subsets/scatter-plot/train/metadata.jsonl

{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_12_fig_8.jpg","caption":"Fig. 8. Tungsten film thickness deposited after three AB cycles versus number of  $\\mathrm{WF}_6$  pulses at  $425 \\mathrm{K}$ . The  $\\mathrm{Si}_2\\mathrm{H}_6$  exposure of  $40 \\mathrm{Si}_2\\mathrm{H}_6$  pulses during each AB cycle was sufficient for a complete  $\\mathrm{Si}_2\\mathrm{H}_6$  half-reaction.","id":"train/atomic-layer-deposition/experimental-usecase/12/fig_8","sample_id":"atomic-layer-deposition/experimental-usecase/12/fig_8","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line plot shows the tungsten film thickness obtained after three AB cycles as a function of the number of WF₆ pulses at 425 K. The thickness reaches saturation after about 10 WF₆ pulses, indicating that additional pulses do not increase growth. The data also confirms that using 40 Si₂H₆ pulses in each AB cycle is sufficient to complete the Si₂H₆ half-reaction.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of WF6 Pulses | Tungsten Film Thickness (Å) |\\n|---|---|\\n| 0 | 0 |\\n| 1 | 3 |\\n| 5 | 7 |\\n| 10 | 7.2 |\\n| 15 | 7.2 |\\n| 20 | 7.2 |\\n| 25 | 6.5 |\\n| 30 | 7.2 |\\n| 40 | 7.2 |\\n| 50 | 7.2 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"To account for minor process variations, a safe setpoint would be slightly above the threshold, such as 12–15 pulses. Programming anything less than 10 pulses risks incomplete surface coverage, while significantly more (e.g., >20) wastes precursor without adding thickness.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"5 Pulses (Slope Region): High precursor efficiency (low waste), but high risk of non-uniformity because growth is sensitive to local gas flux.\\n\\n50 Pulses (Deep Saturation): Confirms maximum uniformity and full coverage, but suffers from very low efficiency due to high precursor usage/waste.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The plateau signifies that all available reactive surface sites (specifically the Si-H bonds established in the previous step) have been consumed and replaced by W-F species. Since WF6 cannot react with itself or the fluorinated surface it creates, the reaction chemically terminates. This site-limitation factor ensures the film thickness is defined strictly by the number of cycles, not the total precursor gas exposure.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The GPC is ~2.5 Å/cycle, so 8 cycles are required for a 20 Å liner.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":598,"height":608}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/12/images/fig_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/12/images/fig_8.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/12/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/12/J.W. Klaus et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"12","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":600,"height":609,"image_format":"jpeg","image_sha256":"4416339a5a164def601a15143b0bd86e76c005bf72af69681c9c27e054aec64b","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_12_fig_9.jpg","caption":"Fig. 9. Tungsten film thickness deposited after three AB cycles versus number of  $\\mathrm{Si}_2\\mathrm{H}_6$  pulses at  $425 \\mathrm{K}$ . The  $\\mathrm{WF}_6$  exposure of nine  $\\mathrm{WF}_6$  pulses during each AB cycle was sufficient for a complete  $\\mathrm{WF}_6$  half-reaction.","id":"train/atomic-layer-deposition/experimental-usecase/12/fig_9","sample_id":"atomic-layer-deposition/experimental-usecase/12/fig_9","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This line  chart represents the tungsten(W) film thickness after 3 ALD- AB cycles as a function of the number of Si₂H₆ pulses per cycle (B-step), while the WF₆ exposure (A-step) is held constant at 9 pulses per cycle. It demonstrates that film growth requires a minimum number of Si₂H₆ pulses to complete the reaction cycle.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of Si₂H₆ Pulses | Tungsten Film Thickness (Å) |\\n|---|---|\\n| 0 | 0 |\\n| 4 | 5 |\\n| 12 | 7.5 |\\n| 24 | 8.1 |\\n| 48 | 7.4 |\\n| 60 | 8.1 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"WF₆ Pulse\\n\\nPurge\\n\\nRepeat steps 1-2 until 9 total WF₆ pulses are delivered.\\n\\nSi₂H₆ Pulse\\n\\nPurge\\n\\nRepeat steps 4-5 a total of 30 times.\\nThe 9 WF₆ pulses ensure the surface is saturated, guaranteeing a complete and reproducible A-step half-reaction regardless of the subsequent B-step condition, isolating the effect of the Si₂H₆ dose.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Diborane (B₂H₆) is likely better for minimizing oxygen. B₂H₆ is a stronger reducing agent that can getter trace oxygen and water more effectively than Si₂H₆. The trade-off is that boron itself can incorporate as an impurity in the W film, potentially altering its electrical resistivity and stability, whereas silicon incorporation from Si₂H₆ may form a more benign tungsten silicide.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Thickness increases with Si₂H₆ pulses, then saturates. From 0 to ~10 pulses, thickness rises sharply because each pulse provides more reductant to convert the W–F surface layer into a metallic W film and regenerate the Si–H sites needed for the next WF₆ pulse. Insufficient Si₂H₆ leaves the surface partially fluorinated, preventing full tungsten deposition in the next cycle.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"This implies the process must use a saturating Si₂H₆ dose (>20 pulses/cycle) to ensure complete surface reaction. To grow 100 Å, you would run the calculated number of full ALD cycles based on the saturated growth per cycle (~20 Å/3 cycles = ~6.7 Å/cycle). Consistency and conformality are achieved only when both half-reactions (9 WF₆ pulses, >20 Si₂H₆ pulses) are operated in their saturated regimes.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":3,"width":599,"height":598}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/12/images/fig_9.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/12/images/fig_9.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/12/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/12/J.W. Klaus et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"12","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":600,"height":603,"image_format":"jpeg","image_sha256":"8cbaa089799ee74d9375dffd2f4cc61dfa62031c8a346286999a72282fce7752","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_16_fig_5.jpg","caption":"FIG. 5. Film growth rates as a function of  $\\mathrm{Zr}$  precursor pulse length with ozone as the oxygen source. The pulsing sequence was  $\\mathrm{x}$ $\\mathrm{slx} + 0.5$  sl1.0 sl1.5 s for  $\\mathrm{Zr}$  pulselpurgel  $\\mathrm{O_3}$  pulselpurge.","id":"train/atomic-layer-deposition/experimental-usecase/16/fig_5","sample_id":"atomic-layer-deposition/experimental-usecase/16/fig_5","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the growth rate of a material, when ozone is used as a precursor, as a function of Zr pulse length at different temperatures: Zr(Me₅Cp)(TEA) at 375°C and 300°C, Zr(Cp)(ᵗBuDAD)(OⁱPr)  and Zr(MeCp)(TMEA) at 250°C. The growth rate for Zr(MeCp)(TMEA) continues to increase, others show self-limiting behaviour.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Zr pulse length (s) | Growth rate (Å/cycle) | Reagent | Temperature (°C) |\\n|---|---|---|---|\\n| 0,5 | 0,28 | Zr(Me_5Cp)(TEA) | 300 |\\n| 1,0 | 0,30 | Zr(Me_5Cp)(TEA) | 300 |\\n| 1,5 | 0,30 | Zr(Me_5Cp)(TEA) | 300 |\\n| 2,0 | 0,31 | Zr(Me_5Cp)(TEA) | 300 |\\n| 2,5 | 0,32 | Zr(Me_5Cp)(TEA) | 300 |\\n| 3,0 | 0,32 | Zr(Me_5Cp)(TEA) | 300 |\\n| 0,5 | 0,34 | Zr(Me_5Cp)(TEA) | 375 |\\n| 1,0 | 0,37 | Zr(Me_5Cp)(TEA) | 375 |\\n| 1,5 | 0,39 | Zr(Me_5Cp)(TEA) | 375 |\\n| 2,0 | 0,41 | Zr(Me_5Cp)(TEA) | 375 |\\n| 2,5 | 0,42 | Zr(Me_5Cp)(TEA) | 375 |\\n| 3,0 | 0,43 | Zr(Me_5Cp)(TEA) | 375 |\\n| 3,5 | 0,43 | Zr(Me_5Cp)(TEA) | 375 |\\n| 1,0 | 0,40 | Zr(Cp)(tBuDAD)(O^iPr) | 250 |\\n| 1,5 | 0,43 | Zr(Cp)(tBuDAD)(O^iPr) | 250 |\\n| 2,0 | 0,50 | Zr(Cp)(tBuDAD)(O^iPr) | 250 |\\n| 2,5 | 0,49 | Zr(Cp)(tBuDAD)(O^iPr) | 250 |\\n| 3,0 | 0,48 | Zr(Cp)(tBuDAD)(O^iPr) | 250 |\\n| 1,0 | 0,57 | Zr(MeCp)(TMEA) | 250 |\\n| 2,0 | 0,66 | Zr(MeCp)(TMEA) | 250 |\\n| 2,5 | 0,69 | Zr(MeCp)(TMEA) | 250 |\\n| 3,0 | 0,71 | Zr(MeCp)(TMEA) | 250 |\\n| 3,5 | 0,74 | Zr(MeCp)(TMEA) | 250 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"No. The growth rate continues to rise as more precursor is dosed, the deposition is therefore  likely influenced by CVD-like  reactions or precursor decomposition. A self-limiting ALD growth would show a plateau.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"2 seconds.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The growth rate decreases by ~ 0.1 Å/cycle.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Zr(Cp)(ᵗBuDAD)(OⁱPr)  and Zr(Me₅Cp)(TEA).\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":656,"height":553}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/16/images/fig_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/16/images/fig_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/16/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/16/Sanni Seppala et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"16","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":656,"height":555,"image_format":"jpeg","image_sha256":"5a29599c0702375ef9f357115da168d1219308d92e6aba95e424c3eef004fadf","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_2_figure_1.jpg","caption":"Figure 1. Dependence of the platinum film growth rate on the  $\\mathrm{MeCpPtMe}_3$  and air pulse times. The air pulse time in the  $\\mathrm{MeCpPtMe}_3$  experiments was  $1.5\\mathrm{s}$  and the  $\\mathrm{MeCpPtMe}_3$  pulse time in the air pulse experiments was  $0.5\\mathrm{s}$ .","id":"train/atomic-layer-deposition/experimental-usecase/2/figure_1","sample_id":"atomic-layer-deposition/experimental-usecase/2/figure_1","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot compares the ALD growth rate of Pt as a function of pulse duration for two independent series: varying MeCpPtMe₃ pulse time at fixed air exposure and varying air pulse time at fixed MeCpPtMe₃ exposure. The MeCpPtMe₃-pulse series shows a modest increase from ~0.44 to ~0.50 Å/cycle with longer precursor pulses, while the air-pulse series shows growth rates in the ~0.36–0.46 Å/cycle range, largely insensitive to air pulse duration beyond ~1.5 s. Error bars highlight the experimental variation for each condition.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Pulse time (s) | MeCpPtMe₃ pulse series (Å cycle⁻¹) | Air pulse series (Å cycle⁻¹)  |\\n|----------------|-----------------------------------|-------------------------------|\\n| 0.0            | ~0.44                             | -                             |  \\n| 0.5            | ~0.45                             | -                             |\\n| 1.0            | ~0.50                             | ~0.36                         |\\n| 1.5            | ~0.49                             | ~0.45                         |\\n| 2.0            | -                                 | ~0.45                         |\\n| 2.5            | -                                 | ~0.42                         |\\n| 3.0            | -                                 | ~0.46                         |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Increasing the MeCpPtMe₃ pulse duration increases the likelihood that all available surface reactive sites are exposed to the Pt precursor before purging. This reduces precursor starvation and moves the system toward surface saturation, which is reflected by the gradual increase in growth rate from ~0.44 to ~0.50 Å/cycle. Once saturation is approached, further increases in pulse time yield diminishing returns, consistent with self-limiting ALD behavior.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The MeCpPtMe₃ pulse-time series shows a stronger dependence, implying that precursor adsorption is more rate-limiting than the air oxidation step under these conditions.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The relatively small and overlapping error bars indicate that the observed changes in growth rate are systematic rather than dominated by experimental noise. This strengthens confidence that the plateaus correspond to true saturation behavior rather than statistical scatter, particularly for the air-pulse series where growth rates remain nearly constant over a wide pulse-time range.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"By identifying the minimum pulse durations required to achieve saturation—about 1.0 s for MeCpPtMe₃ and 1.5 s for air—process engineers can avoid unnecessarily long pulses that increase cycle time without increasing growth per cycle. Operating at these saturation thresholds maximizes throughput while maintaining precise thickness control and film uniformity.\"}]}]","bbox":[{"panel_id":"a","x":1,"y":4,"width":560,"height":424}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/images/figure_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/images/figure_1.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/Aaltonen et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"2","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":561,"height":428,"image_format":"jpeg","image_sha256":"0828deda4229832f1d1f0bbab33983d4d38bdb5f5318eb3261a0bc2bd36e568f","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_2_figure_3.jpg","caption":"Figure 3. Thickness profiles of platinum films grown with short and long  $\\mathrm{MeCpPtMe_3}$  pulses of 0.2 and  $1.5\\mathrm{s}$ , respectively. The profile is measured along the gas flow direction at various distances from the leading edge of the substrate, that is, the edge closest to the precursor inlet.","id":"train/atomic-layer-deposition/experimental-usecase/2/figure_3","sample_id":"atomic-layer-deposition/experimental-usecase/2/figure_3","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"b\",\"text\":\"Figure shows film thickness profiles for platinum ALD using short (0.2 s) and long (1.5 s) MeCpPtMe₃ precursor pulses as a function of substrate position along the gas-flow direction. Long precursor pulses yield consistently thicker films, reflecting greater precursor availability across the surface. Both profiles show a mild decrease in thickness with distance from the leading edge, indicating precursor depletion during transport. The small error bars suggest good reproducibility for both pulse conditions.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Distance from leading edge (cm) | Long-pulse thickness (nm) | Short-pulse thickness (nm) |\\n|---------------------------------|---------------------------|----------------------------|\\n| 1                               | ~78 ± 5                   | ~66 ± 5                    |\\n| 2                               | ~76 ± 5                   | ~64 ± 5                    |\\n| 3                               | ~72 ± 5                   | ~63 ± 5                    |\\n| 4                               | ~68 ± 5                   | ~60 ± 5                    |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Longer MeCpPtMe₃ pulses increase the total precursor dose delivered to the substrate, allowing surface saturation to be reached over a larger downstream distance before precursor depletion occurs. This leads to higher and more uniform film thickness along the gas-flow direction. Shorter pulses provide insufficient precursor flux, causing earlier depletion and reduced surface coverage downstream, which manifests as thinner films and a stronger spatial gradient.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The short-pulse condition shows a stronger thickness decay.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The consistently greater thickness implies more complete surface saturation during each ALD cycle when long pulses are used. This suggests that the process is less transport-limited and that adsorption kinetics dominate over precursor delivery limitations, resulting in a higher effective growth per cycle across the substrate.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The figure demonstrates that longer precursor pulses improve thickness uniformity by overcoming transport limitations. For large-area wafers or high-aspect-ratio features, pulse durations must be sufficiently long to ensure uniform precursor delivery throughout the structure. These insights guide pulse-time optimization to balance throughput against uniformity requirements in industrial ALD reactors.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":2,"width":535,"height":371}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/images/figure_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/images/figure_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/2/Aaltonen et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"2","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":539,"height":375,"image_format":"jpeg","image_sha256":"a844597aa9312b4ebc9d80f5e9d8fd9fce5f4051797867321208966707912d8f","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_26_figure_6.jpg","caption":"Figure 6. Thickness of ALD ITO films versus number of cycles determined using VASE for films deposited on Si(100) at  $275^{\\circ}C$  using  $5\\%$ $\\mathrm{SnO_2}$  cycles.","id":"train/atomic-layer-deposition/experimental-usecase/26/figure_6","sample_id":"atomic-layer-deposition/experimental-usecase/26/figure_6","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart displays a direct linear relationship between the number of ALD cycles and ITO film thickness, demonstrating consistent film growth per cycle.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| ALD Cycles | ITO Thickness (Å) |\\n|------------|-------------------|\\n| 0          | 0                 |\\n| 100        | 200               |\\n| 200        | 400               |\\n| 300        | 600               |\\n| 400        | 800               |\\n| 500        | 1000              |\\n| 600        | 1200              |\\n| 700        | 1400              |\\n| 800        | 1600              |\\n| 900        | 1800              |\\n| 1000       | 2000              |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The ITO thickness increases linearly with the number of ALD cycles, indicating a consistent growth rate.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Number of ALD cycles, ITO Thickness.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":586,"height":570}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/Jeffrey W. Elam et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"26","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":586,"height":570,"image_format":"jpeg","image_sha256":"191f29ab6a4829e33b0ffaeca0d31ba945ff7dbf89798569da4f3e6dc5e424dd","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_26_figure_8.jpg","caption":"Figure 8. Resistivity of ALD ITO films prepared on glass using  $5\\%$ $\\mathrm{SnO_2}$  cycles measured by four-point probe versus (a) number of cycles at  $275^{\\circ}C$  and (b) deposition temperature using 300 cycles.","id":"train/atomic-layer-deposition/experimental-usecase/26/figure_8","sample_id":"atomic-layer-deposition/experimental-usecase/26/figure_8","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows a sharp initial decrease in resistivity with increasing ALD cycles and ITO film thickness, plateauing at around 0.002 Ω·cm beyond 500 Å.\"},{\"panel_id\":\"b\",\"text\":\"The scatter plot shows a gradual decrease in resistivity with increasing deposition temperature, reaching a minimum at 350 °C\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| ALD Cycles | ITO Thickness (Å) | Resistivity (Ohm·cm) |\\n|------------|-------------------|-----------------------|\\n| 0          | 0                 | 0.015                 |\\n| 100        | 100               | 0.002                 |\\n| 200        | 200               | 0.002                 |\\n| 300        | 300               | 0.002                 |\\n| 400        | 400               | 0.002                 |\\n| 500        | 500               | 0.002                 |\\n| 600        | 600               | 0.002                 |\\n| 700        | 700               | 0.002                 |\\n| 800        | 800               | 0.002                 |\\n| 900        | 900               | 0.002                 |\\n| 1000       | 1000              | 0.002                 |\"},{\"panel_id\":\"b\",\"text\":\"| Deposition Temperature (°C) | Resistivity (Ohm·cm) |\\n|-----------------------------|----------------------|\\n| 150                         | 0.01                 |\\n| 200                         | 0.005                |\\n| 250                         | 0.003                |\\n| 300                         | 0.002                |\\n| 350                         | 0.001                |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Yes\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"350 °C\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Resistivity decreases rapidly during early cycles and plateaus at ~0.002 Ω·cm after 500 Å film thickness.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"ALD cycle number, ITO thickness, Depostion temperature\"}]}]","bbox":[{"panel_id":"a","x":7,"y":0,"width":595,"height":623},{"panel_id":"b","x":657,"y":62,"width":579,"height":551}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_8.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/Jeffrey W. Elam et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"26","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":1236,"height":625,"image_format":"jpeg","image_sha256":"85f4410281ead97e0c1e123ec7c69812887a9efb75af3b7523963e9b31baa1ed","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_26_figure_9.jpg","caption":"Figure 9. Optical transmittance for ALD ITO films prepared on glass using 300 ALD cycles with  $5\\%$ $\\mathrm{SnO_2}$  cycles versus deposition temperature. Optical transmittance is given as the average transmission over the wavelength range  $370 - 1000\\mathrm{nm}$ .","id":"train/atomic-layer-deposition/experimental-usecase/26/figure_9","sample_id":"atomic-layer-deposition/experimental-usecase/26/figure_9","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows that optical transmission increases with a higher percentage of Sn cycles, rising steeply at first before plateauing around 92%\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| % of Sn Cycles | Optical Transmission (%) |\\n|----------------|--------------------------|\\n| 0              | 80                       |\\n| 1              | 82                       |\\n| 3              | 87                       |\\n| 5              | 88                       |\\n| 10             | 92                       |\\n| 15             | 92                       |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"As the percentage of Sn cycles increases, the optical transmission initially rises sharply. This rapid improvement occurs up to approximately 5–10% Sn cycles, indicating a significant enhancement in transparency with minimal Sn incorporation. Beyond this point, the curve begins to plateau, and further increases in Sn content do not yield substantial gains. The optical transmission stabilizes near 92%, suggesting a saturation point where additional Sn cycles have a negligible impact on optical performance. This trend implies that optimal transparency can be achieved with a relatively low Sn cycle percentage.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Optical transmission (%)\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":583,"height":584}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_9.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/images/figure_9.