8.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/62/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/62/Mackus et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"62","first_classification_panel_id":"a","first_classification_label":"line chart","caption_source":"content.json:image_caption"},"width":631,"height":1014,"image_format":"jpeg","image_sha256":"ba84bf51a366b7bfe3d648e4960c6e4f1daa3f17e189ccfa02e4a4e9832d9579","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_FIG4_a.jpg","caption":"Capacitively coupled","id":"test/atomic-layer-deposition/experimental-usecase/67/FIG4_a","sample_id":"atomic-layer-deposition/experimental-usecase/67/FIG4_a","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":0,"y":0,"width":614,"height":506}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG4_a.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG4_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":"line chart","caption_source":"content.json:image_caption"},"width":614,"height":506,"image_format":"jpeg","image_sha256":"b155fda1d549675efd7e59508e501798c6f90bba13636448cbdd21c94695bf13","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_FIG6_a.jpg","caption":"FIG. 6. (a) Atomic O density vs  $\\mathrm{O}_2$  flow fraction measured with OES. The y axis plotted on the same scale as other OES measurements for comparison. (b) Ion flux and plasma potential as a function of  $\\mathrm{O}_2$  flow fraction.","id":"test/atomic-layer-deposition/experimental-usecase/67/FIG6_a","sample_id":"atomic-layer-deposition/experimental-usecase/67/FIG6_a","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This figure shows how atomic oxygen density varies as a function of O₂ flow fraction in the plasma. As the O₂ flow fraction increases from 0.0 to 0.3, the atomic oxygen density rises, reaching a maximum near 0.3. At higher O₂ flow fraction (0.5), the atomic oxygen density decreases slightly. This behaviour suggests an optimal oxygen fraction for generating atomic oxygen, beyond which recombination or dilution effects reduce the available atomic species\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| O₂ Flow Fraction | Atomic O density (atoms/m³) |\\n|------------------|----------------------------|\\n| 0.0             | ~1.8 × 10^19                |\\n| 0.1             | ~2.5 × 10^19                |\\n| 0.3             | ~3.0 × 10^19                |\\n| 0.5             | ~2.5 × 10^19                |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Atomic oxygen density initially increases as the O₂ flow fraction rises from 0.0 to 0.3, indicating more efficient generation of atomic oxygen in this range. Beyond this point, increasing the O₂ flow fraction to 0.5 leads to a decrease in atomic oxygen density. This trend suggests that excessive oxygen may promote recombination or reduce plasma efficiency, resulting in fewer free atomic oxygen species.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"~ 0.3 O₂ flow fraction\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Maximizes atomic O availability, Improves efficiency, Avoids recombination, Enhances surface reaction control\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":645,"height":511}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG6_a.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/experimental-usecase/67/images/FIG6_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":"line chart","caption_source":"content.json:image_caption"},"width":645,"height":511,"image_format":"jpeg","image_sha256":"552f28acb4633c87705da7981885d053308f39576d902c21b4ae086b12fb23c8","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_13.jpg","caption":"Fig. 13. Growth rate as a function of V/III ratio at  $1273\\mathrm{K}$ . Note that growth rate decreases as the amount of V-precursor for the reactor shown in Fig. 12.","id":"test/atomic-layer-deposition/simulation-usecase/18/fig_13","sample_id":"atomic-layer-deposition/simulation-usecase/18/fig_13","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart  represents the GaN growth rate versus V/III ratio at 1273 K for the stagnation-flow reactor used by Safvi et al. Unlike the diffusion-limited decrease observed in Figure 13, this configuration exhibits a steady linear increase in growth rate with higher V/III ratios. This indicates that in stagnation-flow reactors, added ammonia effectively enhances surface reaction kinetics without being limited by gas-phase diffusion.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| V/III | Growth Rate (µm/hr) |\\n|---|---|\\n| 35 | 0.58 |\\n| 70 | 0.59 |\\n| 142 | 0.54 |\\n| 214 | 0.45 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Increasing the ratio from 500 to 2000 increases the growth rate from approximately 1.35 μm/hr to about 1.65 μm/hr, corresponding to a ~22% increase. This indicates that supplying additional ammonia enhances the growth rate and confirms that the process is not limited by diffusion in this regime.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"a        The diffusion-limited regime is preferable for large-area uniformity. Here, growth is controlled by mass transport, which can be made uniform with proper reactor design, leading to consistent thickness. The trade-off is that the absolute growth rate is lower and cannot be increased by raising temperature, potentially reducing throughput compared to a reaction-limited process optimized for speed.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The constant growth rate indicates the process has entered a diffusion-limited regime. At lower temperatures, growth is typically reaction-limited, where the rate increases exponentially with temperature. Above a certain temperature, the surface reaction becomes so fast that the rate is instead limited by the slower, weakly temperature-dependent physical transport of reactants through the gas boundary layer to the surface, resulting in the observed plateau.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"a         The flat growth rate profile across a 180 K window demonstrates inherent robustness to temperature fluctuations. This means that minor hot or cold spots in a large reactor will not translate into significant thickness variations, leading to better wafer-scale uniformity and higher manufacturing yield without requiring prohibitively precise temperature control.\"}]}]","bbox":[{"panel_id":"a","x":5,"y":5,"width":585,"height":379}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/18/images/fig_13.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/18/images/fig_13.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":"line chart","caption_source":"content.json:img_caption"},"width":592,"height":383,"image_format":"jpeg","image_sha256":"0a4e85a0f67d092db3b2cd63066c862ff95ab385c4631003fece596a21dcbd4e","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_14.jpg","caption":"Fig. 14. Growth rate as a function of V/III ratio at  $1273\\mathrm{K}$ . Note that growth rate increases as the amount of V-precursor for the reactor shown in Fig. 7.","id":"test/atomic-layer-deposition/simulation-usecase/18/fig_14","sample_id":"atomic-layer-deposition/simulation-usecase/18/fig_14","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This line chart represents shows a positive linear relationship between the V/III precursor ratio and the GaN growth rate at 1273 K in a specific reactor configuration. Higher ammonia (V) flow relative to the gallium (III) precursor leads to a faster growth rate.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| V/III | Growth Rate (µm/hr) |\\n|---|---|\\n| 500 | 1.4 |\\n| 1000 | 1.48 |\\n| 1500 | 1.56 |\\n| 2000 | 1.63 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Doubling the ratio effectively supplies more nitrogen precursor to the surface. Since the trend is linear and positive, this adjustment will directly boost the deposition rate, requiring the engineer to increase the ammonia mass flow controller setpoint while keeping the TMG flow constant.