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/26/Jeffrey W. Elam et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"26","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":583,"height":584,"image_format":"jpeg","image_sha256":"502bf42639e14dc078bee8d3edc500ed6c5d708792184b4881d7201ff5e30145","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_28_figure_12.jpg","caption":"Figure 12. Experimental and simulated  $I - V$  curves of the OTS device. (a)  $I - V$  characteristics by plotting  $I_{\\mathrm{DUT}} - V_{\\mathrm{DUT}}$ . (b)  $I - V$  characteristics for the  $I_{\\mathrm{off}}$  region by voltage sweep. Both experiments were performed using devices fabricated via the ALD-DFM process.","id":"train/atomic-layer-deposition/experimental-usecase/28/figure_12","sample_id":"atomic-layer-deposition/experimental-usecase/28/figure_12","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows a scatter plot of current versus voltage, indicating a non-linear relationship where current increases exponentially with voltage.\"},{\"panel_id\":\"b\",\"text\":\"The figure displays a scatter plot comparing experimental and simulated currents against voltage, highlighting a exponential increase in current with voltage.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Voltage (V) | Current (A) |\\n|---|---|\\n|0.5| 10^-4 |\\n|0.7| 10^-3 |\\n|0.9| 10^-2 |\\n|1.1| 2*10^-2 |\\n|1.3| 2.5*10^-2 |\\n|1.5| 2.5*10^-2 |\\n|1.7| 10^-4 |\\n|1.9| 10^-4 |\"},{\"panel_id\":\"b\",\"text\":\"| Voltage (V) | Current (A) |\\n|---|---|\\n|0| 10^-8 |\\n|0.5| 4*10^-7 |\\n|1| 10^-6 |\\n|1.5| 4*10^-6 |\\n|2| 10^-5 |\\n|2.5| 10^-3 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"slightly below 1 V\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"There seem to be a lot of datapoints where just noise is measured resulting in a huge spread in datapoints.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Yes mostly for low voltages the different types of data overlap, however the switching current is less accurately simulated.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"10^-3 ampere\"}]}]","bbox":[{"panel_id":"a","x":7,"y":38,"width":344,"height":304},{"panel_id":"b","x":350,"y":41,"width":321,"height":302}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/images/figure_12.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/images/figure_12.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/Woohyun Kim et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"28","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":672,"height":342,"image_format":"jpeg","image_sha256":"02ff121b5a035607e98b8138c1264cd8119cacc48148216f3577f68643586281","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_28_figure_9.jpg","caption":"Figure 9. Resistivity-temperature experiments (top) and GAXRD patterns (bottom) of the as-deposited and annealed GeSe films grown through (a), (c) the conventional ALD process and (b), (d) the ALD-DFM process to verify the crystallization temperature.","id":"train/atomic-layer-deposition/experimental-usecase/28/figure_9","sample_id":"atomic-layer-deposition/experimental-usecase/28/figure_9","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"},{"panel_id":"c","label":"stacked spectra chart"},{"panel_id":"d","label":"stacked spectra chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The sheet resistance of ALD grown GeSe is plotted against temperature. There is a sharp decrease in resistance at around 337°C.\"},{\"panel_id\":\"b\",\"text\":\"The sheet resistance of DFM-ALD grown GeSe is plotted against temperature. There is a sharp decrease in resistance at around 348°C.\"},{\"panel_id\":\"c\",\"text\":\"X-ray diffraction patterns are shown for different temperatures, indicating changes in crystallinity for 340 and 350 degrees for ALD grown GeSe.\"},{\"panel_id\":\"d\",\"text\":\"X-ray diffraction patterns are shown for different temperatures, indicating changes in crystallinity for 350 degrees for ALD grown GeSe.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Temperature (°C) | Sheet resistance (Ω/□) |\\n|---|---|\\n| 0 | 3*10^7 |\\n| 50 | 3*10^7 |\\n| 100 | 3*10^7 |\\n| 150 | 3*10^7 |\\n| 200 | 3*10^7 |\\n| 250 | 3*10^7 |\\n| 300 | 3*10^7 |\\n| 350 | 10^4 |\"},{\"panel_id\":\"b\",\"text\":\"| Temperature (°C) | Sheet resistance (Ω/□) |\\n|---|---|\\n| 0 | 3*10^7 |\\n| 50 | 3*10^7 |\\n| 100 | 3*10^7 |\\n| 150 | 3*10^7 |\\n| 200 | 3*10^7 |\\n| 250 | 3*10^7 |\\n| 300 | 3*10^7 |\\n| 350 | 10^4 |\"},{\"panel_id\":\"c\",\"text\":\"| 2θ (°) | Intensity (A.U.) |\\n|---|---|\\n| 20 | - |\\n| 25 | - |\\n| 30 | - |\\n| 32 | Peak for 340 and 350 degrees|\\n| 40 | - |\\n| 45 | - |\\n| 50 | - |\"},{\"panel_id\":\"d\",\"text\":\"| 2θ (°) | Intensity (A.U.) |\\n|---|---|\\n| 20 | - |\\n| 25 | - |\\n| 30 | - |\\n| 32 | Peak for 350 degrees|\\n| 40 | - |\\n| 45 | - |\\n| 50 | - |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The material is more conductive in crystalline state, at higher temperatures the material is more crystalline. Also for these high temperatures lower sheet resistances are measured.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"There are two peaks\"}]},{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"By putting a sample in a RTA or rapid thermal anneal tool.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"348 degrees Celcius\"}]}]","bbox":[{"panel_id":"a","x":29,"y":53,"width":312,"height":307},{"panel_id":"b","x":362,"y":53,"width":310,"height":307},{"panel_id":"c","x":53,"y":393,"width":289,"height":300},{"panel_id":"d","x":383,"y":403,"width":288,"height":290}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/images/figure_9.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/images/figure_9.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/28/Woohyun Kim et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"28","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":672,"height":694,"image_format":"jpeg","image_sha256":"9059f64bc498b64dc9b2bf7cfde370eb7acfd6d03e50bcd1443c240976e54068","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_29_figure_6.jpg","caption":"Figure 6. Ellipsometric measurements of the tungsten nitride film thickness deposited by three AB cycles vs.  $\\mathrm{NH}_3$  exposure at  $600~\\mathrm{K}$ . A  $\\mathrm{WF}_6$  exposure of  $4300~\\mathrm{L}$  was sufficient for a complete  $\\mathrm{WF}_6$  half- reaction at  $600~\\mathrm{K}$ .","id":"train/atomic-layer-deposition/experimental-usecase/29/figure_6","sample_id":"atomic-layer-deposition/experimental-usecase/29/figure_6","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The graph shows the relationship between NH₃ exposure and film thickness at 600 K with 3 AB cycles. The film thickness increases rapidly initially and then plateaus.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| NH₃ Exposure (L) | Film Thickness (Å) |\\n|---|---|\\n| 0 | 0 |\\n| 1000 | 1 |\\n| 2000 | 2 |\\n| 3000 | 3 |\\n| 4000 | 4 |\\n| 5000 | 6 |\\n| 6000 | 7 |\\n| 7000 | 8 |\\n| 8000 | 8 |\\n| 9000 | 8 |\\n| 10000 | 8 |\\n| 11000 | 8 |\\n| 12000 | 8 |\\n| 13000 | 8 |\\n| 14000 | 8 |\\n| 15000 | 8 |\\n| 16000 | 8 |\\n| 17000 | 8 |\\n| 18000 | 8 |\\n| 19000 | 8 |\\n| 20000 | 8 |\\n| 21000 | 8 |\\n| 22000 | 8 |\\n| 23000 | 8 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"3.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"After about 10,000 L.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"7 Å.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":0,"y":1,"width":663,"height":643}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/29/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/29/images/figure_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/29/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/29/J. W. Klaus et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"29","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":666,"height":647,"image_format":"jpeg","image_sha256":"fc4c5da4324f2b2514ab11de50dded83c92060fa1377bf652550e43fc3d06299","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_29_figure_7.jpg","caption":"Figure 7. Ellipsometric measurements of the tungsten nitride film thickness deposited by three AB cycles vs.  $\\mathrm{WF}_6$  exposure at  $600~\\mathrm{K}$ . A  $\\mathrm{NH}_3$  exposure of  $12,000~\\mathrm{L}$  was sufficient for a complete  $\\mathrm{NH}_3$  half- reaction at  $600~\\mathrm{K}$ .","id":"train/atomic-layer-deposition/experimental-usecase/29/figure_7","sample_id":"atomic-layer-deposition/experimental-usecase/29/figure_7","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart shows the relationship between the exposure of WF₆ and film thickness at 3 AB cycles and a temperature of 600 K.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| WF₆ Exposure (L) | Film Thickness (Å) |\\n|---|---|\\n| 0 | 0 |\\n| 1000 | 1 |\\n| 2000 | 2 |\\n| 3000 | 3 |\\n| 6000 | 8 |\\n| 12000 | 8 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"12,000 L.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 4 Å.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"After about 3,000 L.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"About 8 Å.\"}]}]","bbox":[{"panel_id":"a","x":2,"y":0,"width":652,"height":646}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/29/images/figure_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/29/images/figure_7.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/29/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/29/J. W. Klaus et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"29","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":658,"height":650,"image_format":"jpeg","image_sha256":"effb04facf8a3c982346fb17ee8cfd7a7a514ab1b5d6dd444b3cf0c489c16fcd","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_32_figure_1.jpg","caption":"Figure 1. Growth rates of thermal ALD  $\\mathrm{Ta_2O_5}$  as a function of PDMAT exposure time  $(t_{\\mathrm{s}})$  on  $\\mathrm{Si(001)}$  substrate at  $T_{\\mathrm{s}} = 250^{\\circ}\\mathrm{C}$ .","id":"train/atomic-layer-deposition/experimental-usecase/32/figure_1","sample_id":"atomic-layer-deposition/experimental-usecase/32/figure_1","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows growth rate of thermal ALD Ta₂O₅ as a function of PDMAT exposure time at 250°C. The growth rate increases rapidly at short exposure times and saturates at approximately 0.85 Å/cycle for t_s ≥ 1 s, demonstrating self-limiting adsorption characteristic of ideal ALD behavior.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| t_s (s) | Growth rate (Å/cycle) |\\n|---|---|\\n| 0 | 0 |\\n| 0.5 | 0.7 |\\n| 1 | 0.8 |\\n| 1.5 | 0.8 |\\n| 2 | 0.8 |\\n| 2.5 | 0.8 |\\n| 3 | 0.8 |\\n| 3.5 | 0.8 |\\n| 4 | 0.8 |\\n| 4.5 | 0.8 |\\n| 5 | 0.8 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The growth rate increases steeply at short PDMAT exposure times, rising from near zero to about 0.7 Å/cycle at 0.5 s. Beyond 1 s of exposure, the growth rate plateaus at approximately 0.85 Å/cycle. This saturation behavior demonstrates self-limiting adsorption of the PDMAT precursor, which is the defining characteristic of an ideal ALD process.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Approximately 1 second.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Yes. The 0.85 Å/cycle growth rate is almost twice that reported for ethoxide or chloride precursors (both ~0.45 Å/cycle).\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Purge time (t_p = 3 s), water reactant exposure time (t_r = 3 s), substrate temperature (T_s = 250°C).\"}]}]","bbox":[{"panel_id":"a","x":3,"y":7,"width":677,"height":521}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/32/images/figure_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/32/images/figure_1.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/32/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/32/W. J. Maeng et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"32","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":683,"height":528,"image_format":"jpeg","image_sha256":"5d4764f0f32aa9a45efed33704aa21a226f61a5846cccb39c936e51ba3491bae","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_32_figure_4.jpg","caption":"Figure 4. Growth rates of  $\\mathrm{TiO_2}$  as a function of TDMAT exposure time on Si(001) substrate at  $T_{\\mathrm{s}} = 200^{\\circ}\\mathrm{C}$ .","id":"train/atomic-layer-deposition/experimental-usecase/32/figure_4","sample_id":"atomic-layer-deposition/experimental-usecase/32/figure_4","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows growth rate of thermal ALD TiO₂ as a function of TDMAT exposure time at 200°C. Unlike Ta₂O₅ from PDMAT, the growth rate does not saturate but continues to increase with exposure time, indicating incomplete self-limiting behavior. This non-ideal ALD characteristic is attributed to the low thermal decomposition temperature of TDMAT (~180°C).\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| t_S (s) | Growth rate (Å/cycle) |\\n|---|---|\\n| 0 | 0 |\\n| 1 | 1.5 |\\n| 2 | 2 |\\n| 3 | 2.5 |\\n| 4 | 3 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"No. Unlike Ta₂O₅ ALD from PDMAT, which shows clear saturation above 1 s exposure time, the TiO₂ growth rate from TDMAT continues to increase throughout the tested range up to 7 s. Although some slowdown occurs above 1 s, true saturation is never achieved. This indicates that TDMAT does not adsorb with perfect self-limitation at this temperature.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The low thermal decomposition temperature of TDMAT, reported to be approximately 180°C, which is below the growth temperature of 200°C used in this experiment.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"At t_s = 2 s, the growth rate is approximately 2.1 Å/cycle. This is significantly higher than values reported for other precursors such as titanium methoxide (0.5 Å/cycle) and titanium chloride (0.75 Å/cycle). However, the higher rate partly reflects the non-ideal CVD contribution rather than pure ALD growth.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The TDMAT precursor exposure time in seconds.\"}]}]","bbox":[{"panel_id":"a","x":4,"y":2,"width":670,"height":944}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/32/images/figure_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/32/images/figure_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/32/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/32/W. J. Maeng et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"32","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":661,"height":497,"image_format":"jpeg","image_sha256":"b820e3792ed88eb26c1d05640a79b0c38aa6249151fc408441a5a7ceafbd462c","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_33_figure_5.jpg","caption":"Figure 5 Dependence of the saturated ALD cobalt (a) and copper (b) thicknesses per cycle, as a function of the substrate temperature.","id":"train/atomic-layer-deposition/experimental-usecase/33/figure_5","sample_id":"atomic-layer-deposition/experimental-usecase/33/figure_5","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure compares the dependence of ALD growth rate on substrate temperature for two different processes or materials. In subfigure (a), the growth rate increases steadily from approximately 0.01 Å/cycle at 260 °C to 0.12 Å/cycle at 340 °C, suggesting a thermally activated process with relatively linear behavior in this temperature range.\"},{\"panel_id\":\"b\",\"text\":\"In subfigure (b), the growth rate follows a sigmoidal trend, increasing more sharply between 200 °C and 280 °C, indicating a possible transition from a kinetically limited regime to a more saturated or efficient growth regime. Both plots highlight the strong influence of temperature on film growth dynamics, with potential implications for process window optimisation.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Substrate temperature (°C) | Growth rate (Å per cycle) |\\n|----------------------------|----------------------------|\\n| 260                        | 0.01                       |\\n| 280                        | 0.03                       |\\n| 300                        | 0.06                       |\\n| 320                        | 0.09                       |\\n| 340                        | 0.12                       |\"},{\"panel_id\":\"b\",\"text\":\"| Substrate temperature (°C) | Growth rate (Å per cycle) |\\n|----------------------------|----------------------------|\\n| 160                        | 0.0                        |\\n| 200                        | 0.1                        |\\n| 220                        | 0.2                        |\\n| 240                        | 0.35                       |\\n| 280                        | 0.5                        |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The plots suggest that different ALD processes or chemistries have distinct thermal activation behaviors. In (a), the steady increase in growth rate with temperature implies a broad and controllable window, which is advantageous for uniform deposition. In contrast, the sigmoidal trend in (b) indicates that below a certain threshold (~200 °C), the process is kinetically limited, but rapidly becomes efficient above this temperature. This means that process engineers may need to operate within a narrower high-temperature window for maximum efficiency, potentially requiring more precise thermal control. Understanding these trends helps optimize reactor conditions for consistent film quality and throughput\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"320 °C.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Higher temperatures enhance precursor reactivity, improve byproduct desorption\"}]}]","bbox":[{"panel_id":"a","x":12,"y":8,"width":681,"height":533},{"panel_id":"b","x":15,"y":569,"width":666,"height":491}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/33/images/figure_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/33/images/figure_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/33/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/33/BOOYONG S. LIM et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"33","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":694,"height":1070,"image_format":"jpeg","image_sha256":"790c35df622f8449764bda50eb4f2b4ac5eb78446840dbcbf024c508850997e6","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_35_figure_2.jpg","caption":"Figure 2. Growth rate as a function of (a)  $\\mathrm{SbCl}_3$  and (b)  $(\\mathrm{Et}_3\\mathrm{Si})_2\\mathrm{Te}$  pulse length for  $\\mathrm{Sb}_2\\mathrm{Te}_3$  films deposited at  $60^{\\circ}\\mathrm{C}$ . The other precursor pulse lengths were kept at  $1.0~\\mathrm{s}$  and all the purge lengths at  $2.0~\\mathrm{s}$ .","id":"train/atomic-layer-deposition/experimental-usecase/35/figure_2","sample_id":"atomic-layer-deposition/experimental-usecase/35/figure_2","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The growth rate of a material increases with increasing SbCl₃ pulse length up to a certain point and then plateaus.\"},{\"panel_id\":\"b\",\"text\":\"The growth rate of a material increases gradually with increasing (Et₃Si)₂Te pulse length, but shows subsaturation.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| SbCl₃ pulse length (s) | Growth rate (Å/cycle) |\\n|---|---|\\n| 0.0 | 0.0 |\\n|0.2|0.3|\\n| 0.5 | 0.6 |\\n| 1.0 | 0.6 |\\n| 2.0 | 0.6 |\"},{\"panel_id\":\"b\",\"text\":\"| (Et₃Si)₂Te pulse length (s) | Growth rate (Å/cycle) |\\n|---|---|\\n| 0.0 | 0.0 |\\n|0.2|0.0|\\n| 0.5 | 0.45 |\\n| 1.0 | 0.6 |\\n| 1.5 | 0.7 |\\n| 2.0 | 0.6 |\\n| 3.0 | 0.65 |\\n| 4.0 | 0.7 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.6 angstrom per cycle.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"No it does not, there seems to be a form of softsaturation, where there is still an increase in growth rate for ever increasing pulse length.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.6 angstrom per cycle\"}]}]","bbox":[{"panel_id":"a","x":0,"y":9,"width":304,"height":230},{"panel_id":"b","x":336,"y":9,"width":303,"height":230}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/35/images/figure_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/35/images/figure_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/35/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/35/Viljami Pore et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"35","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":639,"height":239,"image_format":"jpeg","image_sha256":"30626e05b6895f64741769959d5203206e736b46bae891424907821887ddfb29","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_42_figure_3.jpg","caption":"Figure 3. ALD saturation curve for 1 as the pulse length of 1 was varied.","id":"train/atomic-layer-deposition/experimental-usecase/42/figure_3","sample_id":"atomic-layer-deposition/experimental-usecase/42/figure_3","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot image represents the growth rate (in A/cycle) against pulse length (in seconds). There seems to be no growth till 2 secs and it  seems to increase only after 3 secs and reaches a peak at 5 secs and slightly decreases.  This saturation is the identity of a self-limiting ALD process, where increasing the exposure time beyond a certain point no longer increases thickness, confirming complete surface reaction.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| pulse length (s) | growth rate (A/cycle) |\\n|---|---|\\n| 1 | 0 |\\n| 2 | 0 |\\n| 3 | 0.15 |\\n| 4 | 0.2 |\\n| 5 | 0.22 |\\n| 6 | 0.21 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Step 1: Pulse of Compound 1 (vapor) onto the hot substrate \\nStep 2: 3 s purge with N₂ \\nStep 3: 6 s pulse of H₂ plasma \\nStep 4: Second N₂ purge. Repeat.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"It increased until it reached a plateau at 0.2 Å/cycle after a 4 s pulse. Shorter pulses (1–2 s) gave no measurable thickness.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The surface became fully covered with precursor after 4 s; longer pulses couldn’t add more copper because all reaction sites were occupied a sign of self-limiting ALD behavior.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Mostly yes—EDS showed it was mainly copper metal with <5% carbon impurity. The growth rate (0.2 Å/cycle) is within the typical range for copper ALD, making it potentially useful for chip interconnects or barrier layers.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":3,"width":560,"height":284}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/42/images/figure_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/42/images/figure_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/42/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/42/Coyle et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"42","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":564,"height":287,"image_format":"jpeg","image_sha256":"df7f976113482658f4df2439bc1b11b8f8f6f3741ef40d2b7fed7f8946901bcf","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_45_figure_4.jpg","caption":"Figure 4. (Color online) The growth per cycle for remote plasma and thermal ALD of Pt and remote plasma ALD of  $\\mathrm{PtO_2}$  as a function of the substrate temperature. At  $100^{\\circ}\\mathrm{C}$  the growth rate of the remote plasma process including a  $\\mathrm{H}_{2}$  gas exposure step in the ALD cycle is shown. For thermal ALD of  $\\mathrm{Pt}$  deposition took place on  $10 \\mathrm{nm}$  thick  $\\mathrm{Pt}$  starting surfaces prepared by remote plasma ALD at  $300^{\\circ}\\mathrm{C}$  The lines serve as guides to the eye.","id":"train/atomic-layer-deposition/experimental-usecase/45/figure_4","sample_id":"atomic-layer-deposition/experimental-usecase/45/figure_4","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows the growth per cycle of different ALD  processes (Thermal ALD Pt , Remote plasma ALD Pt , Remote plasma ALD PtO2)  at varying deposition temperatures. Remote plasma Pt ALD yields a relatively weak temperature dependence, whereas thermal ALD Pt shows little growth at lower temperatures and increases sharply only above ~200 °C. Growth per cycle decreases with temperature for remote plasma PtO2 ALD.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Deposition temperature (°C) | Process | Growth per cycle (nm/cycle) |\\n|---|---|---|\\n| 150 |  Thermal ALD Pt  |  0.001 |\\n| 200 |  Thermal ALD Pt  | 0.01 |\\n| 225 |  Thermal ALD Pt  | 0.035 |\\n| 250 |  Thermal ALD Pt  | 0.037 |\\n| 300 |  Thermal ALD Pt  | 0.044 |\\n| 150 |  Thermal ALD Pt  |  0.001 |\\n| 100 |  Remote plasma ALD Pt  with H2 step | 0.038 |\\n| 200 |  Remote plasma ALD Pt  | 0.044 |\\n| 300 |  Remote plasma ALD Pt  | 0.046 |\\n| 100 |  Remote plasma ALD PtO2 | 0.06 |\\n| 200 |  Remote plasma ALD PtO2  | 0.057 |\\n| 300 |  Remote plasma ALD PtO2  | 0.047 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Thermal ALD Pt.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Temperatures around ~225–300 °C are closest to an ideal ALD window for thermal Pt ALD. In this range, the measured growth per cycle changes only slightly (within the error bars), indicating relatively stable, self-limited behavior. At lower temperatures, the growth rate is  changing rapidly.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Remote plasma ALD Pt is the least temperature-sensitive between 200 and 300 °C.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"At 200 °C, the  values rank as:\\n1. Remote plasma ALD PtO₂ (~0.058–0.060 nm/cycle)\\n2. Remote plasma ALD Pt (~0.045 nm/cycle)\\n3. Thermal ALD Pt (~0.01–0.02 nm/cycle)\"}]}]","bbox":[{"panel_id":"a","x":2,"y":1,"width":597,"height":458}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/45/images/figure_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/45/images/figure_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/45/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/45/Knoops et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"45","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":605,"height":464,"image_format":"jpeg","image_sha256":"3e6912ef0fd19dbc194f0fac994cf8183223dedb2a03bba9cef87d7d44e7e5b6","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_49_figure_1.jpg","caption":"Figure 1. a) The film thickness of the  $\\mathrm{MoS_2}$  film deposited at different temperatures ranging from 175 to  $300^{\\circ}C$  The thickness of 50 ALD cycles  $\\mathrm{MoS_2}$  film as a function of the b)  $\\mathsf{H}_2\\mathsf{S}$  exposure time, c) dose time, and d) exposure time of  $\\mathsf{C}_7\\mathsf{H}_8\\mathsf{Mo}(\\mathsf{CO})_3$  e) Spectroscopic ellipsometry curves for  $\\mathrm{MoS_2}$  films deposited with varying ALD cycles (25, 50, 100, 200, and 300). f) There is a linear dependence between the number of ALD cycles and the deposited film thickness. The ALD reaction temperature was  $250^{\\circ}C$  for all parameter variations except (a).","id":"train/atomic-layer-deposition/experimental-usecase/49/figure_1","sample_id":"atomic-layer-deposition/experimental-usecase/49/figure_1","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"c","label":"scatter plot"},{"panel_id":"e","label":"multiple line chart"},{"panel_id":"b","label":"scatter plot"},{"panel_id":"d","label":"scatter plot"},{"panel_id":"f","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This multi-panel figure characterizes the Atomic Layer Deposition (ALD) of MoS₂ films, showing how thickness depends on various parameters such as  (a) temperature, (b) H₂S exposure time, (c) precursor dose time, (d) Mo-precursor exposure time, (e) optical properties versus cycles, and (f) linear thickness growth with cycle number.  The scatter plot shows the relationship between reaction temperature and layer thickness after 100 ALD cycles. The film thickness after 100 cycles peaks at approximately 250 °C (~9 nm), defining the optimal ALD window. Below 225 °C, growth is slow due to kinetic limitations, while above 275 °C, the thickness drops sharply, likely due to desorption or precursor decomposition.\"},{\"panel_id\":\"b\",\"text\":\"The scatter plot shows the relationship between H₂S exposure time and layer thickness after 50 ALD cycles. The thickness saturates at approximately 30 s exposure time.\"},{\"panel_id\":\"c\",\"text\":\"The scatter plot illustrates the effect of C₇H₈Mo(CO)₃ dose time on layer thickness after 50 ALD cycles. The growth saturates at approximately 0.8 s dose time.\"},{\"panel_id\":\"d\",\"text\":\"The scatter plot illustrates the effect of C₇H₈Mo(CO)₃ exposure time on layer thickness after 50 ALD cycles. The thickness shows a peak at 20 s, followed by a decrease, suggesting possible etching or desorption at very long exposure times.\"},{\"panel_id\":\"e\",\"text\":\"The line chart depicts the change in ellipsometric delta (Δ) with respect to wavelength. The delta parameter shifts monotonically with increasing ALD cycles.\"},{\"panel_id\":\"f\",\"text\":\"The scatter plot shows the linear relationship between ALD cycle number and layer thickness. The graph confirms a highly linear growth per cycle (GPC) of 0.11 nm/cycle with R² = 0.994, demonstrating that the process is stable and self-limiting.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Reaction Temperature (°C) | Layer Thickness After 100 ALD Cycles (nm) |\\n|---|---|\\n| 175      | 0.7       |\\n| 200      | 3         |\\n| 225      | 8         |\\n| 250      | 9         |\\n| 275      | 7.8       |\\n| 285      | 9.1       |\\n| 300      | 4         |\"},{\"panel_id\":\"b\",\"text\":\"| H2S Exposure Time (s) | Layer Thickness After 50 ALD Cycles (nm) |\\n|---|---|\\n| 0.1 | 0.2 |\\n| 1   | 2   |\\n| 5   | 3.1 |\\n| 10  | 3.3 |\\n| 30  | 4.7 |\\n| 40  | 4   |\\n| 60  | 4.7 |\"},{\"panel_id\":\"c\",\"text\":\"| C7H8Mo(CO)3 Dose Time (s) | Layer Thickness After 50 ALD Cycles (nm) |\\n|---|---|\\n| 0        | 0         |\\n| 0.25     | 3         |\\n| 0.5      | 4.2       |\\n| 0.75     | 4.8       |\\n| 1        | 4.7       |\\n| 1.5      | 5.1       |\"},{\"panel_id\":\"d\",\"text\":\"| C7H8Mo(CO)3 Exposure Time (s) | Layer Thickness After 50 ALD Cycles (nm) |\\n|---|---|\\n| 5   | 2.7 |\\n| 10  | 4.9 |\\n| 20  | 6.5 |\\n| 30  | 4.7 |\"},{\"panel_id\":\"e\",\"text\":\"| Wavelength (nm) | Delta (degree) | Number of cycles |\\n|---|---|---|\\n| 370            | 164       | 0 cycle |\\n| 370            | 158       | 25 cycle |\\n| 370            | 145       | 50 cycle |\\n| 370            | 135       | 100 cycle |\\n| 370            | 105       | 200 cycle |\\n| 370            | 72        | 300 cycle |\\n| 400            | 167       | 0 cycle |\\n| 400            | 163       | 25 cycle |\\n| 400            | 150       | 50 cycle |\\n| 400            | 140       | 100 cycle |\\n| 400            | 110       | 200 cycle |\\n| 400            | 83        | 300 cycle |\\n| 500            | 170       | 0 cycle |\\n| 500            | 166       | 25 cycle |\\n| 500            | 154       | 50 cycle |\\n| 500            | 143       | 100 cycle |\\n| 500            | 115       | 200 cycle |\\n| 500            | 93        | 300 cycle |\\n| 600            | 171       | 0 cycle |\\n| 600            | 169       | 25 cycle |\\n| 600            | 157       | 50 cycle |\\n| 600            | 147       | 100 cycle |\\n| 600            | 120       | 200 cycle |\\n| 600            | 98        | 300 cycle |\\n| 700            | 173       | 0 cycle |\\n| 700            | 171       | 25 cycle |\\n| 700            | 160       | 50 cycle |\\n| 700            | 149       | 100 cycle |\\n| 700            | 124       | 200 cycle |\\n| 700            | 104       | 300 cycle |\\n| 800            | --       | 0 cycle |\\n| 800            | --       | 25 cycle |\\n| 800            | --       | 50 cycle |\\n| 800            | --       | 100 cycle |\\n| 800            | --       | 200 cycle |\\n| 800            | 109      | 300 cycle |\\n| 900            | --       | 0 cycle |\\n| 900            | --       | 25 cycle |\\n| 900            | --       | 50 cycle |\\n| 900            | --       | 100 cycle |\\n| 900            | --       | 200 cycle |\\n| 900            | 114      | 300 cycle |\"},{\"panel_id\":\"f\",\"text\":\"| ALD Cycle Number | Layer Thickness (nm) |\\n|---|---|\\n| 0   | 0  |\\n| 25  | 1  |\\n| 50  | 4  |\\n| 101 | 8  |\\n| 200 | 19 |\\n| 301 | 33 |\"}]","vqa":"[{\"panel_id\":\"e\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It confirms precise, atomic-level thickness control and excellent film uniformity. These characteristics are essential for reliably fabricating thin MoS₂ channels or barriers in scalable transistors and memory devices.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The optimal temperature is 250°C (highest thickness, stable growth). The film thickness drops significantly to <4 nm (compared to ~9 nm at 250°C). This indicates that 300°C is outside the stable ALD window, likely causing precursor desorption or parasitic etching, resulting in a much thinner film than intended.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Increasing dose saturates the surface coverage (good). However, extending the residence time (exposure) beyond 20s causes a loss in thickness (bad).\\nThis suggests a competing desorption or etching mechanism at long exposures. The process engineer cannot simply set an arbitrarily long exposure time \\\"just to be safe.\\\" They must choose a value (e.g., 20s) that ensures saturation without crossing into the regime where the film begins to degrade or desorb.\"}]},{\"panel_id\":\"f\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It implies a Nucleation Delay. In an ideal scenario, the line would pass through zero. A negative intercept signifies that during the first ~10 cycles, virtually no film was deposited (incubation period). The precursor struggled to nucleate on the initial substrate surface, requiring several cycles to establish a seed layer before the steady-state growth rate of 0.11 nm/cycle was achieved.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":4,"width":456,"height":350},{"panel_id":"b","x":5,"y":358,"width":450,"height":345},{"panel_id":"c","x":471,"y":2,"width":449,"height":354},{"panel_id":"d","x":471,"y":362,"width":450,"height":343},{"panel_id":"e","x":922,"y":4,"width":447,"height":354},{"panel_id":"f","x":931,"y":357,"width":438,"height":347}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/49/images/figure_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/49/images/figure_1.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/49/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/49/Shen et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"49","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":1372,"height":709,"image_format":"jpeg","image_sha256":"534498c7646df77586a9d45fcf8f97347ad43e1825c35ebf7b4bc7b39c31182d","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_50_fig_11.jpg","caption":"Fig. 11 Optical band gap  $(E_{\\mathrm{g,opt}})$  as a function of DF for ZnO:X  $(X = \\mathsf{Al},\\mathsf{B})$  TIB was used as a B source while DMAI and TMA were used as Al sources. The films were deposited on  $\\mathrm{SiO_2}$  substrates at  $150^{\\circ}C$","id":"train/atomic-layer-deposition/experimental-usecase/50/fig_11","sample_id":"atomic-layer-deposition/experimental-usecase/50/fig_11","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows the linear relationship between doping fraction (DF) and the optical band gap (Eg_opt) for three different materials: TIB, DMAI, and TMA.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Doping fraction, DF | TIB | DMAI | TMA |\\n|---|---|---|---|\\n| 0.00 | 3.25 | 3.25 | 3.25 |\\n| 0.02 | 3.30 | 3.30 | 3.30 |\\n| 0.04 | 3.35 | 3.35 | 3.35 |\\n| 0.06 | 3.40 | 3.40 | 3.40 |\\n| 0.08 | 3.45 | 3.45 | 3.45 |\\n| 0.10 | 3.50 | 3.50 | 3.50 |\\n| 0.12 | 3.55 | 3.55 | 3.55 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The increase in bandgap can be attributed to the Burstein–Moss shift, but only for the range of doping values for which the carrier concentration n increases\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The optical band gap (Eg,opt) of the ZnO:X films was extracted from so-called Tauc plots (i.e., (e2E^2)^2 as a function\\nof the photon energy E, where e2 is the imaginary part of the dielectric function extracted from the SE data\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"For ZnO:B, as well as for ZnO:Al, the optical band gap increases with increasing doping fraction. This increase can be attributed to the Burstein–Moss shift, for the range of DF values for which the carrier concentration n increases.  For higher DF values (DF > 0.034), in the case of ZnO:B and ZnO:Al (DMAI), the increase in the optical bandgap can be attributed to effects related to the grain size, strain and other types of imperfections.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":7,"width":365,"height":289}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/images/fig_11.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/images/fig_11.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/Garcia-Alonso et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"50","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":369,"height":297,"image_format":"jpeg","image_sha256":"a8a021cd8363e79dce5ac504bc29f2859c22b4a4c977fc9c76fb53fac3d99e59","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_50_fig_2.jpg","caption":"Fig. 2 Saturation curves of the  $\\mathrm{ZnO:}$  B process  $(m = 4)$  at  $150^{\\circ}C$  (a) TIB precursor dose, (b) TIB precursor purge, (c)  $\\mathsf{DI - H}_2\\mathsf{O}$  dose, and (d)  $\\mathsf{DI - H}_2\\mathsf{O}$  purge. The growth per supercycle (GPSC) for each condition was averaged over 10 supercycles. The  $\\mathrm{ZnO}$  standard conditions are specified elsewhere (50 ms DEZ dose, 5 s purge, 20 ms  $\\mathsf{H}_2\\mathsf{O}$  dose, 6 s purge) and the process saturated at a GPC of  $0.2\\mathrm{nm}$  per cycle for an intrinsic  $\\mathrm{ZnO}$  film deposited at  $150^{\\circ}C$","id":"train/atomic-layer-deposition/experimental-usecase/50/fig_2","sample_id":"atomic-layer-deposition/experimental-usecase/50/fig_2","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"},{"panel_id":"c","label":"scatter plot"},{"panel_id":"d","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The image plots GPSC as a function of TIB dose\"},{\"panel_id\":\"b\",\"text\":\"The image plots GPSC as a function of TIB purge time\"},{\"panel_id\":\"c\",\"text\":\"The image plots GPSC as a function of DI-H₂O dose\"},{\"panel_id\":\"d\",\"text\":\"The image plots GPSC as a function of  DI-H₂O purge time\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| TIB dose (ms) | GPSC (nm/supercycle) |\\n|---|---|\\n| 0 | 0.7 |\\n| 10 | 0.6 |\\n| 20 | 0.6 |\\n| 30 | 0.6 |\\n| 40 | 0.6 |\\n| 50 | 0.6 |\\n| 60 | 0.6 |\\n| 70 | 0.6 |\\n| 80 | 0.6 |\"},{\"panel_id\":\"b\",\"text\":\"| TIB purge (s) | GPSC (nm/supercycle) |\\n|---|---|\\n| 0 | 0.6 |\\n| 1 | 0.6 |\\n| 2 | 0.6 |\\n| 3 | 0.6 |\\n| 4 | 0.6 |\\n| 5 | 0.6 |\\n| 6 | 0.6 |\\n| 7 | 0.6 |\\n| 8 | 0.6 |\"},{\"panel_id\":\"c\",\"text\":\"| DI-H₂O dose (ms) | GPSC (nm/supercycle) |\\n|---|---|\\n| 0 | 0.5 |\\n| 10 | 0.6 |\\n| 20 | 0.6 |\\n| 30 | 0.6 |\\n| 40 | 0.6 |\\n| 50 | 0.6 |\\n| 60 | 0.6 |\\n| 70 | 0.6 |\\n| 80 | 0.6 |\\n| 90 | 0.6 |\\n| 100 | 0.6 |\"},{\"panel_id\":\"d\",\"text\":\"| DI-H₂O purge (s) | GPSC (nm/supercycle) |\\n|---|---|\\n| 0 | 0.6 |\\n| 1 | 0.6 |\\n| 2 | 0.6 |\\n| 3 | 0.6 |\\n| 4 | 0.6 |\\n| 5 | 0.6 |\\n| 6 | 0.6 |\\n| 7 | 0.6 |\\n| 8 | 0.6 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The reduction in GSPC with increasing TIB dose observed in a could be the result of a slight etching of the ZnO by the TIB  but it could also be due to the inhibition of ZnO growth after a TIB pulse, as observed for DMAI. However, the  experiment carried out in which the film growth after every cycle (either DEZ or TIB cycle), monitored by spectroscopic ellipsometry, showed the change in thickness as a function of the number of cycles deduced from the measurements. From that it is clear that the ZnO was not etched by TIB (i.e., the film thickness did not decrease after one B cycle) but that the TIB\\npulse led to an inhibition in the subsequent ZnO film growth. This growth inhibition took place during several cycles following the TIB pulse, and could also be seen from a second experiment\\nin which a supercycle with m = 19 was used (i.e., a total of 20 cycles comprising 19 ZnO + 1 TIB), it was evident that one TIB pulse affected the ZnO during many cycles.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Growth per super cycle as this ALD process involves doping ZnO with boron. The saturation curves for the gas exposures and purges of the B cycle of an ALD supercycle corresponding to four DEZ\\ncycles and one TIB cycle for the ZnO:B process. These data were obtained by varying the duration of one of the ALD cycle steps, while keeping the duration of other steps sufficiently long to ensure  either saturation of surface reactions or sufficient purging.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"1. Temperature of deposition = 150 °C\\n2. Standard recipe = 50 ms DEZ dose, 5 s purge, 20 ms H2O dose, 6 s purge\\n3. Saturated GPC = 0.2 nm/cycle\"}]}]","bbox":[{"panel_id":"a","x":4,"y":-1,"width":458,"height":317},{"panel_id":"b","x":489,"y":14,"width":391,"height":306},{"panel_id":"c","x":7,"y":330,"width":455,"height":308},{"panel_id":"d","x":494,"y":336,"width":385,"height":304}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/images/fig_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/images/fig_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/Garcia-Alonso et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"50","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":880,"height":644,"image_format":"jpeg","image_sha256":"6f3a643fefa5b48548c14541558e555175a04cb9a4ba495de7d63407ec180772","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_50_fig_3.jpg","caption":"Fig 3 Doping fraction (DF) as a function of the doping cycle ratio  $R_{x}$  for depositions at  $150^{\\circ}C$ , as measured by X-ray photoelectron spectroscopy (XPS). The doping cycle ratio  $R_{x}$  (i.e. the ratio of the number of doping ALD cycles  $(X = \\mathsf{Al}, \\mathsf{B})$  over the total ALD cycles in one supercycle) is calculated as  $1 / (m + 1)$ . The estimated relative error for the doping fraction is  $\\sim 10\\%$ . The lines serve as guides to the eye.","id":"train/atomic-layer-deposition/experimental-usecase/50/fig_3","sample_id":"atomic-layer-deposition/experimental-usecase/50/fig_3","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The plot shows the relationship between dopant cycle ratio and doping fraction (DF) for TIB, DMAI, and TMA.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Dopant Cycle Ratio (R) | TIB (DF) | DMAI (DF) | TMA (DF) |\\n|-------------------------|----------|-----------|----------|\\n| 0.02                   | 0.015    | 0.020     | 0.025    |\\n| 0.04                   | 0.025    | 0.045     | 0.060    |\\n| 0.06                   | 0.035    | 0.065     | 0.095    |\\n| 0.08                   | 0.045    | 0.085     | 0.135    |\\n| 0.10                   | 0.055    | 0.105     | 0.175    |\\n| 0.15                   | 0.075    | 0.115     | 0.215    |\\n| 0.20                   | 0.095    | —         | —        |\\n| 0.25                   | 0.115    | —         | —        |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"1. Trimethylaluminium (TMA) \\n2. Dimethylaluminium isopropoxide (DMAI)\\n3. Triisopropyl borate (TIB)\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"From the figure, comparing all three precursors, TIB results in the lowest incorporation of doping elements while TMA shows the greatest doping incorporation for same dopant cycle ratio.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"This is in agreement with the data obtained by RBS/PIGE analysis, where the areal densities of B or Al per cycle were, on average, ~3.2 at per nm2 per supercycle (TIB), ~4.4 at per nm2 per supercycle (DMAI) and ~9.0 at per nm2 per supercycle (TMA ) for the cycle ratios (m) investigated.  A  combination of precursor reactivity and molecule size is most likely the cause of this trend. The bulkier isopropyl ligands in the TIB and DMAI precursors are likely to cause steric hindrance on the film surface, resulting in a more sparse distribution of the dopants on the surface and therefore a lower doping fraction.36 This is expected from the decreasing size of the monomers (TIB > DMAI > TMA).\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It is worth highlighting that the doping concentrations do not relate directly to the doping efficiency,\\nas some of the doping material might have formed metal oxide or alloy clusters (e.g. Al2O3) rather than merely doping the film. In this case, the dopant atoms do not contribute donating any electrons to the ZnO film.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":5,"width":620,"height":451}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/images/fig_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/images/fig_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/Garcia-Alonso et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"50","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":625,"height":463,"image_format":"jpeg","image_sha256":"34f16fb1abd232eb3f78ef62f3f46346bfef7936d22c1c249cccb57a7d0962ae","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_50_fig_5.jpg","caption":"Fig. 5 Resistivity,  $\\rho$  (a), carrier concentration,  $n$ , and mobility,  $\\mu$  (b) and doping fraction, DF, and doping efficiency,  $\\eta$  (c) of the ZnO:B films as a function of the deposition temperature. The films with thicknesses of  $45 \\pm 5 \\mathrm{nm}$  were deposited on glass substrates. A cycle ratio of  $m = 24$  was employed as it resulted in optimised resistivity in the doping series at  $150^{\\circ}\\mathrm{C}$ . The lines serve as guides to the eye. The estimated relative statistical uncertainties for  $n$ ,  $\\mu$ , DF and  $\\eta$  are  $3\\%$ ,  $5\\%$ ,  $10\\%$  and  $20\\%$  respectively.","id":"train/atomic-layer-deposition/experimental-usecase/50/fig_5","sample_id":"atomic-layer-deposition/experimental-usecase/50/fig_5","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"multi-axis chart"},{"panel_id":"c","label":"multi-axis chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The resistivity  as a funtion of temperature is plotted and it decreases as temperature increases.\"},{\"panel_id\":\"b\",\"text\":\"Both carrier concentration and mobility plotted against temperature increase with temperature, but the mobility decreases slightly at higher temperatures.\"},{\"panel_id\":\"c\",\"text\":\"Doping efficiency versus temperature shows it decreases while doping fraction increases with temperature.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Temperature (°C) | Resistivity, ρ (mΩ cm) |\\n|---|---|\\n| 150 | 3.5 |\\n| 200 | 2.5 |\\n| 250 | 2.2 |\"},{\"panel_id\":\"b\",\"text\":\"| Temperature (°C) | Carrier concentration, n (10^20 cm^-3) | Mobility, μ (cm²V^-1s^-1) |\\n|---|---|---|\\n| 150 | 2.3 | 8 |\\n| 200 | 2.6 | 11 |\\n| 250 | 2.8 | 10 |\"},{\"panel_id\":\"c\",\"text\":\"| Temperature (°C) | Doping fraction, DF | Doping Efficiency, η (%) |\\n|---|---|---|\\n| 150 | 0.02 | 30 |\\n| 200 | 0.025 | 25 |\\n| 250 | 0.03 | 20 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Compared to ZnO:Al films deposited at 250 °C, it appeared that processing at 150 °C led to lower absolute values of conductivity (i.e. a higher resistivity)\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The doping fraction increased with increasing deposition temperature, whereas the doping efficiency decreased slightly, despite the slight increase in carrier concentration with temperature.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It can be concluded that ~200 °C is the optimum deposition temperature for ZnO:B. However, it should be noted that the deposition temperature that is feasible for certain applications can be limited to lower temperatures than 200 °C.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The resistivity of the ZnO:B films decreased from 3.5 mOhm cm at 150 °C to 2.2 mOhm cm when deposited at temperatures of 200–240 °C. This was due to an increase in carrier concentration from 2.2* 10^20 /cm3 to 2.8* 10^20 /cm3 and an increase in mobility from 8 to 10.5 cm2 /Vs.