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"A reactor in a reaction-limited regime (like this one) is more challenging for large-wafer uniformity. Here, growth rate is highly sensitive to local temperature and precursor flux variations. A mass-transport-limited regime offers a natural plateau where growth rate is less sensitive to such local fluctuations, making it inherently better for thickness uniformity across a large area.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The linear increase indicates the growth is reaction-limited and governed by the surface concentration of active nitrogen. A higher V/III ratio delivers more ammonia, which decomposes to provide more NHₓ radicals at the surface. This directly increases the rate of the nitrogen-incorporation reaction, as Ga adatoms are abundant, making growth rate proportional to the nitrogen precursor flux.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"This provides precise thickness control for critical nanoscale layers. By calibrating the linear V/III-to-growth-rate relationship, engineers can accurately deposit a barrier layer to an exact target thickness by controlling the ratio and growth time. This precision is essential for controlling the 2D electron gas density and threshold voltage in the HEMT.\"}]}]","bbox":[{"panel_id":"a","x":5,"y":5,"width":661,"height":421}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/18/images/fig_14.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/18/images/fig_14.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":"line chart","caption_source":"content.json:img_caption"},"width":664,"height":425,"image_format":"jpeg","image_sha256":"281edc0e514dd9bbd3fb1f602085634454bf862894675ceb7c2d2b89efd7da41","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_37_fig_1.jpg","caption":"Fig. 1. Proposed reaction portrait to illustrate ALD growth of alumina from TMA and  $\\mathrm{H}_2\\mathrm{O}$ . The space is defined by chemical concentrations of surface intermediates: the  $y$  axis is the surface coverage of  $\\mathrm{CH}_3$ , adsorbed fragments of the TMA precursor, the  $x$  axis gives the  $\\mathrm{H}$  concentration, as  $\\mathrm{OH}$  or  $\\mathrm{H}_2\\mathrm{O}$ , resulting from the  $\\mathrm{H}_2\\mathrm{O}$  precursor. The units are atoms per unit surface area. The TMA pulse and subsequent purge is indicated as a solid line, the  $\\mathrm{H}_2\\mathrm{O}$  pulse and purge as a dotted line. The cycle as shown therefore represents the deposition of  $1 / 2\\mathrm{Al}_2\\mathrm{O}_3$  and the evolution of  $3\\mathrm{CH}_4$  (diagonal lines) per cell.","id":"test/atomic-layer-deposition/simulation-usecase/37/fig_1","sample_id":"atomic-layer-deposition/simulation-usecase/37/fig_1","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows a phase portrait representing the ALD growth of Al₂O₃ from TMA and H₂O. It plots the surface coverage of reactive intermediates, CH₃ (from TMA) on the y-axis and H (from H₂O) on the x-axis, over the course of a pulse–purge–pulse–purge cycle. Solid lines indicate the TMA pulse and purge, dotted lines indicate the H₂O pulse and purge, and diagonal lines represent CH₄ desorption. The portrait illustrates how the surface concentrations evolve during the cycle, returning to their initial values in a stoichiometric growth cycle.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Surface CH₃ coverage (atoms/unit area) | Surface H coverage (atoms/unit area) |\\n|---|---|\\n| 0   | 0   |\\n| 0.5 | 0.5 |\\n| 1   | 1   |\\n| 1.5 | 1.5 |\\n| 2   | 2   |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"During the ALD cycle, the TMA pulse saturates the surface with CH₃ fragments, increasing the y-axis value, followed by a purge that maintains the surface coverage. The H₂O pulse then introduces H species, increasing the x-axis value, followed by a purge. Diagonal lines in the portrait represent the desorption of CH₄ as a by-product. A successful stoichiometric cycle returns the surface concentrations of both CH₃ and H to their starting values, indicating no accumulation of intermediates.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"CH₃ generally reaches higher surface coverage during the TMA pulse compared to H introduced by the H₂O pulse.\"}]},{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Yes. If the surface concentrations of CH₃ and H do not return to their initial values at the end of the cycle, it indicates accumulation of intermediates and non-stoichiometric growth.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"The height and width of the portrait correspond to the maximum surface coverages of CH₃ and H during the cycle, These dimensions are proportional to the net Al₂O₃ growth rate per cycle, A larger portrait indicates more precursor adsorption and greater film growth, while a smaller portrait suggests reduced surface reactions and slower growth\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":483,"height":447}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/37/images/fig_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/37/images/fig_1.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/37/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-deposition/simulation-usecase/37/Simon D. Elliott.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"37","first_classification_panel_id":"a","first_classification_label":"line chart","caption_source":"content.json:img_caption"},"width":483,"height":447,"image_format":"jpeg","image_sha256":"0d6d76d6cca0d495b22fc099c4e441357fc7545908588186d8f63653b4c0e0ed","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_12_fig_3.jpg","caption":"FIG. 3. Film thickness, measured every five cycles, as a function of number of pulses or cycles performed on  $\\mathrm{Al_2O_3}$ . (a)  $\\mathrm{SF_6}$  plasma without a TMA dose representing half-cycle A. (b) TMA and  $\\mathrm{SF_6}$  gas dosing without striking a plasma representing half-cycle B. (c) Full ALE cycles with alternating  $\\mathrm{SF_6}$  plasma and TMA doses, confirming a linear relationship between the etch depth and the number of cycles. The inset shows a magnification of the transition between half-cycles A and B. Standard ALE recipe conditions are used for each step. The horizontal dashed lines indicate the initial film thickness before processing.","id":"test/atomic-layer-etching/experimental-usecase/12/fig_3","sample_id":"atomic-layer-etching/experimental-usecase/12/fig_3","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"},{"panel_id":"b","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart represents the film thickness (Å), measured every five cycles, as a function of the number of pulses or cycles (0–110) performed on Al₂O₃. (a) 5 seconds SF₆ plasma without a TMA dose, representing half-cycle A. (b) 0.5 seconds TMA and SF₆ gas dosing without striking a plasma, representing half-cycle B. (c) Full ALE cycles with alternating SF₆ plasma and TMA doses, confirming a linear relationship between the etch depth and the number of cycles. The figure shows that the film thickness remains at approximately 200 Å up to cycle 60, followed by a linear decrease at an etch rate of 3.1 Å per cycle.\"},{\"panel_id\":\"b\",\"text\":\"The inset shows a magnified view of the transition between half-cycles A and B at 30 cycles, where a thickness change from 207 Å to 204 Å is observed. The horizontal dashed lines indicate the initial film thickness before processing.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Film thikckness (Å) | Numer of pulses/cycles |\\n|---|---|\\n| 0 | 200 |\\n| 10 | 207 |\\n| 20 | 207 |\\n| 30 | 207 |\\n| 40 | 204 |\\n| 50 | 204 |\\n| 60 | 204 |\\n| 70 | 170 |\\n| 80 | 140 |\\n| 90 | 130 |\\n| 100 | 80 |\\n| 110 | 45 |\"},{\"panel_id\":\"b\",\"text\":\"| Film thikckness (Å) | Numer of pulses/cycles |\\n|---|---|\\n| 20 | 207 |\\n| 30 | 207 |\\n| 40 | 204 |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"After the half-cycle A, the film was exposed to a TMA dose, resulting in a decrease in thickness from 207 Å to 204 Å.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Between these data points actually one full ALE cycle has taken place. This is also the reason that the thickness difference is 3 angstroms.