\"}]}]","bbox":[{"panel_id":"a","x":2,"y":3,"width":538,"height":313},{"panel_id":"b","x":0,"y":335,"width":584,"height":370},{"panel_id":"c","x":2,"y":707,"width":579,"height":355}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/images/fig_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/images/fig_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/Garcia-Alonso et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"50","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":586,"height":1070,"image_format":"jpeg","image_sha256":"e0f526ccecc8cbd299c326a242d068ea3ec7c7a6a124a0e2200e9e14a0b77318","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_50_fig_7.jpg","caption":"Fig. 7 (a) Intensity ratio  $I_{002} / I_{100}$  of the (002) and (100) peaks measured by XRD on the ZnO:X  $(X = \\mathsf{Al},\\mathsf{B})$  films reflecting the change in texture from  $\\langle 002\\rangle$  to  $\\langle 001\\rangle$  and (b)  $c$  -lattice parameter of the ZnO wurtzite unit cell and the corresponding  $2\\theta$  position of the (002) peak as a function of the doping fraction in the ZnO:X  $(X = \\mathsf{Al},\\mathsf{B})$  . The films were deposited at  $150^{\\circ}C$  on glass substrates. The  $c$  -lattice parameter of the wurtzite ZnO crystal is  $5.207\\mathrm{\\AA}$  6","id":"train/atomic-layer-deposition/experimental-usecase/50/fig_7","sample_id":"atomic-layer-deposition/experimental-usecase/50/fig_7","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"multi-axis chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows the variation of I_002/I_100 ratio  with doping fraction\"},{\"panel_id\":\"b\",\"text\":\"The figure shows the variation of  c-lattice parameter with doping fraction, along with the 2θ_002 angle.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Doping fraction, DF | I_002/I_100 TIB |I_002/I_100 DMAI |\\n|---|---|---|\\n| 0.00 | 2.20 | 2.20 | \\n| 0.02 | 1.80 | 1.30 | \\n| 0.04 | 1.60 | 0.60 | \\n| 0.06 | 1.30 | 0.50 | \\n| 0.08 | 0.80 | -        | \\n| 0.10 | 0.40 | -        |\"},{\"panel_id\":\"b\",\"text\":\"| Doping fraction, DF | c-lattice parameter, Å | 2θ_002 |\\n|---|---|---|\\n| 0.00 | 5.22 | 34 |\\n| 0.02 |  5.20 | 34.2 |\\n| 0.04 |  5.18 | 34.4 |\\n| 0.06 |  5.16 | 34.6 |\\n| 0.08 |  5.14 | - |\\n| 0.10 | 5.12 | - |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"It is a guide to the eye, the ratio I002/I100 for a powder pattern, i.e., the pattern characteristic for a film with randomly oriented crystals, has been included to illustrate the change in texture from (002) to (100).\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The ratio I002/I100 for a powder pattern, i.e., the pattern characteristic for a film with randomly oriented crystals, has been included to illustrate the change in texture from (002) to (100). Secondly,\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The shift of the (002) peak as a function of the doping fraction, as a consequence of the\\ncompression of the unit cell (i.e. a decrease of the c-lattice parameter) with increasing doping fractions. It should be noticed that the c-lattice parameter for the pure ZnO films (i.e., DF = 0) was slightly larger (5.219 Å) than the value reported for the wurtzite ZnO crystal lattice (5.207 Å).\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"A significant shift of the peaks to  higher 2theta values was observed for increasing B fractions, which\\nwas likely due to compressive stress in the ZnO unit cell caused by the substitution of Zn2+ (ionic radius of 0.60 Å, assuming a four-coordinate environment as expected for wurtzite structures)\\n132 with smaller B3+ ions (four-coordinate ionic radius of 0.11 Å).132 This effect was significantly smaller in the case of ZnO:Al films doped by DMAI (Al3+, four-coordinate ionic radius of 0.39 Å).\"}]}]","bbox":[{"panel_id":"a","x":0,"y":5,"width":565,"height":272},{"panel_id":"b","x":4,"y":293,"width":618,"height":326}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/images/fig_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/images/fig_7.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/50/Garcia-Alonso et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"50","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":627,"height":619,"image_format":"jpeg","image_sha256":"1a87cbb32c784f80fe2dda489bdb6aea86e83904cf509e7e88f4c776782e08e6","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_54_figure_11.jpg","caption":"Figure 11. (Color online) Galvanostatic cycling for TiN prepared by sputtering and remote plasma ALD at a constant current of  $3\\mu \\mathrm{A} / \\mathrm{cm}^2$ . The remote plasma ALD TiN film shows a lower capacity for Li, indicating better barrier properties.","id":"train/atomic-layer-deposition/experimental-usecase/54/figure_11","sample_id":"atomic-layer-deposition/experimental-usecase/54/figure_11","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart shows the Li-capacity (µAh/cm²) of two different TiN samples over galvanostatic cycles. The blue circles represent 'Sputtered TiN', while the red circles represent 'TiN (H₂-N₂ 30:4 plasma)'. Both samples exhibit a slight increase in capacity over time.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Galvanostatic cycles | Li-capacity (µAh/cm²) Sputtered TiN |  Li-capacity (µAh/cm²) TiN | \\n|---|---|---|\\n| 0 | 0.3 | 0.2 |\\n| 5 | 0.3 | 0.2 |\\n| 10 | 0.3 | 0.2 |\\n| 15 | 0.3 | 0.2 |\\n| 20 | 0.3 | 0.2 |\\n| 25 | 0.3 | 0.2 |\\n| 30 | 0.3 | 0.2 |\\n| 35 | 0.3 | 0.2 |\\n| 40 | 0.3 | 0.2 |\\n| 45 | 0.3 | 0.2 |\\n| 50 | 0.3 | 0.2 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Sputtered TiN and ALD-grown TiN.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"50.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.3  µAh/cm².\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.2  µAh/cm².\"}]}]","bbox":[{"panel_id":"a","x":2,"y":2,"width":661,"height":506}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/54/images/figure_11.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/54/images/figure_11.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/54/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/54/Knoops et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"54","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":666,"height":509,"image_format":"jpeg","image_sha256":"53de9621c45afeae93eb0522998afddf3a1245ea518e7592e560845db1b8ebbc","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_59_fig_4.jpg","caption":"FIG. 4. (Color online) Growth rate per cycle of  $\\mathrm{TaN}_{x,x\\leqslant 1}$  films as a function of the ALD process parameters: (a) PDMAT dosing time, (b)  $\\mathrm{H}_{2}$  plasma exposure time, and (c) deposition temperature. The growth rates are determined by in situ spectroscopic ellipsometry by varying the process parameters during one deposition run (closed symbols) and by the deposition of a thick film under constant process conditions (open symbols). The lines in (a) and (b) are exponential fits to the data.","id":"train/atomic-layer-deposition/experimental-usecase/59/fig_4","sample_id":"atomic-layer-deposition/experimental-usecase/59/fig_4","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"},{"panel_id":"c","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows the growth rate (nm/cycle) versus the PDMAT dosing (s). The growth rate increases with increasing PDMAT dosing time until it plateaus.\"},{\"panel_id\":\"b\",\"text\":\"The figure shows the growth rate (nm/cycle) versus the plasma exposure time (s). The growth rate increases with increasing plasma exposure time.\"},{\"panel_id\":\"c\",\"text\":\"The figure shows the growth rate (nm/cycle) versus the deposition temperature (°C). The growth rate remains constant across different deposition temperatures.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| PDMAT dosing (s) | Growth rate (nm/cycle) |\\n|---|---|\\n| 0 | 0.00 |\\n| 1 | 0.035 |\\n| 2 | 0.035 |\\n| 3 | 0.040 |\\n| 4 | 0.040 |\\n| 5 | 0.040 |\"},{\"panel_id\":\"b\",\"text\":\"| Plasma exposure time (s) | Growth rate (nm/cycle) |\\n|---|---|\\n| 0 | 0.00 |\\n| 5 | 0.04 |\\n| 10 | 0.042 |\\n| 15 | - |\\n| 20 | - |\\n| 25 | - |\\n| 30 | 0.05 |\"},{\"panel_id\":\"c\",\"text\":\"| Deposition temperature (°C) | Growth rate (nm/cycle) |\\n|---|---|\\n| 150 | 0.05 |\\n| 175 | 0.05 |\\n| 200 | 0.05 |\\n| 225 | 0.05 |\\n| 250 | 0.05 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The saturation point can be observed after approximately three seconds of PDMAT.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"To analyze the growth rate for the ALD process, the deposition temperature was changed from 150 °C to 250 °C.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The growth rate remains constant across deposition temperatures.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Yes.\"}]}]","bbox":[{"panel_id":"a","x":7,"y":17,"width":603,"height":437},{"panel_id":"b","x":5,"y":482,"width":605,"height":447},{"panel_id":"c","x":10,"y":950,"width":600,"height":435}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/59/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/59/images/fig_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/59/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/59/Langereis et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"59","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":611,"height":1403,"image_format":"jpeg","image_sha256":"3c028150e7a3abbc5a36e11fe962058bf2dc6405ab120e291e998cf487bc0d69","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_59_fig_6.jpg","caption":"FIG. 6. (Color online) Resistivity of  $\\mathrm{TaN}_{x,x\\leqslant 1}$  films with an  $\\sim 30 \\mathrm{nm}$  thickness deposited by remote plasma ALD using different  $\\mathrm{H}_{2}$  plasma exposure times within the cycle. The resistivity is determined from in situ SE (closed symbols) and ex situ FPP measurements (open symbols). The line serves as a guide to the eyes. The low-resistivity  $\\mathrm{TaN}_{x,x\\leqslant 1}$  film deposited using  $3\\mathrm{~s~H}_{2}$  plasma in the ALD cycle has been indicated separately (open triangle).","id":"train/atomic-layer-deposition/experimental-usecase/59/fig_6","sample_id":"atomic-layer-deposition/experimental-usecase/59/fig_6","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the relationship between resistivity and plasma exposure time (s). The fitted curve indicates that resistivity decreases as plasma exposure time increases.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Plasma exposure time (s) | In-situ SE | Ex-situ SE |\\n| --- | --- | --- |\\n| 0 | 10^4 | 10^5 |\\n| 5 | - | - |\\n| 10 | 10^3 | 10^3 |\\n| 15 | - | - |\\n| 20 | - | - |\\n| 25 | - | - |\\n| 30 | - | 6*10^2 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"This process corresponds to a remote plasma ALD.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The resistivity decreases as the plasma exposure time increases for both measurements (in-situ and ex-situ SE).\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The lowest resistivity is observed after 30 s of plasma exposure.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"There is a slight difference between the two SE measurements. The ex-situ SE shows higher resistivity values, which could be related to environmental contamination.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":3,"width":601,"height":462}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/59/images/fig_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/59/images/fig_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/59/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/59/Langereis et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"59","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":605,"height":466,"image_format":"jpeg","image_sha256":"a6691fc810289d1e368efd517c9850903a722f0cc482e1f0b0632e3620a3aef2","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_60_figure_2.jpg","caption":"Figure 2. Saturation plots of ALD  $\\mathrm{Y}_2\\mathrm{O}_3$  films at a substrate temperature  $200^{\\circ}\\mathrm{C}$  with  $\\mathrm{Y}(\\mathrm{EtCp})_2(\\mathrm{iPr - amd})$  and (a)  $\\mathrm{H}_2\\mathrm{O}$ , (b)  $\\mathrm{O}_2$  plasma, and (c)  $\\mathrm{H}_2\\mathrm{O}$  plasma as a reactant. (d) Film thickness determined by in situ SE versus the ALD cycle number under saturated conditions.","id":"train/atomic-layer-deposition/experimental-usecase/60/figure_2","sample_id":"atomic-layer-deposition/experimental-usecase/60/figure_2","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"},{"panel_id":"c","label":"scatter plot"},{"panel_id":"d","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Scatter plot showing the growth per cycle  (Å/cycle) of Y_2O_3 films as a function of pulse time for precursors Y(EtCp)_2(iPr-amd) and H_2O. The data confirms self-limiting surface reactions\"},{\"panel_id\":\"b\",\"text\":\"Scatter plot showing the growth per cycle  (Å/cycle) of Y_2O_3 films as a function of pulse time for precursors Y(EtCp)_2(iPr-amd) and O_2 plasma. The data confirms self-limiting surface reactions\"},{\"panel_id\":\"c\",\"text\":\"Scatter plot showing the growth per cycle  (Å/cycle) of Y_2O_3 films as a function of pulse time for precursors Y(EtCp)_2(iPr-amd) and H_2O plasma. The data confirms self-limiting surface reactions\"},{\"panel_id\":\"d\",\"text\":\"The chart showing the thickness of the Y_2O_3 films as a function of ALD cycle number for H_2O plasma, O_2 plasma, and H_2O processes.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Pulse time (s) | GPC (Å/cycle) | Precursor|\\n|---|---|---|\\n| 0,00 | 0,00 | Water |\\n| 1,02 | 0,35 | Water |\\n| 1,93 | 0,52 | Water |\\n| 3,86 | 0,62 | Water |\\n| 6,10 | 0,70 | Water |\\n| 7,83 | 0,74 | Water |\\n| 9,97 | 0,73 | Water |\\n| 15,05 | 0,73 | Water |\\n| 18,00 | 0,72 | Water |\\n| 20,03 | 0,72 | Water |\\n| 0,00 | 0,00 | Y(EtCp)2(iPr-amd) |\\n| 1,93 | 0,32 | Y(EtCp)2(iPr-amd) |\\n| 5,19 | 0,43 | Y(EtCp)2(iPr-amd) |\\n| 8,03 | 0,53 | Y(EtCp)2(iPr-amd) |\\n| 9,86 | 0,62 | Y(EtCp)2(iPr-amd) |\\n| 12,10 | 0,70 | Y(EtCp)2(iPr-amd) |\\n| 14,85 | 0,74 | Y(EtCp)2(iPr-amd) |\\n| 18,20 | 0,71 | Y(EtCp)2(iPr-amd) |\\n| 20,14 | 0,73 | Y(EtCp)2(iPr-amd) |\\n| 24,81 | 0,71 | Y(EtCp)2(iPr-amd) |\\n| 29,69 | 0,73 | Y(EtCp)2(iPr-amd) |\"},{\"panel_id\":\"b\",\"text\":\"| Pulse time (s) | GPC (Å/cycle) | Precursor|\\n|---|---|---|\\n| 0,00 | 0,00 | O2 Plasma |\\n| 0,94 | 0,37 | O2 Plasma |\\n| 1,89 | 0,54 | O2 Plasma |\\n| 2,94 | 0,66 | O2 Plasma |\\n| 3,99 | 0,73 | O2 Plasma |\\n| 5,03 | 0,81 | O2 Plasma |\\n| 5,77 | 0,84 | O2 Plasma |\\n| 8,29 | 0,86 | O2 Plasma |\\n| 10,17 | 0,85 | O2 Plasma |\\n| 11,96 | 0,84 | O2 Plasma |\\n| 0,00 | 0,00 | Y(EtCp)2(iPr-amd) |\\n| 1,99 | 0,46 | Y(EtCp)2(iPr-amd) |\\n| 4,83 | 0,66 | Y(EtCp)2(iPr-amd) |\\n| 8,18 | 0,71 | Y(EtCp)2(iPr-amd) |\\n| 9,86 | 0,80 | Y(EtCp)2(iPr-amd) |\\n| 11,96 | 0,84 | Y(EtCp)2(iPr-amd) |\\n| 15,10 | 0,87 | Y(EtCp)2(iPr-amd) |\\n| 17,94 | 0,89 | Y(EtCp)2(iPr-amd) |\\n| 19,72 | 0,87 | Y(EtCp)2(iPr-amd) |\\n| 25,17 | 0,87 | Y(EtCp)2(iPr-amd) |\\n| 30,00 | 0,86 | Y(EtCp)2(iPr-amd) |\"},{\"panel_id\":\"c\",\"text\":\"| Pulse time (s) | GPC (Å/cycle) | Precursor|\\n|---|---|---|\\n| 0,00 | 0,00 | H2O Plasma |\\n| 0,73 | 0,53 | H2O Plasma |\\n| 1,99 | 0,76 | H2O Plasma |\\n| 3,67 | 0,91 | H2O Plasma |\\n| 5,98 | 0,99 | H2O Plasma |\\n| 7,66 | 1,08 | H2O Plasma |\\n| 9,86 | 1,11 | H2O Plasma |\\n| 11,64 | 1,06 | H2O Plasma |\\n| 14,58 | 1,07 | H2O Plasma |\\n| 0,00 | 0,06 | Y(EtCp)2(iPr-amd) |\\n| 0,84 | 0,44 | Y(EtCp)2(iPr-amd) |\\n| 2,94 | 0,69 | Y(EtCp)2(iPr-amd) |\\n| 5,03 | 0,79 | Y(EtCp)2(iPr-amd) |\\n| 7,87 | 0,88 | Y(EtCp)2(iPr-amd) |\\n| 9,86 | 0,95 | Y(EtCp)2(iPr-amd) |\\n| 11,64 | 0,99 | Y(EtCp)2(iPr-amd) |\\n| 14,69 | 1,10 | Y(EtCp)2(iPr-amd) |\\n| 19,93 | 1,04 | Y(EtCp)2(iPr-amd) |\\n| 24,65 | 1,00 | Y(EtCp)2(iPr-amd) |\\n| 29,69 | 1,06 | Y(EtCp)2(iPr-amd) |\"},{\"panel_id\":\"d\",\"text\":\"| ALD cycle number | Thickness (nm) | Precursor|\\n|---|---|---|\\n| 0.00 | 0.0 | H2O |\\n| 40 | 2.65 | H2O |\\n| 68 | 5.05 | H2O |\\n| 100 | 7.68 | H2O |\\n| 0 | 0 | H2O Plasma |\\n| 42 | 4.41 | H2O Plasma |\\n| 100 | 10.39 | H2O Plasma |\\n| 0 | 0 | O2 Plasma |\\n| 50 | 4.88 | O2 Plasma |\\n| 100 | 8.67 | O2 Plasma |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Water achieves saturation faster than the yttrium precursor. The graph shows that the growth per cycle (GPC) for water stabilizes at its saturation point with a pulse time of approximately 8 seconds. In contrast, the yttrium precursor requires a longer pulse time of about 12 seconds to reach  saturation level.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"After ~5 seconds.\"}]},{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"H2O plasma.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"H2O: ~0.74 nm/cycle, O2 palsma: ~0.83 nm/cycle, H2O palsma: ~0.103 nm/cycle\"}]}]","bbox":[{"panel_id":"a","x":2,"y":0,"width":430,"height":340},{"panel_id":"b","x":442,"y":4,"width":404,"height":335},{"panel_id":"c","x":4,"y":336,"width":442,"height":331},{"panel_id":"d","x":452,"y":336,"width":391,"height":330}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/60/images/figure_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/60/images/figure_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/60/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/60/Zhao et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"60","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":847,"height":669,"image_format":"jpeg","image_sha256":"117f6d27ab4f959906f363d46b0624650c5f3dab2f599955a45f228e1d594962","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_60_figure_8.jpg","caption":"Figure 8. Water contact angles of as-deposited and  $\\mathrm{Y}_2\\mathrm{O}_3$  films with a thickness of 5, 10, 20, and  $30~\\mathrm{nm}$  with (a)  $\\mathrm{H}_2\\mathrm{O}$ , (b)  $\\mathrm{O}_2$  plasma, (c)  $\\mathrm{H}_2\\mathrm{O}$  plasma, and (d) the surface free energy.","id":"train/atomic-layer-deposition/experimental-usecase/60/figure_8","sample_id":"atomic-layer-deposition/experimental-usecase/60/figure_8","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"},{"panel_id":"c","label":"scatter plot"},{"panel_id":"deposition","label":"multiple line chart"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":4,"y":3,"width":442,"height":340},{"panel_id":"b","x":458,"y":0,"width":423,"height":350},{"panel_id":"c","x":10,"y":356,"width":435,"height":328},{"panel_id":"d","x":468,"y":349,"width":413,"height":338}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/60/images/figure_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/60/images/figure_8.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/60/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/60/Zhao et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"60","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":886,"height":689,"image_format":"jpeg","image_sha256":"6b568ab6324ec1cc7ac650bfe1a52472c33d48d31a545a31b0ea12ac3b13c24b","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_60_figure_9.jpg","caption":"Figure 9. Water contact angles of as-deposited  $30~\\mathrm{nm}$  thick  $\\mathrm{Y}_2\\mathrm{O}_3$  surfaces (a) before and after annealing at  $500^{\\circ}\\mathrm{C}$  for  $2\\mathrm{~h}$  in static air and (b) after treatment with hexane, toluene, ethanol, and isopropanol for  $24\\mathrm{~h}$ .","id":"train/atomic-layer-deposition/experimental-usecase/60/figure_9","sample_id":"atomic-layer-deposition/experimental-usecase/60/figure_9","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[]","data_extraction":"[]","vqa":"[]","bbox":[{"panel_id":"a","x":8,"y":4,"width":449,"height":330},{"panel_id":"b","x":462,"y":5,"width":423,"height":326}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/60/images/figure_9.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/60/images/figure_9.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/60/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/60/Zhao et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"60","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":889,"height":336,"image_format":"jpeg","image_sha256":"a672ee10f52f2cc0bcea20718393f0b8ff38c53c880882a7f2144296138af89c","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_63_fig_1.jpg","caption":"FIG. 1. Growth rate of  $\\mathrm{Ga}_2\\mathrm{O}_3$  thin films as a function of (a)  $\\mathrm{O}_2$  plasma flow duration at  $250^{\\circ}\\mathrm{C}$  and (b) deposition temperature. TMG dose and  $\\mathrm{O}_2$  plasma flow rate were constant at  $0.015\\mathrm{s}$  and  $25\\mathrm{scm}$  respectively. (c)  $\\mathrm{Ga}_2\\mathrm{O}_3$  film thickness as a function of the number of PEALD cycles.","id":"train/atomic-layer-deposition/experimental-usecase/63/fig_1","sample_id":"atomic-layer-deposition/experimental-usecase/63/fig_1","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"},{"panel_id":"c","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The deposition rate increases rapidly with increasing O₂ plasma duration up to 10 seconds, then plateaus.\"},{\"panel_id\":\"b\",\"text\":\"The deposition rate remains relatively constant across a wide range of temperatures, with a slight decrease at higher temperatures.\"},{\"panel_id\":\"c\",\"text\":\"Film thickness increases linearly with the number of PEALD cycles.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| O2 plasma duration (s) | Deposition rate (nm/cycle) |\\n|---|---|\\n| 2 | 0.020 |\\n| 5 | 0.043 |\\n| 10 | 0.051 |\\n| 20 | 0.053 |\\n| 40 | 0.051 |\\n| 60 | 0.052 |\"},{\"panel_id\":\"b\",\"text\":\"| Temperature (°C) | Deposition rate (nm/cycle) |\\n|---|---|\\n| 30 | 0.061 |\\n| 100 | 0.054 |\\n| 150 | 0.053 |\\n| 200 | 0.053 |\\n| 250 | 0.053 |\\n| 300 | 0.053 |\\n| 350 | 0.052 |\\n| 400 | 0.053 |\"},{\"panel_id\":\"c\",\"text\":\"| Number of PEALD cycles | Film thickness (nm) |\\n|---|---|\\n| 50 | 3.0 |\\n| 150 | 8.0 |\\n| 300 | 16.0 |\\n| 500 | 26.5 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The deposition rate rises rapidly from about 0.02 nm per cycle at 2 s to about 0.05 nm per cycle by about 10 s, then stays near that level up to 60 s. Saturation begins at roughly 10 s.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"The growth per cycle is approximately constant over the tested cycle counts.\\n\\nThere is no strong incubation delay in thickness accumulation over this range.\\n\\nThe process behaves in a self limiting manner consistent with ALD type growth.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The squares are measured deposition rate values at different substrate temperatures. The dashed line indicates the approximate constant growth per cycle within the ALD temperature window, about 0.53 Å per cycle.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"A broad temperature window with stable growth per cycle makes the process easier to transfer across tools and substrates because small temperature variations have limited impact on thickness control. It also enables deposition on temperature sensitive substrates by selecting a lower temperature while keeping similar thickness per cycle. In practice, this supports more predictable film thickness targeting across devices and wafer areas.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":2,"width":548,"height":417},{"panel_id":"b","x":4,"y":424,"width":542,"height":419},{"panel_id":"c","x":0,"y":851,"width":547,"height":411}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/63/images/fig_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/63/images/fig_1.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/63/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/63/Donmez et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"63","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":550,"height":1266,"image_format":"jpeg","image_sha256":"3c23f5ec3c13d73347c70c566bfed5c5845d398f83040cb4bfaa5cf544dd39af","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_64_fig_2.jpg","caption":"FIG. 2. Growth rate of the platinum films grown from  $\\mathrm{MeCpPtMe}_3$  as a function of the deposition temperature, the film thicknesses being measured by EDX and XRR. The  $\\mathrm{MeCpPtMe}_3$  and oxygen pulse times were  $0.7$  and  $1.0\\mathrm{s}$ , respectively, and the purge period between the pulses was  $0.7\\mathrm{s}$ .","id":"train/atomic-layer-deposition/experimental-usecase/64/fig_2","sample_id":"atomic-layer-deposition/experimental-usecase/64/fig_2","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows the growth rate of platinum films deposited from MeCpPtMe₃ as a function of deposition temperature (200–300 °C). The film thicknesses were measured using EDX and XRR. The plot illustrates that the growth rate increases with temperature, from about 0.30 Å/cycle at 200 °C to 0.51 Å/cycle at 300 °C. Pulse and purge times for the precursor and oxygen were 0.7–1.0 s.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Deposition temperature (°C) | Growth rate (Å cycle⁻¹) |\\n|---|---|\\n| 200 | 0.30 |\\n| 220 | 0.36 |\\n| 240 | 0.41 |\\n| 260 | 0.45 |\\n| 280 | 0.48 |\\n| 300 | 0.51 |\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Using pure oxygen instead of air allows deposition of platinum films at significantly lower temperatures. In previous work using air, no film was obtained below 300 °C, whereas with pure oxygen, films could be grown starting at 200 °C. The higher oxygen partial pressure is key to enabling film growth at these lower temperatures without affecting the maximum growth rate at 300 °C.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At 300 °C.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Yes. The growth rate increases as the deposition temperature increases from 200 °C to 300 °C.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Deposition temperature (200–300 °C), Oxygen partial pressure (higher in current study compared to previous air-based study), Pulse times of MeCpPtMe₃ (0.7 s) and oxygen (1.0 s), Purge time between pulses (0.7 s)\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":528,"height":359}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/64/images/fig_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/64/images/fig_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/64/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/64/Aaltonen et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"64","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":528,"height":359,"image_format":"jpeg","image_sha256":"78eed6a2552c212122e78187b5cdcd19473f4a4c14aae1902d89d19b9ab3a46b","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_69_FIG1_a.jpg","caption":"","id":"train/atomic-layer-deposition/experimental-usecase/69/FIG1_a","sample_id":"atomic-layer-deposition/experimental-usecase/69/FIG1_a","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"A scatter plot showing the linear increase in thickness (in nm) as a function of ALD cycle number.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| ALD Cycle (#) | Thickness (nm) |\\n|---|---|\\n| 0 | 0 |\\n| 20 | 2.5 |\\n| 40 | 5 |\\n| 60 | 7 |\\n| 80 | 9 |\\n| 100 | 11 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"300 °C.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Roughly 9 nm.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"For both O2 plasma exposure steps, roughly 0.16 nm is removed.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"During the TDMAP and TMA exposure steps, material is deposited, roughly 0.20 and 0.24 nm, respectively. During O2-plasma exposure, material is removed, roughly 0.16 and 0.17 nm for the first and second exposure, respectively. This yields a net growth of 0.11 nm/cycle.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":4,"width":673,"height":382}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/images/FIG1_a.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/images/FIG1_a.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/Ruben Blomme et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"69","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"not_found"},"width":703,"height":392,"image_format":"jpeg","image_sha256":"3e440481ef613a49314e31f8ac3cf4194a2ac140076c8a554fb17b03009458ec","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_69_FIG5.jpg","caption":"FIG. 5. In situ ellipsometry data for the different combinations of precursors that were tested at  $300^{\\circ}\\mathrm{C}$ . Depicted are the thicknesses obtained from the SE measurements conducted at the end of each full cycle.","id":"train/atomic-layer-deposition/experimental-usecase/69/FIG5","sample_id":"atomic-layer-deposition/experimental-usecase/69/FIG5","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows the thickness of as a function of the number of ALD cycles for 5 different precursor combinations.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| ALD Cycle (#) | Thickness (nm) TDMAP - TMA | Thickness (nm) TDMAP - O₂* | Thickness (nm) TDMAP - O₂* - TMA | Thickness (nm) TDMAP - TMA - O₂* | Thickness (nm) TDMAP - O₂* - TMA - O₂* |\\n|---|---|---|---|---|---|\\n| 0 | 0 | 0 | 0 | 0 | 0 |\\n| 20 | 0 | 2.0 | 2.0 | 2.0 | 3.5 |\\n| 40 | 0 | 2.0 | 4.0 | 4.0 | 6.0 |\\n| 60 | 0 | 2.0 | 6.0 | 6.0 | 8.0 |\\n| 80 | 0 | 2.0 | 7.0 | 7.0 | 10.0 |\\n| 100 | 0 | 2.0 | 8.0 | 8.0 | 12.0 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"TP–O–TAl–O (TDMAP–O₂*–TMA–O₂*).\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The TP–O–TAl–O process yields a phosphorus content of approximately 11%, which is very similar to that obtained for the TP–TAl–O process.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The TP–TAl process shows no sustained growth beyond an initial thickness increase of about 0.37 nm after the first cycle, indicating very low reactivity without a coreactant.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The TP–O (TDMAP–O₂*) process shows limited growth that saturates at a total film thickness of approximately 2 nm.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":4,"width":674,"height":450}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/images/FIG5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/images/FIG5.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/Ruben Blomme et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"69","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":678,"height":456,"image_format":"jpeg","image_sha256":"bd145982728ec2d269e69b6df440afd9ad141a7374a762806439a0465648b2a4","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_69_FIG6_a.jpg","caption":"FIG. 7. In situ ellipsometry data of three processes with different AIO-PO ratios. As the contribution of PO increases, the overall growth of the material also increases.","id":"train/atomic-layer-deposition/experimental-usecase/69/FIG6_a","sample_id":"atomic-layer-deposition/experimental-usecase/69/FIG6_a","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot displays the ratio of oxygen to phosphorus (O/P) against various elements (Al, Ti, Ca, Li, Fe, La, Zn, Co, Na, Mg, V, Ni). The points are color-coded, and there's a horizontal dashed line indicating a specific value.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| M (Element) | TMP | TMP* | DEPA | DEPA* | TDMAP | This work |\\n|---|---|---|---|---|---|---|\\n| Al | 4.8 | 3.4 |  |  |  | 5.6 |\\n| Ti | 3.9 | 2.6 |  | 1.9 |  |  |\\n| Ca | 3.7 | 3.3 |  |  |  |  |\\n| Li | 3.8 | 2.2 | 2.4 |  |  |  |\\n| Fe | 3.9 | 2.5 |  |  |  |  |\\n| La | 4.9 |  |  |  |  |  |\\n| Zn |  | 2.0 |  |  |  |  |\\n| Co | 4.5 | 2.9 |  |  |  |  |\\n| Na |  |  | 3.7 |  |  |  |\\n| Mg |  |  |  |  | 3.2 |  |\\n| V  |  | 3.4 |  |  |  |  |\\n| Ni |  | 2.8 |  |  |  |  |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The aluminum phosphate deposited in this work has a P/M ratio of 0.51 and an O/P ratio of 5.65, which deviates strongly from an ideal orthophosphate composition (P/M = 1 and O/P = 4), indicating phosphorus deficiency and oxygen excess.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The aluminum phosphate in this work shows a lower P/M ratio (0.51 vs 0.67) and a higher O/P ratio (5.65 vs 3.16) than the TDMAP-based magnesium phosphate, highlighting a substantial difference in phosphorus incorporation between the two systems.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The comparison demonstrates that both the phosphorous precursor and the metal precursor significantly influence the final film composition. In particular, the TDMAP-based aluminum phosphate deposited in this work exhibits an unusually low P/M ratio and a very high O/P ratio, placing it far from the stoichiometry of an orthophosphate. This behavior contrasts with TMP-, TMP*-, DEPA-, and DEPA*-based processes, which generally yield compositions closer to P/M = 1. The stark difference with the TDMAP-based magnesium phosphate further suggests that the metal precursor itself can dictate growth chemistry and ligand exchange efficiency in ALD, reinforcing that phosphorus incorporation is not solely controlled by the phosphorous precursor but emerges from the combined metal–phosphorus chemistry.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Processes based on TMP, TMP*, DEPA, and DEPA* generally show higher P/M ratios than TDMAP-based processes, indicating more efficient phosphorus incorporation.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":650,"height":423}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/images/FIG6_a.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/images/FIG6_a.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/Ruben Blomme et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"69","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":650,"height":423,"image_format":"jpeg","image_sha256":"cf6c54d202284429dd0ac51de342cf1b266b7fd49eab7dd3077b4f91abfee166","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_69_FIG7.jpg","caption":"","id":"train/atomic-layer-deposition/experimental-usecase/69/FIG7","sample_id":"atomic-layer-deposition/experimental-usecase/69/FIG7","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart illustrates the growth of thickness over ALD supercycles for different AIO-PO ratios.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| ALD Supercycle (#) | Thickness (nm) AlO-PO (1-1) | Thickness (nm) AlO-PO (1-3) | Thickness (nm) AlO-PO (1-6) |\\n|---|---|---|---|\\n| 0 | 0 | 0 | 0 |\\n| 20 | 3 | 5 | 7 |\\n| 40 | 5 | 8 | 12 |\\n| 60 | 8 | 12 | 16 |\\n| 80 | 10 | 16 | 22 |\\n| 100 | 13 | 20 | 28 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Roughly 7 nm.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Roughly 13 nm.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Roughly 17 nm.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Figure 7 illustrates that splitting the original four-step process into separate TDMAP–O₂* and TMA–O₂* cycles and combining them in a supercycle provides a practical lever to control film composition and growth behavior. Increasing the number of TDMAP–O₂* cycles per supercycle (from 1 to 3 and 6, with one TMA–O₂* cycle fixed) systematically increases both the growth per supercycle and the phosphorus content of the deposited film. Unlike the standalone TP–O process, which saturates at a thickness of approximately 2 nm, the periodic insertion of a TMA–O₂* cycle effectively resets the surface chemistry, allowing continued chemisorption and further growth during subsequent TDMAP–O₂* cycles. As a result, the supercycle approach overcomes growth saturation while enabling tunable phosphorus incorporation, analogous to previously reported TMP-based supercycle strategies.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":672,"height":456}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/images/FIG7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/images/FIG7.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/69/Ruben Blomme et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"69","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"not_found"},"width":672,"height":456,"image_format":"jpeg","image_sha256":"8c73efde7c9f7a1c19019166b8c31da91f6cc551c669baf7e46fce65365c8915","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_8_figure_3.jpg","caption":"Figure 3. Process development steps of Eu-HQA thin films that show temperature scanning (left), surface saturation of precursors (middle), and thickness control over the number of ALD/MLD cycles (right). (Reproduced under CC licence from ref. [55], 2023, Royal society of chemistry).","id":"train/atomic-layer-deposition/experimental-usecase/8/figure_3","sample_id":"atomic-layer-deposition/experimental-usecase/8/figure_3","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"multiple scatter plot"},{"panel_id":"c","label":"line chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The graph shows the relationship between deposition temperature (T<sub>dep</sub>) and GPC (Å/cycle). As the temperature increases, GPC decreases.\"},{\"panel_id\":\"b\",\"text\":\"The graph illustrates the impact of pulse length on GPC. Longer pulse lengths result in higher GPC values.\"},{\"panel_id\":\"c\",\"text\":\"The line chart demonstrates the increase in thickness with the number of cycles. There's a linear relationship indicating that more cycles lead to thicker layers.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| T<sub>dep</sub> (°C) | GPC (Å/cycle) |\\n|---|---|\\n| 210 | 6.8 |\\n| 220 | 6.5 |\\n| 230 | 6.0 |\\n| 240 | 5.5 |\"},{\"panel_id\":\"b\",\"text\":\"| Pulse length (s) | GPC (Å/cycle) |\\n|---|---|\\n| 1 | 5.7 |\\n| 2 | 6.5 |\\n| 3 | 7.0 |\\n| 4 | 7.2 |\\n| 5 | 7.3 |\"},{\"panel_id\":\"c\",\"text\":\"| Cycles | Thickness (nm) |\\n|---|---|\\n| 50 | 50 |\\n| 100 | 75 |\\n| 150 | 100 |\"}]","vqa":"[{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Yes. The linear relationship between thickness and cycle number, passing through the origin, confirms self-limiting, layer-by-layer growth characteristic of ideal ALD/MLD processes.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Both Eu-thd and HQA precursors reach saturation at approximately 3 seconds pulse length.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Blue points represent Eu-thd (europium tris-tetramethylheptanedionate) precursor saturation, and green points represent HQA (2-hydroxyquinoline-4-carboxylic acid) organic precursor saturation.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"This decreasing GPC trend is typical for ALD/MLD processes involving organic precursors. At higher temperatures, organic molecules may partially desorb from the surface before reacting, or undergo thermal decomposition. Additionally, increased thermal energy can enhance desorption of physisorbed species, reducing the effective surface coverage per cycle.\"}]}]","bbox":[{"panel_id":"c","x":658,"y":0,"width":341,"height":688},{"panel_id":"b","x":332,"y":0,"width":326,"height":688},{"panel_id":"a","x":0,"y":0,"width":331,"height":694}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/8/images/figure_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/8/images/figure_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/8/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/8/Adv Materials Inter - 2024 - Ghazy - Atomic and Molecular Layer Deposition of Functional Thin Films Based on Rare Earth.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"8","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":1003,"height":689,"image_format":"jpeg","image_sha256":"7576d8f8bc10778c087e6e009b4ec436f752e2a9f4e87184681e0dbcfa8e1593","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_experimental-usecase_8_figure_6.jpg","caption":"Figure 6. Growth rates of different  $\\Upsilon_{2}\\bigcirc_{3}$  processes (left) with different Y-precursors and  $\\mathsf{H}_2\\mathsf{O}$  as co-reactant, and (right) with different co-reactants and  $\\Upsilon (^{(\\mathsf{IPr2}}\\mathsf{amd})(^{\\mathsf{Et}}\\mathsf{Cp})_2$  as Y-precursor.","id":"train/atomic-layer-deposition/experimental-usecase/8/figure_6","sample_id":"atomic-layer-deposition/experimental-usecase/8/figure_6","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Comparison of GPC values across ten yttrium precursors with H₂O as co-reactant, showing Yᴱᵗ(Cp)₃ achieves the highest GPC (~1.7 Å) and Y(ᵗBu₂-Famd)₃ the lowest (~0.67 Å).\"},{\"panel_id\":\"b\",\"text\":\"Effect of five different co-reactants on GPC when using Y(ⁱPr₂amd)(ᴱᵗCp)₂ precursor, with O₃ giving the highest (~1.1 Å) and O₂ plasma the lowest (~0.83 Å).\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| GPC (Å) | Compound Name |\\n| --- | --- |\\n| 0.75 | Y(iPr2-amd)3 |\\n| 0.85 | Y(Me)(Cp)3 |\\n| 1.00 | Y(Cp)3 |\\n| 1.10 | Y(iPr2-amd)(EtCp)2 |\\n| 1.20 | Y(Bu)(Cp)3 |\\n| 1.25 | Y(DPDMG)3 |\\n| 1.35 | Y(iPr2-amd)(iPrCp)2 |\\n| 1.45 | Y(iPr2-Famd)3 |\"},{\"panel_id\":\"b\",\"text\":\"| GPC (Å) | Gas Type |\\n| --- | --- |\\n| 0.75 | O2 plasma |\\n| 0.85 | H2O plasma |\\n| 0.95 | O3 |\\n| 1.05 | O2 |\\n| 1.15 | H2O |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Ozone produces the highest GPC (~1.1 Å), followed by H₂O and H₂O plasma (~1.03–1.05 Å), then O₂ (~0.9 Å), with O₂ plasma yielding the lowest (~0.83 Å). This suggests that chemical reactivity of the oxidizer toward the precursor ligands, rather than plasma activation alone, primarily determines growth rate.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The y-axis represents growth-per-cycle (GPC) measured in angstroms (Å), indicating the film thickness deposited per ALD cycle.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Yᴱᵗ(Cp)₃ achieves the highest GPC of approximately 1.7 Å.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Key factors include: (1) steric bulk of ligand substituents affecting surface packing; (2) ligand lability influencing reaction kinetics with H₂O; (3) molecular footprint controlling maximum surface coverage per cycle.\"}]}]","bbox":[{"panel_id":"b","x":583,"y":4,"width":580,"height":396},{"panel_id":"a","x":0,"y":2,"width":574,"height":393}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/8/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/8/images/figure_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/8/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/experimental-usecase/8/Adv Materials Inter - 2024 - Ghazy - Atomic and Molecular Layer Deposition of Functional Thin Films Based on Rare Earth.pdf"],"main_category":"atomic-layer-deposition","sub_category":"experimental-usecase","paper_id":"8","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":1167,"height":400,"image_format":"jpeg","image_sha256":"b259f81d80fcc7b8644d5a9da60c8086aa7b9362ff7f505ca05a3ca3298f14bd","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_32_37a3a22fa00063f75c2dfec274e4e445732ee1bc19ab810e090c416eadd5d937.jpg","caption":"Fig. B2 - Comparison between pressure readings by capacitance manometer and Pirani gauge.","id":"train/atomic-layer-deposition/simulation-usecase/32/37a3a22fa00063f75c2dfec274e4e445732ee1bc19ab810e090c416eadd5d937","sample_id":"atomic-layer-deposition/simulation-usecase/32/37a3a22fa00063f75c2dfec274e4e445732ee1bc19ab810e090c416eadd5d937","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot compares the capacitance and Pirani guage readings with Argon flow\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Pirani Gauge (Torr) | Capacitance Manometer (Torr) |\\n|----------------------|------------------------------|\\n| 0.00                | 0.00                        |\\n| 0.02                | 0.03                        |\\n| 0.04                | 0.06                        |\\n| 0.06                | 0.09                        |\\n| 0.08                | 0.11                        |\\n| 0.10                | 0.14                        |\\n| 0.12                | 0.17                        |\\n| 0.14                | 0.20                        |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The capacitance manometer readings measure pressure consistently with the Pirani gauge with respect to the argon flow\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The model predictions for the pressure to a value(Pconv) to be compared with the Pirani gauge measurements.Although the above assumption is not validated for H2O orTMA, experimental measurements show that it can be used for the Ar flow. However, only by using  the same assumption for the reactants, as it is the only way to compare the outlet measurements with the model predictions\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"A capacitance manometer is a highly accurate electronic gauge that measures pressure (especially vacuum) by detecting changes in electrical capacitance caused by the deflection of a thin diaphragm, converting this physical movement into a precise electrical signal, making it gas-independent and ideal for lab use\"}]}]","bbox":[{"panel_id":"a","x":4,"y":4,"width":1008,"height":647}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/images/37a3a22fa00063f75c2dfec274e4e445732ee1bc19ab810e090c416eadd5d937.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/images/37a3a22fa00063f75c2dfec274e4e445732ee1bc19ab810e090c416eadd5d937.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/reactorGP Gakis et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"32","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":1017,"height":656,"image_format":"jpeg","image_sha256":"c45015685cae00e243628d083121e3ad6b2920ebc99edc94ecd7cce76e3b0ab1","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_32_fig_4.jpg","caption":"Fig. 4 - Outlet pressure vs inlet Ar flow rate: model predictions (line) and experimental measurements (dots).","id":"train/atomic-layer-deposition/simulation-usecase/32/fig_4","sample_id":"atomic-layer-deposition/simulation-usecase/32/fig_4","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart compares the experimental data and model prediction on the relationship between the inlet argon flow rate (in sccm) and the outlet pressure (in Torr). There is a linear increase in outlet pressure as the inlet argon flow rate increases.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Inlet Argon Flow Rate (sccm) | Outlet Pressure - Experiments (Torr) | Outlet Pressure - Model (Torr) |\\n|------------------------------|--------------------------------------|--------------------------------|\\n| 0                            | 0.000                               | 0.000                         |\\n| 50                           | 0.015                               | 0.012                         |\\n| 75                           | 0.022                               | 0.020                         |\\n| 100                          | 0.030                               | 0.028                         |\\n| 150                          | 0.045                               | 0.043                         |\\n| 200                          | 0.060                               | 0.058                         |\\n| 250                          | 0.075                               | 0.073                         |\\n| 300                          | 0.085                                | 0.08                         |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The outlet pressure varied linearly with respect to inlet argon flow rate. As the argon flow rate increases, outlet pressure increases.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"in order to compare the predicted values, Pcalc,with the Pirani measurements, Pexp, the former is converted using the formula , Pconv = Pcalc ( kmixture/kNitrogen),  which accounts for the dependence, assumed linear, of the pressure reading on the ratio of the thermal conductivities of the gas mixture and nitrogen\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"kmixture is the thermal conductivity of the gas mixture,kNitrogen is the thermal conductivity of nitrogen, Pcalc is the calculated pressure. Using the kinetic gas theory, the thermal conductivity of pure nitrogen at the outlet temperature, i.e270◦C, is estimated 0.041106 W/mK.