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"In subfigure (c), the film thickness exhibits a continuous, linear decrease with a constant negative slope, characteristic of a controlled layer-by-layer etching process. In contrast, subfigures (a) and (b) show saturation behavior where the thickness stabilizes (plateaus) after an initial change. This comparison highlights that the linear removal of material is exclusive to the cyclic alternation of the two reactants.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The Etch Per Cycle (EPC) is 3.1 Å/cycle.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At 30 cycles, there is a distinct step-change where the thickness drops from 207 Å to 204 Å (a reduction of 3 Å), marking the transition between the SF₆ plasma-only phase and the TMA/SF₆ gas phase.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Al2O3 is resistant to spontaneous etching by SF6 plasma, with etching observed only when alternating between SF6 plasma and TMA doses.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The exposure of Al2O3 to multiple SF6 plasma pulses without dosing TMA (half-cycle A) causes an initial 1A˚ increase in film thickness, which corresponds to the formation of a thin fluorinated surface layer. When subsequently exposing the substrate to TMA pulses (b), the thickness initially decreases by 3A˚  , and this is attributed to the removal of the modified AlF3 surface layer. After the initial 3A˚ of etching, the film thickness remains constant during further TMA doses, confirming that the TMA exposure is self-limiting.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"First, the substrates were exposed to 30 pulses of SF6 plasma (Half-cycle A), followed by 30 pulses of TMA (half-cycle B), and finally 50 ALE cycles (SF6 plasma + TMA dose).\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The exposure to SF6 plasma, which is related to the formation of a thin fluorinated layer.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The film thickness is decreasing with an etch rate of 3.1 Å per cycle.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"By performing a linear fit in the last section of the graph and taking the slope.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Synergy is high if one of the two doses on there own do not etch. This seems the case so the synergy is high.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"12 nm, or 120 angstrom would require 39 cycles.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The constant thickness indicates that fluorination alone does not cause spontaneous material removal. The fluorinated Al₂O₃ surface is chemically stable under these conditions and does not desorb volatile species by itself. This confirms that SF₆ plasma acts only as a surface activation step rather than an etching step.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"High thermal and chemical stability of the fluorinated surface.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The complete absence of thickness change during prolonged SF₆ plasma exposure proves that the fluorinated surface is highly stable against thermal decomposition and ion-induced damage. This thermal and chemical robustness means that no unintended etching or roughening occurs during the activation step, preserving atomic-scale smoothness. It is precisely this stable-yet-reactive fluorine-terminated surface that allows the subsequent TMA step to selectively and completely remove exactly one monolayer per cycle without over-etching.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The data in region (a) shows that exposure to SF₆ plasma alone results in a slight mass gain (thickness increase) due to surface fluorination but does not cause continuous etching. Similarly, region (b) shows that after the removal of the fluorinated layer, prolonged exposure to TMA does not result in further thickness loss. Together, these results demonstrate that neither reactant can spontaneously etch the bulk material on its own; etching only occurs through the synergistic, sequential application of both reactants.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The thickness increase is attributed to the fluorination of the Al₂O₃ surface, forming a lower-density AlF₃ (or AlOₓFᵧ) layer, which involves a volume expansion compared to the original oxide.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Half-cycle A (0–30 cycles): Exposure to 5 seconds of SF₆ plasma without any TMA dose.\\n\\nHalf-cycle B (30–60 cycles): Dosing with 0.5 seconds of TMA and SF₆ gas without striking a plasma.\\n\\nFull ALE Cycle (60–110 cycles): Alternating exposures of SF₆ plasma and TMA doses.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"During the Half-cycles (A and B), the film thickness remains largely static (staying at approximately 200–207 Å), indicating that the individual components alone do not sustain etching. In contrast, the Full ALE cycles (c) exhibit a linear decrease in thickness, demonstrating that the combination of both steps is required to achieve a continuous etch rate (specifically 3.1 Å per cycle).\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Yes. The image clearly shows a linear relationship between the etch depth and the number of cycles with a specific rate of 3.1 Å per cycle. This linearity implies that manufacturers can precisely calculate the total etch depth simply by counting cycles, which is ideal for high-precision semiconductor fabrication.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Alternating pulses of SF6 plasma and TMA show a linear dependence between the number of cycles and etch depth of the Al2O3 film (c).\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 207 Å.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"30 for both.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At a rate of 3.1 Å per cycle.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"A key requirement for ALE is that etching should only occur if the substrate is exposed to alternating half-cycles, which is why this experiment was conducted. The figure shows that the SF6 half-cycles did not affect the film thickness. When the substrate was exposed to the TMA half-cycles, a decrease was expected for the first half-cycle as the AlF3 layer formed during SF6 exposure should react with TMA. This was indeed the case and the substrate thickness did not change during the remainder of the TMA half-cycles. The data shown in this figure demonstrate that the process conditions meet this key requirement.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The nearly unchanged thickness during both half-cycles indicates that neither the SF₆ plasma step nor the TMA step continues to etch (or grow) the film on its own once the surface reaction is complete. This is consistent with self-limited half-reactions. A non-self-limiting process would show a sustained thickness drift during a single half-cycle.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The thickness remains essentially unchanged during the isolated SF₆ half-step and the isolated TMA half-step, indicating little net removal from either reactant alone. When the steps are alternated in repeated full cycles, the thickness decreases steadily with cycle number and yields a consistent etch per cycle. This difference shows that net etching is driven by the combined, cyclic chemistry of the two steps (reactant synergy).\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Etch rate is constant in the full ALE cycle region.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Right after cycle 30 (the transition into the TMA half-cycle).\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Alternating SF₆ plasma exposure, TMA dosing, ALE cycles combining  half-steps\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"During the individual half-cycles (SF₆ plasma or TMA exposure alone), the film thickness remains constant, indicating no net etching. In contrast, when the two steps are combined into a full ALE cycle, the film thickness decreases linearly with cycle number\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Approximately 3.1 Å per cycle.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Alternating SF6 plasma and TMA doses (full ALE cycles) result in a linear decrease of film thickness with cycle number, at an etch rate of approximately 3.1 Å per cycle, which is more systematic than the isolated changes in half-cycle A or B.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"SF6 plasma exposure creates a thin fluorinated surface layer (thickness increase), TMA dosing removes the fluorinated layer (thickness decrease), The self-limiting behavior of TMA indicates controlled surface reactions, The synergy of alternating cycles ensures consistent ALE etching\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The ALE process is highly efficient, as indicated by the linear decrease in film thickness during alternating SF6 plasma and TMA cycles, yielding an etch rate of approximately 3.1 Å per cycle. The initial thickness changes during individual half-cycles demonstrate controlled surface reactions, with the SF6 plasma forming a fluorinated layer and TMA removing it. The calculated ALE synergy of 99.9% confirms that the process reliably etches Al2O3 with precise control over material removal.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Yes. During SF6 plasma exposure (half-cycle A), the film thickness slightly increases due to the formation of a thin fluorinated surface layer. During TMA dosing (half-cycle B), the thickness decreases initially as the fluorinated layer is removed and then remains constant, confirming self-limiting behavior.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":583,"height":447}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/12/images/fig_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/12/images/fig_3.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/12/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/12/Isotropic plasma atomic layer etching of Al2O3 using a fluorine containing plasma and Al(CH3)3.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"12","first_classification_panel_id":"a","first_classification_label":"line chart","caption_source":"content.json:img_caption"},"width":583,"height":447,"image_format":"jpeg","image_sha256":"da7670607a127e6b6d4f27f24f593d4b2e497c38ea95132fd0fd012a3b9083be","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_13.jpg","caption":"Figure 13. AlN film thickness vs number of  $\\mathrm{XeF}_2 / \\mathrm{BCl}_3 / \\mathrm{H}_2\\mathrm{O}$  cycles on single-crystal AlN film at  $255^{\\circ}C$ . The measured AlN etch rate is  $0.49\\mathrm{\\AA}$  cycle.","id":"test/atomic-layer-etching/experimental-usecase/20/figure_13","sample_id":"atomic-layer-etching/experimental-usecase/20/figure_13","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows the decrease in AlN thickness over different number of ALE cycles with XeF₂/BCl₃/H₂O etching at 255°C, with an etch rate of 0.49 Å/cycle.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| XeF₂/BCl₃/H₂O Cycles | AIN Thickness (Å) |\\n|---|---|\\n| 0 | 2935 |\\n| 25 | 2923 |\\n| 50 | 2910 |\\n| 75 | 2900 |\\n| 100 | 2880 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"First, there is an XeF2 exposure, followed by BCl3 and H2O at 255 °C.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"As the number of XeF2/BCl3/H2O cycles increases at 255 °C, the film thickness decreases.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"An AlN thickness of 2900 Å can be achieved after 75 cycles.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.049 Å per cycle.\"}]}]","bbox":[{"panel_id":"a","x":4,"y":5,"width":657,"height":550}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/20/images/figure_13.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/20/images/figure_13.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":"line chart","caption_source":"content.json:img_caption"},"width":662,"height":558,"image_format":"jpeg","image_sha256":"719e70bea3fb7f6eed3ad38915a2072898017881971c6aa0e76b333fb1a72216","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_6.jpg","caption":"Figure 6. AlN film thickness vs number of  $\\mathrm{HF / BCl}_3$  cycles on single-crystal AlN film at  $298^{\\circ}C$  using (a) viscous flow of  $\\mathrm{BCl}_3$  precursor at  $500~\\mathrm{mTorr}$  for  $1\\mathrm{s}$  and (b) static exposure of  $\\mathrm{BCl}_3$  precursor at  $2.5$  Torr for  $20~\\mathrm{s}$ . The HF exposure was at  $100~\\mathrm{mTorr}$  for  $1\\mathrm{s}$ .","id":"test/atomic-layer-etching/experimental-usecase/20/figure_6","sample_id":"atomic-layer-etching/experimental-usecase/20/figure_6","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"},{"panel_id":"b","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart shows the thickness of AlN (Aluminum Nitride) over HF/BCl₃ cycles using a viscous flow of BCl3, with a gradual increase.\"},{\"panel_id\":\"b\",\"text\":\"The line chart shows the thickness of AlN (Aluminum Nitride) over HF/BCl₃ cycles with a static exposure of BCl3, which remains almost constant over the number of cycles.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| HF/BCl₃ Cycle | AlN Thickness (Å) |\\n|---|---|\\n| 0 | 2827.5 |\\n| 10 | 2828.0 |\\n| 20 | 2828.5 |\\n| 30 | 2829.0 |\"},{\"panel_id\":\"b\",\"text\":\"| HF/BCl₃ Cycle | AlN Thickness (Å) |\\n|---|---|\\n| 0 | 2743.0 |\\n| 10 | 2742.5 |\\n| 20 | 2742.5 |\\n| 30 | 2743.0 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The thickness of the AlN increases with an increase in the number of HF/BCl3 cycles under a viscous flow.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Two types of flow were studied: a viscous and static.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Static flow of BCl3.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"As can be seen in both subfigures, the changes in film thickness are small. For this reason, it can be concluded that no etching is achieved under these conditions.\"}]}]","bbox":[{"panel_id":"a","x":1,"y":5,"width":659,"height":216},{"panel_id":"b","x":3,"y":225,"width":656,"height":255}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/20/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/20/images/figure_6.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":"line chart","caption_source":"content.json:img_caption"},"width":661,"height":481,"image_format":"jpeg","image_sha256":"fb967773566e142afe6270f3c3b48981f6cedd8b0342bb0bf64a38aa1bd7ed2c","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_7.jpg","caption":"Figure 7. AlN film thickness vs number of  $\\mathrm{XeF}_2 / \\mathrm{BCl}_3$  cycles on single-crystal AlN film. (a) Viscous flow at  $298^{\\circ}C$  for  $\\mathrm{BCl}_3$  at  $500~\\mathrm{mTorr}$  for  $1\\mathrm{s}$  and  $\\mathrm{XeF}_2$  at  $50~\\mathrm{mTorr}$  for  $1\\mathrm{s}$ . (b) At  $255^{\\circ}C$ , static exposure of  $\\mathrm{BCl}_3$  at  $2.5$  Torr for  $20~\\mathrm{s}$  and viscous flow of  $\\mathrm{XeF}_2$  at  $50~\\mathrm{mTorr}$  for  $1\\mathrm{s}$ .","id":"test/atomic-layer-etching/experimental-usecase/20/figure_7","sample_id":"atomic-layer-etching/experimental-usecase/20/figure_7","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"},{"panel_id":"b","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"A line chart showing the thickness of AlN  versus the number of XeF2/BCl3 2 cycles at 298°C  and under viscous flow.\"},{\"panel_id\":\"b\",\"text\":\"A line chart showing the decrease in AlN thickness at 255°C during BCl₃ static etching with an etch rate of 0.62 Å/cycle.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| XeF₂/BCl₃ Cycles | AlN Thickness (Å) |\\n|---|---|\\n| 0 | 2911 |\\n| 10 | 2912 |\\n| 20 | 2912 |\\n| 30 | 2913 |\\n| 40 | 2914 |\"},{\"panel_id\":\"b\",\"text\":\"| XeF₂/BCl₃ Cycles | AlN Thickness (Å) |\\n|---|---|\\n| 0 | - |\\n| 100 | 2850 |\\n| 200 | 2810 |\\n| 300 | - |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"As the number of cycles increases, the thickness of the AlN shows a slight increase.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Viscous flow for both BCl3 and XeF2 at 298 °C, and static flow for BCl3 and viscous flow of XeF2 at 255 °C.