\"}]}]","bbox":[{"panel_id":"a","x":2,"y":1,"width":1004,"height":560}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/images/fig_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/32/reactorGP Gakis et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"32","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":1012,"height":567,"image_format":"jpeg","image_sha256":"55e0b26b04d5b876fbcf3c24722474db7e1790970347edcd319e793d0ddc3601","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_35_fig_1.jpg","caption":"Fig. 1. Effect of  $\\mathrm{AlMe}_3$  reaction temperature on the amount of aluminum adsorbed,  $\\Delta c_{\\mathrm{Al}}$ , in the  $\\mathrm{AlMe}_3$  reaction with alumina heat-treated at  $560^{\\circ}\\mathrm{C}$  (adapted from reference [21]). Results up to reaction at  $300^{\\circ}\\mathrm{C}$  are shown; above  $300^{\\circ}\\mathrm{C}$ ,  $\\mathrm{AlMe}_3$  decomposes [21,29]. A line has been fitted to the experimental data; the confidence limits represent 1 standard deviation.","id":"train/atomic-layer-deposition/simulation-usecase/35/fig_1","sample_id":"atomic-layer-deposition/simulation-usecase/35/fig_1","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the amount of aluminum adsorbed (Δc_Al) as a function of AlMe₃ reaction temperature on alumina pretreated at 560°C. The fitted line (y = a + bx) yields a = 2.9 ± 0.1 nm⁻² and b = −0.0003 ± 0.0005, indicating that Δc_Al remains essentially constant (~2.9 nm⁻²) across the 80–300°C range, demonstrating temperature independence of aluminum adsorption in this regime.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| AlMe₃ reaction temperature [°C] | Δcₐ [nm²] |\\n|---|---|\\n| 100 | 2.9 ± 0.1 |\\n| 150 | 2.9 ± 0.1 |\\n| 200 | 2.9 ± 0.1 |\\n| 250 | 2.9 ± 0.1 |\\n| 300 | 2.9 ± 0.1 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"No. The fitted slope b = −0.0003 ± 0.0005 is statistically indistinguishable from zero, meaning Δc_Al remains constant at approximately 2.9 nm⁻² regardless of reaction temperature in this range.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Above 300°C, AlMe₃ undergoes thermal decomposition, which would introduce non-ALD growth mechanisms and invalidate the self-limiting surface reaction assumption.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The temperature independence indicates that the AlMe₃ chemisorption reaction is not kinetically limited within the 80–300°C window; rather, it proceeds to completion regardless of thermal energy input. This behavior is characteristic of self-terminating ALD reactions where surface site availability (OH groups) and steric hindrance by adsorbed methyl groups, not reaction kinetics, determine the saturation coverage. The constant Δc_Al value of ~2.9 nm⁻² therefore reflects a thermodynamically controlled endpoint defined by the fixed OH concentration on the 560°C pretreated alumina surface.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The y-axis shows Δc_Al, the surface concentration of aluminum atoms adsorbed per ALD cycle, expressed in units of nm⁻² (atoms per square nanometer).\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":403,"height":228}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/35/images/fig_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/35/images/fig_1.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/35/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/35/Riikka L. Puurunen.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"35","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":403,"height":228,"image_format":"jpeg","image_sha256":"74bfca2ba037653b9f6e6d42b2190db9052c22dba79e3fff710f7013362344b1","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-deposition_simulation-usecase_38_figure_8.jpg","caption":"Figure 8. Hacac pressure contours across the wafer in (a) R1 at  $2\\mathrm{s}$  and (b) R4 at  $3\\mathrm{s}$ . The red region in (a,b) indicates the area that is covered with the reagent, Hacac. Distribution of exposure time for all nodes in the wafer for (c) R1 and (d) R4.  $\\mu$  and  $\\sigma$  represent the average and standard deviation, respectively. There are 10,410 nodes in R1 and 6737 nodes in R4.","id":"train/atomic-layer-deposition/simulation-usecase/38/figure_8","sample_id":"atomic-layer-deposition/simulation-usecase/38/figure_8","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"},{"panel_id":"c","label":"bar chart"},{"panel_id":"d","label":"bar chart"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Hacac partial pressure contour on wafers in reactor R1 at 2 s, showing coverage regions in red and illustrating spatial distribution of the reagent.\"},{\"panel_id\":\"b\",\"text\":\"Hacac partial pressure contour on wafers in reactor R4 at 3 s, showing coverage regions in red and illustrating spatial distribution of the reagent.\"},{\"panel_id\":\"c\",\"text\":\"Bar chart of Hacac exposure times for all wafer nodes in reactor R1, showing a broader distribution with average 2.717 s and standard deviation 0.1925 s.\"},{\"panel_id\":\"d\",\"text\":\"Bar chart of Hacac exposure times for all wafer nodes in reactor R4, showing a tightly centered distribution with average 2.701 s and standard deviation 0.0079 s.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| x-coordinate|y-coordinate|\\n|-------------|------------|\\n|0.1|-0.05|\\n|0.15|-0.1|\\n|0.2|-0.15|\\n|0.25|-0.2|\\n|0.3|-0.25|\"},{\"panel_id\":\"b\",\"text\":\"| x-coordinate|y-coordinate|\\n|-------------|------------|\\n|0.05|-0.1|\\n|0.1|-0.15|\\n|0.15|-0.2|\\n|0.2|-0.25|\\n|0.25|-0.3|\"},{\"panel_id\":\"c\",\"text\":\"| Exposure time (s)|Number of nodes|\\n|------------------|---------------|\\n|2.0|0|\\n|2.1|0|\\n|2.2|0|\\n|2.3|0|\\n|2.4|0|\\n|2.5|0|\\n|2.6|0|\\n|2.7|0|\\n|2.8|0|\\n|2.9|0|\\n|3.0|0|\\n|3.1|0|\\n|3.2|0|\\n|3.3|0|\\n|3.4|0|\\n|3.5|0|\\n|3.6|0|\\n|3.7|0|\\n|3.8|0|\\n|3.9|0|\\n|4.0|0|\"},{\"panel_id\":\"d\",\"text\":\"| Exposure time (s)|Number of nodes|\\n|------------------|---------------|\\n|2.0|0|\\n|2.1|0|\\n|2.2|0|\\n|2.3|0|\\n|2.4|0|\\n|2.5|0|\\n|2.6|0|\\n|2.7|0|\\n|2.8|0|\\n|2.9|0|\\n|3.0|0|\\n|3.1|0|\\n|3.2|0|\\n|3.3|0|\\n|3.4|0|\\n|3.5|0|\\n|3.6|0|\\n|3.7|0|\\n|3.8|0|\\n|3.9|0|\\n|4.0|0|\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Reactor geometry (shape of reaction zones), Wafer rotation speed (angular velocity), Radial position of nodes on the wafer, Design of inlet and outlet configurations, Arc length of reaction zones relative to radial distance\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Yes. R1 exhibits a wider exposure time distribution with higher standard deviation, whereas R4 has a tightly centered distribution with low variability, despite similar average exposure times.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Exposure time is determined by recording the Hacac partial pressure at every node of the wafer as it passes through the reaction zone. Each node is considered exposed as long as the pressure is above a threshold, and the total time in this state defines the exposure time. The wafer rotation and geometry of the reaction zones influence how long each node remains in contact with the reagent, which is then visualized as a distribution across all nodes.\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"R1: Average exposure time = 2.717 s, Standard deviation = 0.1925 s\\nR4: Average exposure time = 2.701 s, Standard deviation = 0.0079 s\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":538,"height":388},{"panel_id":"b","x":557,"y":0,"width":532,"height":388},{"panel_id":"c","x":0,"y":481,"width":498,"height":389},{"panel_id":"d","x":557,"y":481,"width":500,"height":357}],"source":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/38/images/figure_8.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/38/images/figure_8.json","source_content":"icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/38/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-deposition/simulation-usecase/38/Sungil Yun et al.pdf"],"main_category":"atomic-layer-deposition","sub_category":"simulation-usecase","paper_id":"38","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":1089,"height":908,"image_format":"jpeg","image_sha256":"7f3c849e07f505efb0cad2eed2b8bdf45e16f04d6290851f9009f7d2161ba9ca","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_11_figure_2.jpg","caption":"Figure 2.  $\\mathrm{ZnO}$  thickness evolution as a function of the number of cycles for (a) sequence 1: Hacac and  $\\mathrm{O}_2$  gas (circles); (b) sequence 2: Hacac and  $\\mathrm{O}_2$  plasma (squares); and (c) sequence 3: only  $\\mathrm{O}_2$  plasma pulses (diamonds). Significant  $\\mathrm{ZnO}$  thickness decrease was observed only for sequence 2. All of the experiments were carried out at a temperature of  $250^{\\circ}C$","id":"train/atomic-layer-etching/experimental-usecase/11/figure_2","sample_id":"atomic-layer-etching/experimental-usecase/11/figure_2","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"},{"panel_id":"c","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The ZnO thickness remains constant at approximately 585 Å for Hacac + O₂ gas and O₂ plasma treatments, while it decreases linearly for Hacac + O₂ plasma treatment.\"},{\"panel_id\":\"b\",\"text\":\"The ZnO thickness decreases linearly with increasing number of cycles for Hacac + O₂ plasma treatment.\"},{\"panel_id\":\"c\",\"text\":\"The ZnO thickness remains constant at approximately 585 Å for O₂ plasma treatment.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of cycles | Hacac + O₂ gas |\\n|------------------|----------------|\\n| 0               | 584            | \\n| 10              | 585            | \\n| 20              | 585            | \\n| 30              | 585            | \\n| 40              | 585            |\"},{\"panel_id\":\"b\",\"text\":\"| Number of cycles | Hacac + O₂ plasma |\\n|------------------|--------------------|\\n| 0                 | 585                |\\n| 10              | 572                |\\n| 20              | 560                |\\n| 30              | 550                |\\n| 40              | 530                |\"},{\"panel_id\":\"c\",\"text\":\"| Number of cycles | O₂ plasma |\\n|------------------|-----------|\\n| 0               | 585       |\\n| 10              | 585       |\\n| 20              | 585       |\\n| 30              | 585       |\\n| 40              | 585       |\"}]","vqa":"[{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"From the figure, image b when Hacac and O2 plasma are exposed sequentially, we observe actual etching taking place as the thickness decreases.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"1. Hacac\\n, 2. O2 gas\\n, 3. O2 plasma.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Upon exposure to Hacac an apparent thickness increase of ∼2 Å was detected after the first 10 cycles, whereas the thickness decreases slightly afterward. This apparent thickness increase can be attributed to the adsorption of Hacac molecules onto the ZnO surface, similarly to what has been observed on Al2O3 substrates for the ABC-type area-selective ALD of SiO2. The subsequent thickness decrease might be due to partial decomposition of acac species on the ZnO surface . Note that the measured thickness does not decrease below the starting ZnO thickness during sequence 1.\"}]}]","bbox":[{"panel_id":"c","x":366,"y":11,"width":139,"height":355},{"panel_id":"b","x":223,"y":13,"width":133,"height":359},{"panel_id":"a","x":71,"y":11,"width":135,"height":363}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/11/images/figure_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/11/images/figure_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/11/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/11/Isotropic Atomic Layer Etching of ZnO Using Acetylacetone and O2 Plasma.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"11","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":506,"height":403,"image_format":"jpeg","image_sha256":"eef0cb05047163a5f0bb13e029928058d5b43a3889b68f281974d3c052da4083","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_11_figure_3.jpg","caption":"Figure 3. (a) EPC as a function of the Hacac exposure time for a fixed  $\\mathrm{O}_2$  plasma step of  $5\\mathrm{s}$ . Saturation (self-limiting behavior) was reached for a total Hacac exposure of  $27\\mathrm{s}$ , resulting in an EPC of  $1.31\\mathrm{\\AA / cycle}$ . (b) EPC as a function of the  $\\mathrm{O}_2$  plasma exposure time for a fixed Hacac dose of  $27\\mathrm{s}$ . A saturated EPC value was measured for all the investigated  $\\mathrm{O}_2$  plasma exposure times  $(\\geq 2\\mathrm{s})$ . The processing temperature was  $250^{\\circ}\\mathrm{C}$ . The dashed lines serve as a guide to the eye.","id":"train/atomic-layer-etching/experimental-usecase/11/figure_3","sample_id":"atomic-layer-etching/experimental-usecase/11/figure_3","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart shows the relationship between Hacac exposure time and EPC (Å/cycle). Saturation achieved after 20 s\"},{\"panel_id\":\"b\",\"text\":\"The line chart illustrates the impact of O₂ plasma exposure time on EPC (Å/cycle). Saturation achieved after 3s.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Hacac Exposure (s) | EPC (Å/cycle) |\\n|---------------------|---------------|\\n| 0                  | 0.0           |\\n| 5                  | 0.4           |\\n| 10                 | 0.8           |\\n| 15                 | 1.0           |\\n| 20                 | 1.2           |\\n| 30                 | 1.3           |\\n| 40                 | 1.4           |\"},{\"panel_id\":\"b\",\"text\":\"| O₂ Plasma Exposure (s) | EPC (Å/cycle) |\\n|-------------------------|---------------|\\n| 0                      | 0.0           |\\n| 1                      | 1.2           |\\n| 2                      | 1.3           |\\n| 4                      | 1.3           |\\n| 6                      | 1.3           |\\n| 8                      | 1.3           |\\n| 10                     | 1.3           |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"1. Hacac exposure was observed to saturate at a value of 1.31 ± 0.08 Å for a total Hacac exposure time of 27 s\\n2. O2 plasma exposure time shows saturation already after 2 s  at EPC of 1.31 Å/cycle\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"By observing the saturation curves, as the process saturates with respect to the Hacac and O2 plasma exposure, the ALE rate will remain the same irrespective of longer duration indicating the process has hit its reactive limit, therefore the images depict the self-limiting behaviour.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The processing temperature was 250 °C.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"b","x":459,"y":3,"width":365,"height":303},{"panel_id":"a","x":8,"y":6,"width":427,"height":308}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/11/images/figure_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/11/images/figure_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/11/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/11/Isotropic Atomic Layer Etching of ZnO Using Acetylacetone and O2 Plasma.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"11","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":841,"height":352,"image_format":"jpeg","image_sha256":"1d4e3ded5c3e55a171309cab58c75059e3ad8480514d147748f3615cd5cf2c45","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_13_figure_5.jpg","caption":"Figure 5. Etch depth versus number of etching cycles.","id":"train/atomic-layer-etching/experimental-usecase/13/figure_5","sample_id":"atomic-layer-etching/experimental-usecase/13/figure_5","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the relationship between the number of cycles and etch depth. As the number of cycles increases, the etch depth also increases.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of cycles | Etch depth [nm] |\\n|---|---|\\n| 0 | 0 |\\n| 5 | 3 |\\n| 7 | 5 |\\n| 16 | 11 |\\n| 24 | 17 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"3 nm.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"5 nm.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"11 nm.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"17 nm.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":6,"width":579,"height":445}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/13/images/figure_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/13/images/figure_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/13/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/13/Low-Damage and Self-Limiting (Al)GaN Etching Process.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"13","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":583,"height":450,"image_format":"jpeg","image_sha256":"e0505b85e5d8bb12c0bbeba848cabc4aaa1ff00cbda800da52f38a4676235f15","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_17_figure_4.jpg","caption":"Figure 4. Mass change per cycle (MCPC) showing the self-limiting nature of the  $\\mathrm{SF}_4$  and  $\\mathrm{Sn(acac)}_2$  reactions during  $\\mathrm{Al}_2\\mathrm{O}_3$  ALE at 200  $^\\circ \\mathrm{C}$ . (a)  $\\mathrm{SF}_4$  exposure time,  $X_{r}$  is varied as the  $\\mathrm{Sn(acac)}_2$  exposure time is held constant at  $2.5\\mathrm{s}$ . (b)  $\\mathrm{Sn(acac)}_2$  exposure time,  $X_{r}$  is varied as the  $\\mathrm{SF}_4$  exposure time is held constant at  $2.5\\mathrm{s}$ . Purge times after each reactant exposure are  $55\\mathrm{s}$ .","id":"train/atomic-layer-etching/experimental-usecase/17/figure_4","sample_id":"atomic-layer-etching/experimental-usecase/17/figure_4","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This dual-panel figure illustrates how MCPC (mass change per cycle) varies with precursor exposure time for both SF₄ and Sn(acac)₂ in the Al₂O₃ ALE process. In both cases, the MCPC decreases rapidly with increasing exposure time and asymptotically approaches a saturation point, indicating that each half-reaction becomes self-limiting beyond a certain exposure threshold. These saturation behaviors confirm that both precursors operate via self-limiting surface chemistry, a defining feature of atomic layer etching processes.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| SF₄ Exposure Time (s) | MCPC (ng/cm²) |\\n|------------------------|----------------|\\n| 0                      | 3.8            |\\n| 0.5                    | 2.1            |\\n| 1                      | 1.1            |\\n| 2                      | 0.7            |\\n| 3                      | 0.5            |\"},{\"panel_id\":\"b\",\"text\":\"| Sn(acac)₂ Exposure Time (s) | MCPC (ng/cm²) |\\n|-----------------------------|----------------|\\n| 0                           | 3.6            |\\n| 0.5                         | 2.3            |\\n| 1                           | 1.2            |\\n| 2                           | 0.6            |\\n| 3                           | 0.4            |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"SF₄.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The decreasing trends in MCPC for both SF₄ and Sn(acac)₂ exposures show that each precursor reaction is self-limiting, a hallmark of ALE chemistry. Initially, as exposure time increases, there is a sharp drop in MCPC due to rapid surface reactions. However, beyond ~2 seconds, additional exposure yields diminishing returns, indicating saturation of available reactive sites. This suggests that both the fluorination and ligand-exchange steps are well-controlled and reach completion under sufficient exposure durations, enabling precise layer-by-layer material removal.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Self-limiting reaction, efficient early-stage removal, surface saturation by ~2–3 s, stable MCPC plateau for repeatability\"}]}]","bbox":[{"panel_id":"a","x":2,"y":3,"width":669,"height":474},{"panel_id":"b","x":4,"y":484,"width":664,"height":478}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/17/images/figure_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/17/images/figure_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/17/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/17/SF4 as the Fluorination Reactant for Al2O3 and VO2.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"17","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":672,"height":964,"image_format":"jpeg","image_sha256":"7825a92902e1dc15869af58b329d4aa442aabe19b53803ece7b81fb3a2c8bb1e","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_22_e491b4d038fdd578fa1d521ec6367993f5d8e0ca7e5b91f94752170244c6b21d.jpg","caption":"","id":"train/atomic-layer-etching/experimental-usecase/22/e491b4d038fdd578fa1d521ec6367993f5d8e0ca7e5b91f94752170244c6b21d","sample_id":"atomic-layer-etching/experimental-usecase/22/e491b4d038fdd578fa1d521ec6367993f5d8e0ca7e5b91f94752170244c6b21d","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Panel (a) shows the mass loss (ng/cm²) of material as a function of MoF₆ dose duration, revealing rapid initial etching followed by saturation beyond ~1 second.\"},{\"panel_id\":\"b\",\"text\":\"Panel (b) plots mass loss against H₂O dose time, similarly showing an initial sharp increase that levels off near 4 seconds. The graphs indicate self-limiting behaviour characteristic of atomic layer etching (ALE), where further dosing does not result in additional material removal beyond a certain threshold.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| MoF₆ dose (s) | Mass Loss (ng/cm²) |\\n|---|---|\\n| 0.0 | 0.0 |\\n| 0.5 | 16.0 |\\n| 1.0 | 18.0 |\\n| 1.5 | 18.0 |\\n| 2.0 | 18.0 |\\n| 2.5 | 18.0 |\\n| 3.0 | 18.0 |\"},{\"panel_id\":\"b\",\"text\":\"| H₂O dose (s) | Mass Loss (ng/cm²) |\\n|---|---|\\n| 0.0 | 0.0 |\\n| 0.5 | 9.0 |\\n| 1.0 | 16.0 |\\n| 2.0 | 19.0 |\\n| 3.0 | 22.0 |\\n| 4.0 | 22.0 |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"MoF₆: Mass loss quickly reaches saturation around 0.5–1.0 s, H₂O: Mass loss increases over a wider dose range, saturating near 3–4 s.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"They exhibit self-limiting behaviour, indicating atomic layer etching (ALE) characteristics\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The data show that MoF₆ achieves most of its etching effect within the first second, after which further increases in dose do not increase mass loss. This suggests a surface-limited reaction where all available reactive sites are quickly consumed. In contrast, H₂O dosing results in a more gradual increase in mass loss, reaching a plateau around 4 seconds. This indicates that the removal step may depend on the availability of MoF₆-reacted surface species and the diffusion/reactivity of H₂O, making it the rate-limiting factor in the overall ALE cycle.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":507,"height":394},{"panel_id":"b","x":0,"y":398,"width":502,"height":396}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/images/e491b4d038fdd578fa1d521ec6367993f5d8e0ca7e5b91f94752170244c6b21d.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/images/e491b4d038fdd578fa1d521ec6367993f5d8e0ca7e5b91f94752170244c6b21d.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/22/Thermal Atomic Layer Etching of MoS2 Using MoF6 and H2O.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"22","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"not_found"},"width":506,"height":800,"image_format":"jpeg","image_sha256":"9c21ae02a40bfa3e8c2fd41a3898e76c4d48bc310ed6d9d83c714da666003cee","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_29_fig_2.jpg","caption":"FiG. 2. EPC of ALE for the AlGaN film as a function of Ar plasma exposure time in the removal step with a self-bias voltage of  $45~\\mathrm{V}$ .","id":"train/atomic-layer-etching/experimental-usecase/29/fig_2","sample_id":"atomic-layer-etching/experimental-usecase/29/fig_2","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the relationship between Ar plasma exposure time and etch depth per cycle. As the exposure time increases, the etch depth also increases, till it soft-saturates.