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Subfigure b shows the BCl3 static flow at 255 °C.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"A static flow of BCl3 and viscous flow of XeF2 at 255 °C are required to etch the surface.\"}]}]","bbox":[{"panel_id":"a","x":6,"y":4,"width":651,"height":133},{"panel_id":"b","x":3,"y":140,"width":656,"height":339}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/20/images/figure_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/20/images/figure_7.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":"line chart","caption_source":"content.json:img_caption"},"width":661,"height":478,"image_format":"jpeg","image_sha256":"22ce8eae3d3d58206547f57df62b0e5aa59f14008f22870efdab9acaab18675c","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_84ec0afa1681d61dad16457f11e33680c3ceb0560ff64acb7baa4f5a6cea16ef.jpg","caption":"","id":"test/atomic-layer-etching/experimental-usecase/25/84ec0afa1681d61dad16457f11e33680c3ceb0560ff64acb7baa4f5a6cea16ef","sample_id":"atomic-layer-etching/experimental-usecase/25/84ec0afa1681d61dad16457f11e33680c3ceb0560ff64acb7baa4f5a6cea16ef","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"},{"panel_id":"b","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The etch rate of WO₃ increases with increasing BCl₃ exposure at a constant temperature and HF exposure, and saturates at 4 angstrom per cycle.\"},{\"panel_id\":\"b\",\"text\":\"The etch rate of WO₃ increases with increasing HF exposure at a constant temperature and BCl₃ exposure, and soft-saturates at 4.2 angstrom per cycle.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| BCl₃ Exposure (mTorr s) | WO₃ Etch Rate (Å/cycle) |\\n|---|---|\\n| 0 | 0 |\\n| 200 | 3.7 |\\n| 300 | 4.2 |\\n| 500 | 4.1 |\"},{\"panel_id\":\"b\",\"text\":\"| HF Exposure (mTorr s) | WO₃ Etch Rate (Å/cycle) |\\n|---|---|\\n| 0 | 0 |\\n| 100 | 2.7 |\\n| 150 | 3.8 |\\n| 200 | 4.2 |\\n| 300 | 4.3 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The etch rate would also 4 angstrom per cycle. This is the saturated value so further increasing the exposure would not result in a higher etch rate.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Usually an exponential function is used to fit the datapoints, where fit parameter in the exponent are related to dosing constants.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"No, there is no full saturation observed. The curve has not flattened fully.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"207 degrees celcius\"}]}]","bbox":[{"panel_id":"a","x":6,"y":8,"width":650,"height":543},{"panel_id":"b","x":4,"y":560,"width":652,"height":549}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/84ec0afa1681d61dad16457f11e33680c3ceb0560ff64acb7baa4f5a6cea16ef.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/25/images/84ec0afa1681d61dad16457f11e33680c3ceb0560ff64acb7baa4f5a6cea16ef.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":"line chart","caption_source":"not_found"},"width":658,"height":1114,"image_format":"jpeg","image_sha256":"24c051a3104aa52ebcb0c6be34020c79a62f5eb525915534aaeb9f6f75b83145","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_3.jpg","caption":"Fig. 3. (a) The progression in thickness as a function of the number of pulses or cycles for Hhfac-only pulses, consecutive Hhfac and  $\\mathrm{O_2}$  gas pulses, and ALE cycles incorporating both Hhfac and O radical pulses. (b) Real-time QCM measurements during 12 cycles of repeated Hhfac and  $\\mathrm{O_2}$  gas dosing, followed by 40 ALE cycles on  $\\mathrm{HfO_2}$  at a substrate temperature of  $300^{\\circ}\\mathrm{C}$ . The inset shows the magnified region from 1600 to  $1700\\mathrm{~s~}$","id":"test/atomic-layer-etching/experimental-usecase/31/fig_3","sample_id":"atomic-layer-etching/experimental-usecase/31/fig_3","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"},{"panel_id":"b","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This figure illustrates the effects of different ALE processing conditions on film etching and mass loss. Panel (a) compares HfO₂, ZrO₂, and Al₂O₃ across three treatments: Hhfac only, Hhfac + O₂ gas, and Hhfac + O radicals. In the first two cases, film thickness remains constant over 50 cycles, indicating negligible etching. However, when O radicals are introduced, HfO₂ film thickness decreases steadily with each cycle, while ZrO₂ and Al₂O₃ remain unchanged—demonstrating material selectivity and process activation via radical chemistry. Panel (b) tracks the mass change during a similar ALE sequence at 300 °C, with a sharp mass decrease occurring only after the O radical step is introduced. The inset FTIR spectra confirm removal of Hhfac and reaction byproducts, supporting the proposed etch mechanism.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of process cycles | HfO₂ (nm) |\\n|--------------------------|-----------|\\n| 0                        | 20        |\\n| 10                       | 18        |\\n| 20                       | 16        |\\n| 30                       | 14        |\\n| 40                       | 12        |\\n| 50                       | 10        |\"},{\"panel_id\":\"b\",\"text\":\"| Number of process cycles | HfO₂ (nm) |\\n|--------------------------|-----------|\\n| 0                        | 20        |\\n| 10                       | 18        |\\n| 20                       | 16        |\\n| 30                       | 14        |\\n| 40                       | 12        |\\n| 50                       | 10        |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Hhfac exposure, O₂ exposure, O radical introduction, Mass loss etching, FTIR byproduct detection.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Under the Hhfac only and Hhfac + O₂ gas conditions, the film thicknesses of HfO₂, ZrO₂, and Al₂O₃ remain constant, indicating that no etching occurs. However, when Hhfac is combined with O radicals, a steady decrease in HfO₂ film thickness is observed over multiple cycles, while ZrO₂ and Al₂O₃ remain unaffected. This trend suggests that O radicals are necessary to activate etching selectively in HfO₂, and the process is both material- and chemistry-dependent. The findings confirm that etch activation and selectivity are tunable through radical exposure.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"HfO₂.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":33,"y":0,"width":912,"height":557},{"panel_id":"b","x":5,"y":619,"width":942,"height":545}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/31/images/fig_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/31/images/fig_3.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":"line chart","caption_source":"content.json:image_caption"},"width":950,"height":1169,"image_format":"jpeg","image_sha256":"7be4a5c3fec51a05c4050bdb5844d34c307918a0f5c1c46f6b1a8eed2f7b9ef1","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_1.jpg","caption":"FIG. 1. (Color online) Film thickness measured by spectroscopic ellipsometry vs number of AlN ALE reaction cycles at  $275^{\\circ}C$  The etch rate increases after removing the  $\\mathrm{AlO_xN_y}$  layer on the AlN film.","id":"test/atomic-layer-etching/experimental-usecase/44/fig_1","sample_id":"atomic-layer-etching/experimental-usecase/44/fig_1","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line plot shows the etched thickness of a sample versus the number of Atomic Layer Etching (ALE) cycles at 275°C. The etch process reveals two distinct regions: an initial slow-etch region attributed to a surface AlOₓNᵧ layer, and a subsequent faster-etch region attributed to pure AlN.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Cycle Number | Thickness (Å) |\\n|-------|-------------|\\n|   0   |    491      |\\n| 100   |    479      |\\n| 200   |    472      |\\n| 301   |    460      |\\n| 400   |    435      |\\n| 500   |    400      |\\n| 600   |    355      |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Surface AlOₓNᵧ (oxynitride) layer (Cycles 0-300).\\n\\nTransition region to purer AlN (Cycles ~300-400).\\n\\nBulk, pure AlN film (Cycles 400-600).\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The AlOₓNᵧ region etches slowly (~0.07 Å/cycle), while pure AlN etches over 5x faster (~0.36 Å/cycle). This creates a significant trade-off: higher material purity (pure AlN) enables faster, more efficient etching, but the presence of even a thin, impure surface layer (AlOₓNᵧ) drastically reduces the effective etch rate and requires many cycles to clear, impacting process throughput.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The transition from an oxygen-containing surface layer (AlOₓNᵧ) to oxygen-free, pure AlN.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The \\\"knee\\\" marks the point where the inconsistent surface oxide is fully removed and etching of the target AlN begins. To achieve precise depth control, the process must first clear this variable surface layer. Therefore, the total etch cycle count must account for this non-productive \\\"overetch\\\" phase to ensure accurate etching into the pure AlN bulk.\"}]}]","bbox":[{"panel_id":"a","x":2,"y":4,"width":634,"height":533}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/44/images/fig_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/44/images/fig_1.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":"line chart","caption_source":"content.json:image_caption"},"width":639,"height":539,"image_format":"jpeg","image_sha256":"9e6057ee7ab88704be13324d3bef81ade7508c6d44f464e8f249e21f34b2a08e","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_3.jpg","caption":"FIG. 3. (Color online) Etch rate vs reactant exposure during AlN ALE at  $275^{\\circ}C$  a Etch rate vs HF exposure with  $\\mathrm{Sn(acac)}_2$  exposure fixed at 900 mTorr s and (b) etch rate vs  $\\mathrm{Sn(acac)}_2$  exposure with HF exposure fixed at  $1270\\mathrm{mTorr}$  s.","id":"test/atomic-layer-etching/experimental-usecase/44/fig_3","sample_id":"atomic-layer-etching/experimental-usecase/44/fig_3","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"},{"panel_id":"b","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The image represents the etch rate (Å/cycle) against the exposure (mTorr.s)  Etch rate vs HF exposure with Sn(acac)₂ exposure fixed at 900 mTorr. s\"},{\"panel_id\":\"b\",\"text\":\"This image shows Atomic Layer Etching (ALE) saturation curves for a sequential, two-step Sn(acac)₂ / HF thermal etching process of a metal oxide (likely SnO₂). It demonstrates the self-limiting nature of ALE: each reactant exposure saturates, leading to a constant etch rate per cycle once a sufficient exposure dose is reached. Etch rate vs Sn(acac)₂ exposure with HF exposure fixed at 1270 mTorr. s.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Exposure (mTorr·s) | Etch Rate (Å/cycle) |\\n|---|---|\\n| 0 | 0.0 |\\n| 300 | 0.1 |\\n| 600 | 0.2 |\\n| 900 | 0.3 |\\n| 1200 | 0.35 |\\n| 1500 | 0.35 |\\n| 1800 | 0.4 |\"},{\"panel_id\":\"b\",\"text\":\"| Exposure (mTorr·s) | Etch Rate (Å/cycle) |\\n|---|---|\\n| 0 | 0.0 |\\n| 200 | 0.14 |\\n| 400 | 0.23 |\\n| 600 | 0.26 |\\n| 800 | 0.32 |\\n| 1000 | 0.35 |\\n| 1200 | 0.35 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"HF Pulse: Exposure ≥ ~966 mTorr·s to form a saturated AlF₃ layer.\\n\\nPurge: 130 s with N₂ to remove HF and by-products.\\n\\nSn(acac)₂ Pulse: Exposure ≥ ~900 mTorr·s for complete ligand exchange.\\n\\nPurge: 130 s with N₂ to remove Sn(acac)₂ and volatile etch products.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"HF saturates more sharply and shows a risk of Chemical Vapor Etching  at high exposure, while Sn(acac)₂ saturates more gradually. This implies HF has a narrower safe process window; precise dose control is critical to avoid runaway etching. The trade-off is that HF's high reactivity enables fast surface fluorination, but demands stricter process control compared to the more forgiving Sn(acac)₂.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"A saturated, monolayer-thick AlF₃ surface layer.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The saturation plateau represents a stable process window in which the etch rate remains constant and is unaffected by small changes in reactant dose or pressure. This natural self-limiting behavior enables uniform etching across the wafer and ensures excellent run-to-run repeatability, which is far more important for achieving high yield than a faster but poorly controlled etch process.\"}]}]","bbox":[{"panel_id":"a","x":1,"y":0,"width":631,"height":357},{"panel_id":"b","x":0,"y":367,"width":626,"height":387}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/44/images/fig_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/experimental-usecase/44/images/fig_3.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":"line chart","caption_source":"content.json:image_caption"},"width":636,"height":758,"image_format":"jpeg","image_sha256":"bd493fe1376d063fb1178ed06e964c4b29180b36e3e78d24797ba803b6752b7d","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_11_4547e83349e7a9a1caf5fbbebd8da3978ec45a7a4a981e4142a90de337b8fc18.jpg","caption":"","id":"test/atomic-layer-etching/simulation-usecase/11/4547e83349e7a9a1caf5fbbebd8da3978ec45a7a4a981e4142a90de337b8fc18","sample_id":"atomic-layer-etching/simulation-usecase/11/4547e83349e7a9a1caf5fbbebd8da3978ec45a7a4a981e4142a90de337b8fc18","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"},{"panel_id":"b","label":"multiple line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows simulated mass change during 3 complete thermal ALE cycles of Al2O3 at 573 K. Each cycle exhibits a characteristic step pattern: mass increases during HF exposure (Step A, fluorination) by approximately 13 ng/cm², followed by mass decrease during TMA exposure (Step B, etching) by approximately 29 ng/cm², with 30 s purge times between steps.\"},{\"panel_id\":\"b\",\"text\":\"Compares mass change profiles for 3 ALE cycles at three temperatures (543 K, 573 K, 593 K). Higher temperatures result in faster cycle completion, with the 593 K process finishing around 175 s versus approximately 225 s at 543 K, demonstrating the Arrhenius temperature dependence of reaction kinetics.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Time (s) | Mass change (ng/cm²) |\\n|---|---|\\n| 0 | 0 |\\n| 2 | 13 |\\n| 30 | 13 |\\n| 35 | -16 |\\n| 60 | -16 |\\n| 70 | -3 |\\n| 90 | -3 |\\n| 100 | -32 |\\n| 125 | -32 |\\n| 135 | -19 |\\n| 155 | -19 |\\n| 165 | -48 |\\n| 190 | -48 |\"},{\"panel_id\":\"b\",\"text\":\"| Time (s) | T = 543 K (ng/cm²) | T = 573 K (ng/cm²) | T = 593 K (ng/cm²) |\\n|---|---|---|---|\\n| 0 | 0 | 0 | 0 |\\n| 10 | 13 | 13 | 13 |\\n| 28 | 13 | 13 | 13 |\\n| 34 | 13 | -16 | -16 |\\n| 40 | -16 | -16 | -16 |\\n| 60 | -16 | -16 | -16 |\\n| 66 | -16 | -3 | -3 |\\n| 80 | -16 | -3 | -3 |\\n| 85 | -3 | -3 | -3 |\\n| 92 | -3 | -3 | -3 |\\n| 98 | -3 | -32 | -32 |\\n| 115 | -3 | -32 | -32 |\\n| 125 | -32 | -32 | -19 |\\n| 130 | -32 | -19 | -19 |\\n| 150 | -32 | -19 | -19 |\\n| 158 | -32 | -19 | -48 |\\n| 165 | -19 | -48 | -48 |\\n| 190 | -19 | -48 | -48 |\\n| 205 | -48 | - | - |\\n| 240 | -48 | - | - |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Each ALE cycle consists of two sequential half-reactions with distinct mass changes. During Step A (HF exposure), fluorine atoms replace oxygen atoms on the Al2O3 surface, forming AlF3 while releasing water, resulting in a net mass increase of approximately 13 ng/cm² because fluorine is heavier than oxygen. During Step B (TMA exposure), trimethylaluminum reacts with the fluorinated surface through ligand exchange, producing volatile dimethylaluminum fluoride (DMAF) that desorbs, causing a mass decrease of approximately 29 ng/cm². The flat regions correspond to the 30 s purge periods.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The horizontal plateaus represent the 30 s purge periods where inert gas removes unreacted precursors and byproducts from the reactor before the next half-cycle begins.