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Ar Plasma Exposure Time (s) | Etch Depth per Cycle (nm) |\\n|---|---|\\n| 0 | 0 |\\n|10|0.3|\\n| 25 | 0.43 |\\n|30|0.55|\\n| 50 | 0.5 |\\n| 100 | 0.6 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The etching is probably induced by the ion influence from the plasma. If there is no plasma, there is also no etching.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"With ALE only the modified layer is selectively etched. It seems that the modification depth is roughly 0.5 nm, so there is soft-saturation at this level.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"No there is no such benefit, saturation means that there is no more significant increase.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Plasma pressure, with high pressures, there are usually less energetic ions bombarding the substrate, whereas low-pressure plasmas do show this.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":3,"width":533,"height":460}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/29/images/fig_2.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/29/images/fig_2.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/29/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/29/Atomic layer etching of AlGaN using Cl2 and Ar gas.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"29","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":536,"height":464,"image_format":"jpeg","image_sha256":"aac5e9311a01a29cb3df5c5509a3d2e65b598aac45acf77e4f48cd7b6e07f888","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_30_fig_1.jpg","caption":"Fig. 1. EPC of GaN as a function of bias voltage for a fixed Ar step time of  $5\\mathrm{s}$ .","id":"train/atomic-layer-etching/experimental-usecase/30/fig_1","sample_id":"atomic-layer-etching/experimental-usecase/30/fig_1","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart shows GaN etch per cycle (EPC) versus bias voltage for an Ar removal step time of 5 s. EPC increases at low bias, is nearly constant around 0.4 nm from about 50 to 100 V (self limited ALE window), and rises strongly above about 100 V.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Bias Voltage [V] | EPC [nm] |\\n|----------------|---------|\\n| 0              | 0.0     |\\n| 50             | 0.3     |\\n| 100            | 0.4     |\\n| 150            | 1.1     |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The x axis is bias voltage in V, and the y axis is EPC in nm.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The ALE window is about 50 to 100 V. In this region EPC is almost constant at about 0.4 nm, consistent with a self limited removal per cycle rather than steadily increasing sputter yield.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It suggests a transition toward physical sputtering, where ion energy exceeds the sputtering threshold for GaN. In that regime, EPC rises with increasing bias because removal is no longer limited mainly by the prior chlorination step.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Ion energy is insufficient to fully remove the chlorinated, modified surface layer each cycle.\\n\\nSome modified species remain, so net removal per cycle stays low until bias is increased into the self limited regime.\"}]}]","bbox":[{"panel_id":"a","x":6,"y":7,"width":598,"height":386}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/images/fig_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/images/fig_1.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/30/Atomic layer etching of GaN and AlGaN.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"30","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":603,"height":395,"image_format":"jpeg","image_sha256":"c1b3c0271e04ea25c2cafacef1320ebb00e3d0b6ff3917a3d37b1552ac799ca0","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_32_figure_3.jpg","caption":"Figure 3. InGaAs etch depth measured as a function of (a) first grid voltage (acceleration grid voltage) of the ICP ion gun for chlorine species (ions/radicals) exposure time of  $15\\mathrm{min}$  and (b) the chlorine adsorption time for the first grid voltages of  $+10$  and  $+15\\mathrm{V}$  during the chlorine adsorption step. The other conditions are the same as those in figure 2.","id":"train/atomic-layer-etching/experimental-usecase/32/figure_3","sample_id":"atomic-layer-etching/experimental-usecase/32/figure_3","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Line chart showing InGaAs etch depth as a function of acceleration grid voltage for a fixed Cl species exposure time of 15 min. Etching is negligible below 10 V and increases sharply between 15 and 30 V, indicating a threshold-like voltage dependence.\"},{\"panel_id\":\"b\",\"text\":\"Multiple line chart showing InGaAs etch depth as a function of Cl species exposure time at acceleration voltages of 15 V and 10 V. Etching proceeds more rapidly at 15 V, while the 10 V condition shows a delayed onset and substantially lower etch depth over the same time range.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Acceleration Voltage (V) | Etch Depth (Å) |\\n|---|---|\\n| 0 | 0 |\\n| 5 | 0 |\\n| 10 | 0 |\\n| 15 | ~45 |\\n| 20 | ~75 |\\n| 30 | ~150 |\"},{\"panel_id\":\"b\",\"text\":\"| Exposure Time (min) | 15V Etch Depth (Å) | 10V Etch Depth (Å) | \\n|---|---|---| \\n| 0 | 0 | 0 | \\n| 5 | 0 | 0 | \\n| 10 | 0 | 0 | \\n| 15 | ~5 | 0 | \\n| 25 | ~20 | ~5 | \\n| 35 | ~65 | ~15 |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"After ~10–15 min, 15 V increases much faster, reaching ~60–70 Å by ~35 min, while 10 V rises slowly to only ~10–20 Å by ~35 min. This indicates a strong voltage/ion-energy dependence of the etch step.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"15 V.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"15 min.\"}]}]","bbox":[{"panel_id":"b","x":6,"y":545,"width":658,"height":527},{"panel_id":"a","x":2,"y":1,"width":662,"height":526}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/32/images/figure_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/32/images/figure_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/32/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/32/Atomic layer etching of InGaAs.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"32","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":667,"height":1075,"image_format":"jpeg","image_sha256":"db23ff27671621577ed208467a5e144d1f45ce41359de9ca1d073be5ac563ede","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_32_figure_4.jpg","caption":"Figure 4. Sputter etch depth of InGaAs itself during the  $\\mathrm{Ar^{+}}$  ion gun operation as a function of (a) different first grid voltages for the  $\\mathrm{Ar^{+}}$  ion exposure time of 50 s/cycle and (b)  $\\mathrm{Ar^{+}}$  ion exposure time for the first grid voltages of  $+10$  and  $+30\\mathrm{V}$ . The sputtering was cyclic processed for 100 cycles without adsorption of chlorine during the chlorine adsorption step. The power of the ICP ion gun for the  $\\mathrm{Ar^{+}}$  ion was maintained at  $200\\mathrm{W}$  at the process chamber pressure of  $3.0\\mathrm{mTorr}$  Ar. The second grid voltage to the ICP ion gun was maintained at  $-100\\mathrm{V}$  for a directional ion beam while the third grid voltage was grounded.","id":"train/atomic-layer-etching/experimental-usecase/32/figure_4","sample_id":"atomic-layer-etching/experimental-usecase/32/figure_4","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Scatter plot of Ar⁺ sputter etch depth vs acceleration grid voltage (50 s/cycle × 100 cycles). Etching is negligible up to 10 V, becomes measurable by ~30 V, and increases sharply by 50 V, indicating a strong voltage/ion-energy dependence.\"},{\"panel_id\":\"b\",\"text\":\"Multiple scatter plot showing InGaAs etch depth as a function of Ar⁺ ion exposure time at acceleration grid voltages of 10 V and 30 V. The 30 V condition produces substantially higher etch depths over the same exposure times, indicating a strong dependence of sputter yield on ion energy.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Acceleration Grid Voltage (V) | Etch Depth (Å) |\\n|---|---|\\n| 0 | 0 |\\n| 10 | 0 |\\n| 30 | ~10 |\\n| 50 | ~60 |\"},{\"panel_id\":\"b\",\"text\":\"| Ar+ Ion Exposure Time (sec) | 30V Etch Depth (Å) | 10V Etch Depth (Å) | \\n|---|---|---| \\n| 0 | 0 | 0 | \\n| 50 | ~20 | 0 | \\n| 100 | ~95 | 0 | \\n| 150 | ~210 | ~75 |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Etch depth increases non-linearly with exposure time for both, but 30 V increases much faster (≈20 → 95 → 210 Å) while 10 V stays ~0 up to 100 s and only rises by 150 s (~75 Å). This implies sputtering is strongly ion-energy dependent; at lower voltage there’s an effective “low-yield” regime over much of the time range.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 30 V.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"50 s/cycle × 100 cycles.\"}]}]","bbox":[{"panel_id":"b","x":4,"y":524,"width":653,"height":467},{"panel_id":"a","x":0,"y":0,"width":660,"height":513}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/32/images/figure_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/32/images/figure_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/32/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/32/Atomic layer etching of InGaAs.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"32","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":661,"height":994,"image_format":"jpeg","image_sha256":"fc6eb0908efcbf69b8ce10b29aff9a348ccf44188b35b627d5c0007ac0e62b00","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_33_fig_4.jpg","caption":"FIG. 4. EPC of the TiN films as a function of the radical exposure time. The dashed line is provided as a guide to the eye only.","id":"train/atomic-layer-etching/experimental-usecase/33/fig_4","sample_id":"atomic-layer-etching/experimental-usecase/33/fig_4","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the relationship between Radical Exposure Time (in seconds) and EPC (in nm/cycle). There is a positive correlation where EPC increases as the Radical Exposure Time increases, but form as saturation is observed.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Radical Exposure Time (s) | EPC (nm/cycle) |\\n|---|---|\\n| 0 | 0 |\\n| 10 | 0.9 |\\n| 30 | 0.6|\\n| 50 | 1.2 |\\n| 60 | 2.8 |\\n| 120 | 3 |\\n| 240 | 3.3 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It seems as if the EPC is soft-saturated, meaning that additional radical exposure time after 250 s will still increase the EPC.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"No, there is mostly overlap for longer exposure times. The shorter exposure times shows less overlap.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Somewhere between 50 and 100 of radical exposure time.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"3.4 nm/cycle.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":1,"width":673,"height":534}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/33/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/33/images/fig_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/33/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/33/Atomic layer etching of titanium nitride.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"33","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":669,"height":556,"image_format":"jpeg","image_sha256":"072141a5663eb4ac2d7a9decfef3d37409acf772c6bd9a038ad28b2174472489","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_33_fig_6.jpg","caption":"FIG. 6. Etching amount of TiN films as a function of the number of ALE cycles. The radical exposure time and IR annealing time were set at 60 and  $30~\\mathrm{s}$ , respectively. The dashed line on the plot was obtained from least squares fitting to the data points.","id":"train/atomic-layer-etching/experimental-usecase/33/fig_6","sample_id":"atomic-layer-etching/experimental-usecase/33/fig_6","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows a positive linear relationship between the number of cycles and etching amount, indicating that the etching amount increases proportionally with the number of cycles.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of Cycles | Etching Amount (nm) |\\n|-----------------|--------------------|\\n| 0               | 0                  |\\n| 2               | 3                  |\\n| 4               | 10                 |\\n| 6               | 12                 |\\n| 8               | 17                 |\\n| 10              | 19                 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"There is no delay in the beginning or enhanced etching indicating that there is nbo starting effect.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The EPC can be determined by calculating the slope of the linear fit through the data points.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The EPC is 1.9 nm/cycle.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 5 cycles.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":2,"width":667,"height":598}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/33/images/fig_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/33/images/fig_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/33/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/33/Atomic layer etching of titanium nitride.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"33","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":669,"height":603,"image_format":"jpeg","image_sha256":"aec139ef67325da47105d391e8ce8f0248b4f9b6770728ad5f7a1eb8c22a9e36","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_34_fig_1.jpg","caption":"FIG. 1. Etching rate of  $\\mathrm{CF_4}$  dry etching for  $\\mathrm{Al}_2\\mathrm{O}_3$  films at various  $\\mathrm{CF_4}$  plasma pressures.","id":"train/atomic-layer-etching/experimental-usecase/34/fig_1","sample_id":"atomic-layer-etching/experimental-usecase/34/fig_1","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure presents a scatter plot showing how the etching rate of Al₂O₃ during CF₄ dry etching varies with working pressure. The data reveal a steep decline in etching rate as pressure increases from a high rate at near-zero pressure to almost negligible etching above ~1 Torr indicating strong pressure-dependent suppression of etching efficiency.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Working pressure (Torr)|Etching rate (nm/min)|\\n|------------------------|---------------------|\\n|0.0|4.5|\\n|0.2|2.8|\\n|0.3|1.1|\\n|0.6|0.3|\\n|1.0|0.5|\\n|1.4|0.2|\\n|2.0|0.2|\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Understanding the pressure dependence is essential because it allows engineers to precisely control the etching process for high selectivity and accuracy. Lower pressures provide higher etching efficiency, useful for rapid material removal, while higher pressures reduce etching, enabling self-limiting or low-damage processing. This knowledge helps optimize process windows for applications such as device patterning, surface preparation, and atomic layer etching, ensuring high-quality results and consistency across wafers.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At 0 Torr, the etching rate is approximately 4.5 nm/min, while at 2 Torr, it decreases to about 0.2 nm/min.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Reduced mean free path of reactive species, Increased collision frequency in the plasma, Lower energy of reactive ions reaching the surface, Decreased dissociation efficiency of CF₄ at higher pressure\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":666,"height":506}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/images/fig_1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/images/fig_1.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/CF4 plasma-based atomic layer etching of Al2O3.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"34","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":666,"height":506,"image_format":"jpeg","image_sha256":"a544a73ab1cc9aa0aa9082112e164a1d4746673aaaeb772605cb63f8ff8d37f5","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_34_fig_5.jpg","caption":"FIG. 5. Saturation curve of the  $\\mathrm{CF}_4$  plasma dosage.","id":"train/atomic-layer-etching/experimental-usecase/34/fig_5","sample_id":"atomic-layer-etching/experimental-usecase/34/fig_5","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure presents a saturation curve showing how the etching-per-cycle rate increases sharply during the initial seconds of CF₄ plasma treatment and then gradually plateaus as the exposure time reaches 20 seconds. This indicates a self-limiting etching behavior characteristic of atomic layer processes.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| CF4 plasma treatment (s)|Etching per cycle (nm)|\\n|-------------------------|----------------------|\\n|0|0.00|\\n|5|0.20|\\n|10|0.25|\\n|20|0.24|\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Understanding saturation helps ensure precise, self-limiting etching in ALE processes. By identifying the plasma dose at which etching stops increasing (~10 s), operators can optimize process time, reduce energy consumption, and avoid overexposure that may damage substrates. This allows tight control of etch depth and improves film uniformity and reproducibility in nanofabrication.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At 5 s, the etching is ~0.20 nm/cycle;\\nat 10 s, it rises to ~0.25 nm/cycle;\\nat 20 s, it remains approximately unchanged (~0.24 nm/cycle), showing saturation.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Higher plasma dosage activates more reactive fluorine species, Increased activation enhances etching-per-cycle at lower times, Beyond ~10 s, all reactive sites saturate, As a result, etching-per-cycle plateaus.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":675,"height":511}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/images/fig_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/images/fig_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/CF4 plasma-based atomic layer etching of Al2O3.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"34","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":675,"height":511,"image_format":"jpeg","image_sha256":"17d2c7c5e0054bc5d0f4703fbdfc08db569fb5987defb9a02e73fa2c13366d6e","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_34_fig_6.jpg","caption":"FIG. 6. Saturation curve of the DMAC dosage.","id":"train/atomic-layer-etching/experimental-usecase/34/fig_6","sample_id":"atomic-layer-etching/experimental-usecase/34/fig_6","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows etching per cycle as a function of DMAC pulse time: etching increases rapidly at short pulse times and then approaches a plateau (~0.23–0.25 nm per cycle) beyond about 1 s, indicating saturation of the DMAC dosage.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| DMAC pulse time (s)|Etching per cycle (nm)|\\n|--------------------|----------------------|\\n|0.0|0.00|\\n|0.2|0.15|\\n|0.5|0.16|\\n|1.0|0.25|\\n|2.0|0.23|\"}]","vqa":"[{\"panel_id\":\"d\",\"items\":[{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The plateau indicates that after approximately 1 second of DMAC pulsing, the etching process reaches its maximum efficiency, meaning the surface becomes fully saturated and additional pulse time does not significantly enhance etching. This helps identify the optimal processing window where the reaction is both time-efficient and stable, avoiding unnecessary overexposure of DMAC.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The etching increases quickly at short pulse times but remains nearly constant at around 0.23–0.25 nm/cycle at longer times.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]},{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Self-limiting reaction behavior, Saturation of available reactive sites, Limited additional etching even with longer DMAC exposure\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":669,"height":500}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/images/fig_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/images/fig_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/34/CF4 plasma-based atomic layer etching of Al2O3.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"34","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":669,"height":500,"image_format":"jpeg","image_sha256":"9c4ac08943d54a961cd43ca1654ab85ca51ed08f6e637d522befcd3ac7d03292","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_36_fig1.jpg","caption":"Fig.1. Etched amount of the blanket  $\\mathrm{TiO_2}$  as a function of the  $\\mathrm{CF_4}$  plasma treatment time (30 and 60 s).","id":"train/atomic-layer-etching/experimental-usecase/36/fig1","sample_id":"atomic-layer-etching/experimental-usecase/36/fig1","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart shows the relationship between plasma treatment time (s) and the etched amount (nm), indicating a linear increase.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| plasma treatment (s) | etched amount (nm) |\\n|---|---|\\n| 0 | 0 |\\n| 15 | 2.58 |\\n| 30 | 6.6 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The error bar spans nearly 1 nm or a 6.6 nm measurement. It is a non-negligible variation. This suggests that it might be difficult to precisely control the process at the angstrom level often required for device manufacturing.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It is not self-limiting. The linearity indicates this is a continuous etching process, where the amount removed is directly proportional to time. Self-limiting process would show a plateau after a certain time.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"It stays roughly constant over time.\"}]}]","bbox":[{"panel_id":"a","x":4,"y":2,"width":597,"height":510}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig1.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig1.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/Cyclic C4F8 and O2 plasma etching of TiO2.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"36","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":608,"height":514,"image_format":"jpeg","image_sha256":"591da360b646f6cdf507c6630eab0c2a38123045a8b66bfcb19c1aac97b8474a","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_36_fig_4.jpg","caption":"Fig. 4. Cyclic number dependence of the amount of blanket  $\\mathrm{TiO_2}$  etching with cyclic  $\\mathrm{C_4F_8}$  and  $\\mathrm{O_2}$  plasma.","id":"train/atomic-layer-etching/experimental-usecase/36/fig_4","sample_id":"atomic-layer-etching/experimental-usecase/36/fig_4","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The chart shows a linear relationship between the number of cycles and the amount of blanket TiO2 etched (in nm), indicating a consistent increase with each cycle.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of cycles | blanket TiO2 etched amount (nm) |\\n|---|---|\\n| 0 | 0 |\\n| 1 | 0.69 |\\n| 4 | 2.89 |\\n| 8 | 5.66 |\\n| 12 | 8.65 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Accumulated process variations.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"No\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Approx. 0.71 nm/cycle.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"It stays at approximately the same speed.\"}]}]","bbox":[{"panel_id":"a","x":1,"y":3,"width":669,"height":445}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/Cyclic C4F8 and O2 plasma etching of TiO2.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"36","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":672,"height":450,"image_format":"jpeg","image_sha256":"a2787eebb6ec5273de9e07707ac89bef8be4e24b4183edb22bb11bd4f793e670","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_36_fig_5.jpg","caption":"Fig. 5. CF polymer thickness on the blanket  $\\mathrm{TiO_2}$  film for different  $\\mathrm{C_4F_8}$  plasma treatment times.","id":"train/atomic-layer-etching/experimental-usecase/36/fig_5","sample_id":"atomic-layer-etching/experimental-usecase/36/fig_5","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart shows the relationship between the duration of C₄F₈ plasma treatment (in s) and the thickness of the CF polymer formed (in nm), indicating a linear relationship between them.