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Mass change simulations provide critical parameters for process optimization without costly experiments. They reveal half-cycle times needed to achieve saturation at different operating conditions, enabling selection of temperature and pressure combinations that balance throughput with process quality. The data also validates whether the self-limiting behavior characteristic of ALE is achieved, as indicated by the flat plateaus. Additionally, the net mass loss per cycle can be correlated with etch per cycle measurements to verify the microscopic model accuracy.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Higher temperatures significantly reduce the total cycle time, with 3 cycles completing in approximately 175 s at 593 K compared to approximately 225 s at 543 K.\"}]}]","bbox":[{"panel_id":"a","x":1,"y":4,"width":603,"height":506},{"panel_id":"b","x":3,"y":514,"width":600,"height":504}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/images/4547e83349e7a9a1caf5fbbebd8da3978ec45a7a4a981e4142a90de337b8fc18.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/images/4547e83349e7a9a1caf5fbbebd8da3978ec45a7a4a981e4142a90de337b8fc18.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/Microscopic and data-driven modeling andoperation.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"11","first_classification_panel_id":"a","first_classification_label":"line chart","caption_source":"not_found"},"width":608,"height":1022,"image_format":"jpeg","image_sha256":"da6365720babd13569196b6c704148af4dbf8dbd48923b3be812180277fe59bb","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_11_538f430c664805721e436d2436803b1712d3747d91728885c13fa415252c98a7.jpg","caption":"","id":"test/atomic-layer-etching/simulation-usecase/11/538f430c664805721e436d2436803b1712d3747d91728885c13fa415252c98a7","sample_id":"atomic-layer-etching/simulation-usecase/11/538f430c664805721e436d2436803b1712d3747d91728885c13fa415252c98a7","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Convergence test for the kinetic energy cutoff (ecutwfc) parameter in DFT calculations for θ-Al2O3. The total energy decreases from approximately -2388 Ry at 20 Ry cutoff to about -2454 Ry, converging when ecutwfc reaches 50 Ry. This determines the minimum plane wave basis set size needed for accurate electronic structure calculations.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Kinetic energy cut-off (Ry) | Total energy (Ry) |\\n|---|---|\\n| 20 | -2390 |\\n| 25 | -2430 |\\n| 30 | -2450 |\\n| 35 | -2450 |\\n| 40 | -2450 |\\n| 45 | -2450 |\\n| 50 | -2450 |\\n| 55 | -2450 |\\n| 60 | -2450 |\\n| 65 | -2450 |\\n| 70 | -2450 |\\n| 75 | -2450 |\\n| 80 | -2450 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The kinetic energy cutoff (ecutwfc) determines how many plane waves are included in the basis set expansion of the electronic wavefunctions. A higher cutoff includes more plane waves, improving accuracy but increasing computational cost. The optimization finds the minimum cutoff value where the total energy is converged, balancing accuracy against efficiency. Using a value below convergence yields unreliable energies, while using excessively high values wastes computational resources without improving results.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The total energy difference is approximately 66 Ry, decreasing from about -2388 Ry at 20 Ry cutoff to about -2454 Ry at 50 Ry cutoff.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The x-axis represents the kinetic energy cutoff (ecutwfc) in Rydberg units, which controls the number of plane waves used in the DFT basis set.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"A value of 50 Ry was selected because it is the point where total energy convergence is achieved, meaning further increases provide no significant improvement in accuracy.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":5,"width":581,"height":422}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/images/538f430c664805721e436d2436803b1712d3747d91728885c13fa415252c98a7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/images/538f430c664805721e436d2436803b1712d3747d91728885c13fa415252c98a7.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/Microscopic and data-driven modeling andoperation.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"11","first_classification_panel_id":"a","first_classification_label":"line chart","caption_source":"not_found"},"width":586,"height":425,"image_format":"jpeg","image_sha256":"b52b6e0f9ffa02231d7717c6ad46ab79cd1861f6a81421f6ccd1dcf48e5f24bb","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_11_f29fb0936e8517e60638237bcc9fde1e5c001e80bfc85cf97feddcf84eb574c3.jpg","caption":"","id":"test/atomic-layer-etching/simulation-usecase/11/f29fb0936e8517e60638237bcc9fde1e5c001e80bfc85cf97feddcf84eb574c3","sample_id":"atomic-layer-etching/simulation-usecase/11/f29fb0936e8517e60638237bcc9fde1e5c001e80bfc85cf97feddcf84eb574c3","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Convergence test for the kinetic energy cutoff for charge density and potential (ecutrho) in DFT calculations for θ-Al2O3 (2̄ 0 1). Unlike ecutwfc, the total energy increases (becomes less negative) with increasing ecutrho, rising from approximately -2454.038 Ry at 100 Ry to a converged value around -2454.006 Ry at ecutrho=200 Ry. This parameter controls the grid density for representing charge density and potentials.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Kinetic energy cut-off for charge density and potential (Ry) | Total energy (Ry) |\\n|---|---|\\n| 100 | -0.040 |\\n| 150 | -0.010 |\\n| 200 | -0.005 |\\n| 250 | -0.005 |\\n| 300 | -0.005 |\\n| 350 | -0.005 |\\n| 400 | -0.005 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The two parameters show opposite convergence trends. For ecutwfc, the total energy decreases (becomes more negative) with increasing cutoff values. For ecutrho, the total energy increases (becomes less negative) as the cutoff increases. Both parameters eventually reach convergence plateaus, but they approach their final values from opposite directions. This reflects their different roles in the calculation: ecutwfc controls wavefunction expansion while ecutrho controls charge density representation.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Ecutrho controls the kinetic energy cutoff for representing the charge density and potential on the real-space grid.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"A value of 200 Ry was selected because the total energy no longer changes significantly beyond this point, indicating convergence.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"In this study, ecutrho (200 Ry) is set to 4 times ecutwfc (50 Ry), which is a common ratio used for PAW pseudopotential calculations to ensure adequate charge density representation.\"}]}]","bbox":[{"panel_id":"a","x":6,"y":7,"width":571,"height":388}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/images/f29fb0936e8517e60638237bcc9fde1e5c001e80bfc85cf97feddcf84eb574c3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/images/f29fb0936e8517e60638237bcc9fde1e5c001e80bfc85cf97feddcf84eb574c3.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/Microscopic and data-driven modeling andoperation.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"11","first_classification_panel_id":"a","first_classification_label":"line chart","caption_source":"not_found"},"width":578,"height":394,"image_format":"jpeg","image_sha256":"a34587de75f1d637b6773a4e056e0b9467bafe9297effb84855ff40a846f5ff3","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_11_fig_1.jpg","caption":"(c) The  $k$ -point optimization of  $\\theta -\\mathrm{Al}_2\\mathrm{O}_3$  (2 0 1). Fig. 1 - The total energy converges when ecutwfc is 50 (a), ecutrho is 200 (b), and the number of  $k$ -points is 4 (c), respectively.","id":"test/atomic-layer-etching/simulation-usecase/11/fig_1","sample_id":"atomic-layer-etching/simulation-usecase/11/fig_1","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Convergence test for the k-point grid density in DFT calculations for θ-Al2O3 (2̄ 0 1). The total energy decreases from approximately -2451 Ry at k-point=1 to about -2454.7 Ry, achieving convergence at k-point=4. This parameter determines the sampling density for Brillouin zone integration in the electronic structure calculations.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| K-point | Total energy (RY) |\\n|---|---|\\n| 1 | -2451.0 |\\n| 2 | -2454.5 |\\n| 3 | -2454.0 |\\n| 4 | -2454.0 |\\n| 5 | -2454.0 |\\n| 6 | -2454.0 |\\n| 7 | -2454.0 |\\n| 8 | -2454.0 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"K-points define the sampling grid for integrating electronic properties over the Brillouin zone in reciprocal space. Periodic DFT calculations require evaluating wavefunctions at multiple k-points to accurately compute total energies and other properties. Insufficient k-point sampling leads to inaccurate results, while excessive sampling increases computational cost without improving accuracy. The convergence test identifies the minimum k-point density that yields reliable results.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The total energy decreases by approximately 3.7 Ry, from about -2451 Ry at k-point=1 to approximately -2454.7 Ry at convergence.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Convergence is achieved at k-point=4, after which the total energy remains essentially constant.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"These parameters control different aspects of calculation accuracy: ecutwfc determines basis set completeness while k-points determine Brillouin zone sampling, so each must be independently optimized.\"}]}]","bbox":[{"panel_id":"a","x":6,"y":7,"width":572,"height":418}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/images/fig_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/images/fig_1.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/Microscopic and data-driven modeling andoperation.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"11","first_classification_panel_id":"a","first_classification_label":"line chart","caption_source":"content.json:img_caption"},"width":586,"height":425,"image_format":"jpeg","image_sha256":"df7d68a91522c9fd93e91498b85c4ca14ec7ced6cb079a792e0ed89cabd8de6c","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_11_fig_5.jpg","caption":"Fig. 5 - Film thickness with respect to number of cycles. It is assumed that the initial thickness is 100 Å. An etch thickness per cycle of 0.46 Å/cycle is applied.","id":"test/atomic-layer-etching/simulation-usecase/11/fig_5","sample_id":"atomic-layer-etching/simulation-usecase/11/fig_5","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The thickness decreases linearly with increasing number of cycles.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of cycle | Thickness (angstrom) |\\n|----------------|---------------------|\\n| 0              | 100                 |\\n| 25             | 90                  |\\n| 50             | 80                  |\\n| 75             | 70                  |\\n| 100            | 60                  |\\n| 125            | 50                  |\\n| 150            | 40                  |\\n| 175            | 30                  |\\n| 200            | 20                  |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The y-axis represents the Al2O3 film thickness measured in angstroms (Å).\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The linear relationship reflects the self-limiting nature of the ALE process. Each cycle removes a consistent amount of material (0.46 Å) regardless of how many cycles have been performed, because the surface reactions saturate after complete monolayer conversion. This contrasts with continuous etching processes where the etch rate may vary with film thickness or process time. The linearity is a defining characteristic of ALE and enables precise, predictable thickness control.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Approximately 217 cycles would be needed to completely etch the 100 Å film (100 Å ÷ 0.46 Å/cycle ≈ 217 cycles).\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The etch per cycle (EPC) value of 0.46 Å/cycle is taken from experimental measurements reported in the literature, as the kMC model itself does not directly compute EPC.\"}]}]","bbox":[{"panel_id":"a","x":4,"y":6,"width":606,"height":440}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/images/fig_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/images/fig_5.json","source_content":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/content.json","source_pdf":["icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/11/Microscopic and data-driven modeling andoperation.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"11","first_classification_panel_id":"a","first_classification_label":"line chart","caption_source":"content.json:img_caption"},"width":611,"height":447,"image_format":"jpeg","image_sha256":"def13787f39744adad77ed3b35b3560d775b543490439eb20f2349f6fec870c8","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_4.jpg","caption":"Figure 4. Data for (a) the Ta etch rate in  $\\mathrm{Cl}_2$  plasma (i.e., Step A only) as a function of temperature, showing that etching is suppressed at  $0^{\\circ}\\mathrm{C}$ . (b) Etch per cycle (ETC) as a function of ion energy for the Ta substrate at  $0^{\\circ}\\mathrm{C}$  for ALE (in gray), averaged over 40 full ALE cycles, and for Ar sputter (i.e., Step B only) (in blue).","id":"test/atomic-layer-etching/simulation-usecase/21/figure_4","sample_id":"atomic-layer-etching/simulation-usecase/21/figure_4","subset":"line-chart","split":"test","classification":[{"panel_id":"a","label":"line chart"},{"panel_id":"b","label":"multiple line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Ta chemical etching in Cl₂ plasma (Step A only) is suppressed at 0 °C and increases with wafer temperature, reaching ~2 nm/s at 60 °C. This indicates strong thermal activation of the Step-A chemistry.\"},{\"panel_id\":\"b\",\"text\":\"At 0 °C, Ta ALE shows an ALE window with EPC near ~0.35–0.45 nm/cycle across roughly 20–90 eV, followed by a rise at higher energy. Ar-only sputtering increases steadily with ion energy and lacks a plateau, consistent with non-self-limiting removal.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Wafer temperature (°C) | Etch rate (nm/s) |\\n|---|---|\\n| 0  | ~0.00 |\\n| 10 | ~0.10 |\\n| 30 | ~0.50 |\\n| 60 | ~2.00 |\"},{\"panel_id\":\"b\",\"text\":\"| Ion energy (eV) | EPC in ALE mode (nm/cycle) | EPC in Ar-only mode (nm/cycle) |\\n|---|---|---|\\n| 0   | ~0.00 | ~0.00 |\\n| 20  | ~0.42 | ~0.00 |\\n| 40  | ~0.36 | ~0.02 |\\n| 60  | ~0.44 | ~0.06 |\\n| 90  | ~0.43 | ~0.25 |\\n| 115 | ~0.65 | ~0.40 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At 0 °C the surface reaction pathway for chlorination/volatile product formation is kinetically inhibited, so net chemical etching is negligible.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 10 °C.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"-EPC remains approximately constant (~0.35 nm/cycle) between ~20–80 eV.\\n-Increasing ion energy within this range does not increase removal.\\n-The Ar-only curve lacks a plateau, underscoring the chemical selectivity of ALE.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The stable EPC plateau in the ALE curve indicates self-limiting removal once the surface has been chemically modified, with ion energy serving only to trigger detachment. The monotonic increase in the Ar-only curve reflects purely physical sputtering, where removal scales continuously with energy. This contrast shows that ALE achieves controlled, low-damage removal by separating chemical preparation from physical activation.\"}]}]","bbox":[{"panel_id":"a","x":50,"y":7,"width":486,"height":308},{"panel_id":"b","x":47,"y":351,"width":492,"height":293}],"source":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/21/images/figure_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/test/gold_standard_test_set/atomic-layer-etching/simulation-usecase/21/images/figure_4.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":"line chart","caption_source":"content.json:img_caption"},"width":539,"height":644,"image_format":"jpeg","image_sha256":"168e06982f9d9bd48181fbbd8b492b2b66ab94aaed4cbe74355cbb0efbf6f18c","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"}