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| C₄F₈ plasma treatment time (s) | Thickness of CF polymer (nm) |\\n|---|---|\\n| 0 | 0 |\\n| 10 | 3.53 |\\n| 30 | 13.11 |\\n| 60 | 26.91 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"It stays roughly constant.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"30 seconds.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The approximate deposition rate is 0.46 nm/s (based on the slope).\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"No, it is not. If the process were self-limiting, the graph would rise and then plateau, once the surface is fully covered. The plot shows a linear increase, meaning the polymer continues to grow thicker and thicker indefinitely as long as the plasma is applied.\"}]}]","bbox":[{"panel_id":"a","x":1,"y":4,"width":668,"height":445}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/images/fig_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/36/Cyclic C4F8 and O2 plasma etching of TiO2.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"36","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":672,"height":452,"image_format":"jpeg","image_sha256":"8518c9801fabbe0cdf5dba12c68b8114acc2d300ba4926bd746f7bb6fb23dfd3","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_40_fig_3.jpg","caption":"FIG. 3. (a) TiN ALE etch per cycle (EPC) vs substrate table temperature. (b) EPC vs  $\\Omega_2$  gas exposure time with in situ HF exposure time fixed at  $10s$  at  $300^{\\circ}C$  c) EPC vs in situ HF time with  $\\Omega_2$  exposure time fixed at  $2s$  at  $300^{\\circ}C$  The etch rates are observed to saturate with exposure time, demonstrating the self-limiting nature of the ALE process. The dashed lines are guides to the eye.","id":"train/atomic-layer-etching/experimental-usecase/40/fig_3","sample_id":"atomic-layer-etching/experimental-usecase/40/fig_3","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"},{"panel_id":"c","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The etch rate increases with increasing temperature.\"},{\"panel_id\":\"b\",\"text\":\"The etch rate increases with oxygen exposure time up to 2 seconds, then plateaus.\"},{\"panel_id\":\"c\",\"text\":\"The etch rate increases with in-situ HF exposure time up to 15 seconds, then plateaus\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Temperature (°C) | Etch rate (Å/cycle) |\\n|---|---|\\n| 150 | 1.10 |\\n| 175 | 1.90 |\\n| 200 | 2.45 |\\n| 250 | 3.10 |\\n| 300 | 3.30 |\\n| 350 | 3.28 |\"},{\"panel_id\":\"b\",\"text\":\"| Oxygen exposure time (s) | Etch rate (Å/cycle) |\\n|---|---|\\n| 0.0 | 0.10 |\\n| 0.5 | 1.65 |\\n| 1.0 | 2.40 |\\n| 2.0 | 2.95 |\\n| 4.0 | 2.95 |\\n| 6.0 | 2.92 |\"},{\"panel_id\":\"c\",\"text\":\"| In-situ HF exposure time (s) | Etch rate (Å/cycle) |\\n|---|---|\\n| 0.0 | -0.20 |\\n| 1.0 | 0.55 |\\n| 2.5 | 1.25 |\\n| 5.0 | 2.30 |\\n| 10.0 | 3.05 |\\n| 15.0 | 3.10 |\\n| 20.0 | 3.05 |\"}]","vqa":"[{\"panel_id\":\"c\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"2 seconds for O2 gas exposure and 10 seconds for in-situ HF exposure, both at the onset of saturation.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The saturation curves demonstrate the self-limiting nature of both half-reactions in the ALE process. In subfigure (b), the oxidation step saturates because oxygen diffusion slows as the TiO2 layer thickens. In subfigure (c), the etching step saturates because the in-situ HF selectively removes only the TiO2 layer and stops when it reaches the underlying TiN. This dual self-limiting behavior is essential for achieving precise, reproducible etch depths with angstrom-scale control in each cycle.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The etch rate increases from 1.1 Å/cycle at 150°C to 3.2 Å/cycle at 300°C due to enhanced oxygen diffusion into TiN at higher temperatures, which produces thicker TiO2 layers during each oxidation step. Above 300°C, the etch rate plateaus because the oxide thickness becomes limited by factors other than diffusion kinetics, such as the thermodynamic stability of the oxide layer or surface reaction saturation. This plateau behavior is consistent with observations in other thermal ALE processes for TiN.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Approximately 3 Å/cycle.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":42,"width":479,"height":387},{"panel_id":"b","x":484,"y":45,"width":457,"height":384},{"panel_id":"c","x":953,"y":43,"width":452,"height":382}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/40/images/fig_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/40/images/fig_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/40/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/40/Isotropic plasma-thermal atomic layer etching of.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"40","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":1408,"height":433,"image_format":"jpeg","image_sha256":"6ea1080e1582cd6113625c32aabf7de6b7ae8c9f9fba7f7837ddb25bfc1ae1c3","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_43_fig_5.jpg","caption":"Fig. 5. EPC versus temperature plot for the ALE of  $\\mathrm{Al_2O_3}$  by 30 cycles of  $\\mathrm{NF_3}$  remote plasma for  $4\\mathrm{s}$  and TMA for  $3\\mathrm{s}$ . The etching amount was determined by ellipsometry.","id":"train/atomic-layer-etching/experimental-usecase/43/fig_5","sample_id":"atomic-layer-etching/experimental-usecase/43/fig_5","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This line chart represents the Etch Per Cycle (EPC) Å/cycle for Atomic Layer Etching (ALE) of Al₂O₃ as a function of substrate temperature. The process uses cycles of NF₃ remote plasma followed by TMA exposure(3 s duration). EPC slowly increases with temperature up to a point (~250°C), then a rapid etch rate is observed and then stabilizes. This thickness loss was measured via ellipsometry.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Substrate Temperature (°C) | Etch Per Cycle (Å/cycle) |\\n|----------------------------|--------------------------|\\n| 150                        | -0.1                     |\\n| 175                        | -0.1                     |\\n| 200                        |  0.0                     |\\n| 225                        |  0.4                     |\\n| 250                        |  1.4                     |\\n| 275                        |  2.6                     |\\n| 290                        |  3.8                     |\\n| 300                        |  4.2                     |\\n| 320                        |  4.6                     |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Exposure to NF₃ remote plasma for 4 s to fluorinate the surface. \\n\\nExposure to TMA for 3 s to perform ligand exchange and remove the surface species.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The thermal ALE process is mandatory. A plasma-enhanced removal step (Ar⁺ sputter) causes physical bombardment damage, which would destroy the atomic lattice of graphene. The thermal TMA step in this process provides a chemically selective and gentle removal mechanism via volatile AlF₃ formation, enabling damage-free etching necessary for 2D materials, despite the low or moderate temperature requirement.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Higher temperatures typically increase the thermal energy available for diffusion and reaction.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"NO, At 150°C, the EPC is very low (~0.5 Å/cycle) and is on the steep part of the curve. This means tiny temperature fluctuations in the reactor could cause significant percentage variations in the etch rate, destroying the atomic-scale depth control required. For reliable control, one must operate in the temperature-independent saturation regime (near 250-300°C) where the rate is stable.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":7,"width":654,"height":652}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/43/images/fig_5.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/43/images/fig_5.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/43/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/43/Surface reaction during thermal atomic layer etching of aluminum oxide.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"43","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":656,"height":661,"image_format":"jpeg","image_sha256":"91f35c295f4360424302162140d58d683c8d58e95201dea4181c258e40a9a4bb","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_45_figure_4.jpg","caption":"Figure 4. (a) Etch rates for TiN versus  $\\mathrm{O}_3$  exposure time with constant HF exposure time of  $1.0\\mathrm{~s}$ . (b) Etch rates for TiN versus HF exposure time with constant  $\\mathrm{O}_3$  exposure time of  $3.0\\mathrm{~s}$ .","id":"train/atomic-layer-etching/experimental-usecase/45/figure_4","sample_id":"atomic-layer-etching/experimental-usecase/45/figure_4","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The etch rate increases with increasing O₃ exposure time at 250°C\"},{\"panel_id\":\"b\",\"text\":\"The etch rate increases with increasing HF exposure time at 250°C\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| O3 Exposure Time (s) | Etch Rate (Å/cycle) |\\n|---|---|\\n| 0.0 | 0.00 |\\n| 0.5 | 0.08 |\\n| 1.0 | 0.11 |\\n| 1.5 | 0.17 |\\n| 2.0 | 0.17 |\\n| 2.5 | 0.20 |\\n| 3.0 | 0.19 |\\n| 4.0 | 0.22 |\"},{\"panel_id\":\"b\",\"text\":\"| HF Exposure Time (s) | Etch Rate (Å/cycle) |\\n|---|---|\\n| 0.0 | 0.00 |\\n| 0.5 | 0.14 |\\n| 0.75 | 0.18 |\\n| 1.0 | 0.19 |\\n| 1.25 | 0.20 |\\n| 1.5 | 0.21 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The saturation behavior confirms that both half-reactions are self-limiting. The O₃ reaction self-terminates because the TiO₂ layer acts as a diffusion barrier preventing further oxidation. The HF reaction self-terminates because it can only remove the finite TiO₂ layer formed during oxidation—HF does not etch TiN directly.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Neither reactant alone etches TiN-O₃ only oxidizes the surface, and HF cannot etch TiN directly. Both half-reactions are required for etching.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Self-limiting behavior ensures that etch depth depends only on cycle count, not on precise exposure timing or local reactant concentration variations. This enables atomic-level thickness control and uniform etching across large wafers, which is essential for advanced device fabrication.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"3 s for O₃ and 1 s for HF, yielding the optimized sequence 3-30-1-30 (with 30 s N₂ purges).\"}]}]","bbox":[{"panel_id":"a","x":3,"y":4,"width":667,"height":472},{"panel_id":"b","x":1,"y":493,"width":667,"height":477}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/45/images/figure_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/45/images/figure_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/45/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/45/Thermal Atomic Layer Etching of Titanium Nitride.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"45","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":675,"height":969,"image_format":"jpeg","image_sha256":"c48c3689e5e24b18e8fc6c836ce3e55ff34230ef465f3b4d25ea58fe01dfb9c6","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_46_figure_13.jpg","caption":"Figure 13. Etch rates at  $170^{\\circ}\\mathrm{C}$  for  $\\mathrm{TiO_2}$ : (a)  $\\mathrm{BCl_3}$  with a constant  $\\mathrm{WF_6}$  dose time of  $0.2\\mathrm{s}$  and (b)  $\\mathrm{WF_6}$  dose time with a constant  $\\mathrm{BCl_3}$  exposure time of  $1\\mathrm{s}$ . The ALE reactor was conditioned with  $200\\mathrm{TiO_2}$  ALD cycles after every 3 subsequent ALE runs.","id":"train/atomic-layer-etching/experimental-usecase/46/figure_13","sample_id":"atomic-layer-etching/experimental-usecase/46/figure_13","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure characterizes the saturation behavior of a thermal Atomic Layer Etching (ALE) process for Titanium Dioxide , Etch rate as a function of BCl₃ dose time at 170 °C, showing clear saturation and self-limiting behavior characteristic of thermal ALE.\"},{\"panel_id\":\"b\",\"text\":\"Etch rate as a function of WF₆ dose time at 170 °C, also exhibiting saturation, confirming that the TiO₂ etching process follows an ALE mechanism rather than continuous etching.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| BCl₃ Dose Time (s) | Etch Rate (Å/cycle) |\\n|------------------:|-------------------:|\\n| 0.0 | 0.00 |\\n| 0.1 | 0.44 |\\n| 0.5 | 0.65 |\\n| 1.0 | 0.68 |\\n| 1.5 | 0.72 |\"},{\"panel_id\":\"b\",\"text\":\"| WF₆ Dose Time (s) | Etch Rate (Å/cycle) |\\n|-----------------:|-------------------:|\\n| 0.0 | 0.10 |\\n| 0.1 | 0.50 |\\n| 0.2 | 0.68 |\\n| 0.4 | 0.60 |\\n| 0.5 | 0.64 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"1 .   WF₆ pulse: 0.2 s\\n\\n, 2.   N₂ purge: 45 s\\n\\n, 3.   BCl₃ pulse: 1.0 s\\n\\n, 4.   N₂ purge: 45 s\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"At 0.1 s, the process operates in the unsaturated regime (the steep slope of the curve), where the etch rate is highly sensitive to small fluctuations in gas flow or pressure. In contrast, 1.0 s is in the saturated regime (plateau), where the rate is stable. Operating at 0.1 s risks severe non-uniformity, especially in high-aspect-ratio structures where gas diffusion is slow.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Chamber Memory Effect. Without conditioning, the reactant gases (WF₆ andBCl₃) would partially react with or adsorb onto the chamber walls instead of the sample, effectively lowering the partial pressure at the wafer surface. This would likely cause the saturation curves to shift to the right (requiring longer doses) or show artificially lower etch rates until the walls eventually became seasoned.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Surface Site Saturation. The surface is fully fluorinated (or modified) by the WF₆; once all available surface bonds have reacted, no further chemical modification can occur regardless of excess exposure, limiting the material removed in the subsequent step.\"}]}]","bbox":[{"panel_id":"a","x":2,"y":0,"width":465,"height":367},{"panel_id":"b","x":458,"y":2,"width":455,"height":366}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/46/images/figure_13.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/46/images/figure_13.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/46/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/46/Thermal Selective Vapor Etching of TiO2.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"46","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":914,"height":369,"image_format":"jpeg","image_sha256":"0b7388362a627306225c52e0dd3c30090d1834a753ffa788bd0369b67b6fce9f","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_experimental-usecase_46_figure_6.jpg","caption":"Figure 6. (a) Thickness and (b) refractive index at  $632.8~\\mathrm{nm}$  of samples with  $5.5\\mathrm{nm}$ $\\mathrm{TiO_2}$  followed by 0, 15, 25, 50, and  $100\\mathrm{WF}_6$  doses at  $220^{\\circ}\\mathrm{C}$ , measured with spectroscopic ellipsometry. Lines are a guide to the eye.","id":"train/atomic-layer-etching/experimental-usecase/46/figure_6","sample_id":"atomic-layer-etching/experimental-usecase/46/figure_6","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This line chart represents the change in film thickness with increasing WF₆ doses at 220 °C, showing a linear decrease that indicates progressive etching or consumption of the 5.5 nm TiO₂ film.\"},{\"panel_id\":\"b\",\"text\":\"This line chart represents the change in refractive index (at 632.8 nm) with increasing WF₆ doses, showing a sharp initial drop (0–25 doses) followed by stabilization at ~1.85, indicating initial chemical modification before steady-state etching.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of WF₆ Doses | Thickness (Å) |\\n|--------------------:|--------------:|\\n| 0   | 56.00 |\\n| 10  | 58.64 |\\n| 25  | 53.11 |\\n| 50  | 50.34 |\\n| 100 | 33.71 |\"},{\"panel_id\":\"b\",\"text\":\"| Number of WF₆ Doses | Refractive Index @ 632.8 nm |\\n|--------------------:|----------------------------:|\\n| 0   | 2.31 |\\n| 10  | 2.29 |\\n| 25  | 1.88 |\\n| 50  | 1.83 |\\n| 100 | 1.90 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Deposition: Deposition of a 5.5 nm TiO₂ starting layer.\\n\\nTreatment: Exposure to a specific number of  WF₆ doses (0, 15, 25, 50, or 100) at 220°C.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Yes. The stepwise, dose-dependent index change allows spatial patterning of the refractive index by controlling the number of WF₆ doses across different regions of the waveguide. This enables the fabrication of gradient-index profiles for improved light confinement and mode shaping.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"WF₆ fluorination is superior for independent control. It changes the index by altering composition while also thinning the film. Simply depositing a thinner film only changes thickness, not the intrinsic index. Fluorination offers two tuning knobs (dose for index, cycles for thickness).\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"WF₆ fluorinates the TiO₂, converting it into a titanium oxyfluoride (TiOFₓ) or fluoride. Fluorine incorporation increases the film's band gap and lowers its density/polarizability, both of which directly cause a decrease in the refractive index.\"}]}]","bbox":[{"panel_id":"a","x":5,"y":2,"width":465,"height":385},{"panel_id":"b","x":514,"y":1,"width":469,"height":385}],"source":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/46/images/figure_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/46/images/figure_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/46/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/experimental-usecase/46/Thermal Selective Vapor Etching of TiO2.pdf"],"main_category":"atomic-layer-etching","sub_category":"experimental-usecase","paper_id":"46","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":986,"height":389,"image_format":"jpeg","image_sha256":"32a4fa18ba9b0b54af0dd0de1536eba71f513996e938fc613e4df6ab900f413a","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_15_fig_7.jpg","caption":"FIG. 7. Etch per cycle as a function of  $\\mathrm{Cl}_2$  impacts from MD simulations.  $1000\\mathrm{Ar}^+$  ion impacts are used for the ion bombardment step in every case. The energy of the  $\\mathrm{Ar}^+$  ions is  $70\\mathrm{eV}$ .","id":"train/atomic-layer-etching/simulation-usecase/15/fig_7","sample_id":"atomic-layer-etching/simulation-usecase/15/fig_7","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot shows the relationship between the number of Cl₂ impacts and etch per cycle.  The EPC initially increases rapidly with chlorine dose due to incomplete surface coverage, then saturates beyond ~1000 Cl₂ i impacts at ~0.9 Å/cycle. This plateau demonstrates the self-limiting behavior of ideal atomic layer etching, where additional chlorine does not increase material removal once the surface is fully chlorinated.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Number of Cl₂ Impacts | Etch per Cycle (Å/cycle) |\\n|----------------------|--------------------------:|\\n| 0                    | 14.52                     |\\n| 110                  | 0.53                      |\\n| 250                  | 0.68                      |\\n| 500                  | 0.84                      |\\n| 1000                 | 0.85                      |\\n| 2265                 | 0.89                      |\\n| 3000                 | 0.92                      |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Because 250 impacts is in the non-saturated region (steep slope). Small fluctuations in gas flow or pressure at this point would cause large changes in the etch rate. At 1500 impacts, the process is on the plateau (saturated), so it is robust against small variations.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Speed vs. Precision. The ALE process is extremely slow (only 0.9 Å per full cycle) compared to continuous etching. The trade-off is accepting low throughput to gain atomic-level precision and control.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Site Saturation. At 500 impacts, active surface sites are still available. By 2500 impacts, every available surface bond is already occupied by a Chlorine atom (saturated). Since no new Chlorine can stick, the subsequent ion etch cannot remove any additional material, leading to the plateau.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Spontaneous Etching. It would indicate that the excess Chlorine is beginning to etch the silicon chemically on its own (without ions), destroying the self-limiting behavior that defines Atomic Layer Etching.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":8,"width":675,"height":469}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/15/images/fig_7.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/15/images/fig_7.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/15/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/15/Molecular dynamics study of silicon atomic layer etching by chlorine gas and argon ions.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"15","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":675,"height":477,"image_format":"jpeg","image_sha256":"40acd0fa0dde7cbb89aeec51231ce66adea79a6f94b15d28ba93a3a96129aa33","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_16_fig_11.jpg","caption":"Fig. 11. (a) The effect of the substrate velocity on the etching per cycle. (b) The effect of the moving substrate on etching per cycle to illustrate the uniformity of the etch.","id":"train/atomic-layer-etching/simulation-usecase/16/fig_11","sample_id":"atomic-layer-etching/simulation-usecase/16/fig_11","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"multiple scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"A scatter plot showing the relationship between velocity (mm/s) and etching per cycle (Å/cycle). The etching decreases as the velocity increases.\"},{\"panel_id\":\"b\",\"text\":\"A scatter plot showing the etching per cycle (Å/cycle) at different substrate positions (mm) for velocities of 80 mm/s, 120 mm/s, and 160 mm/s. The etching remains relatively constant across the substrate positions for each velocity.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Velocity (mm/s) | Etching per cycle (Å/cycle) |\\n|---|---|\\n| 0 | 0.47 |\\n| 50 | 0.42 |\\n| 100 | 0.26 |\\n| 150 | 0.15 |\"},{\"panel_id\":\"b\",\"text\":\"| Substrate position (mm) | Etching per cycle (Å/cycle) 80 |Etching per cycle (Å/cycle) 120 |Etching per cycle (Å/cycle) 160 |\\n|---|---|---|---|\\n| 0 | 0.36 | 0.22 | 0.1 |\\n| 0.1 | 0.36 | 0.22 | 0.1 |\\n| 0.2 | 0.36 | 0.22 | 0.1 |\\n| 0.3 | 0.36 | 0.22 | 0.1 |\\n| 0.4 | 0.36 | 0.22 | 0.1 |\\n| 0.5 | 0.36 | 0.22 | 0.1 |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"With increasing velocity, the exposure time is decreasing. Usually long exposure times result in better uniformity.\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The parameter that is changed is the dosing time or exposure time. The exposure time scales with 1/velocity.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The saturated EPC is 0.47 A/cycle, this is reached with a velocity below 40 mm/s.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"0.36 mm/s\"}]}]","bbox":[{"panel_id":"b","x":3,"y":587,"width":658,"height":535},{"panel_id":"a","x":0,"y":0,"width":666,"height":522}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/16/images/fig_11.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/16/images/fig_11.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/16/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/16/Multiscale computational fluid dynamics modeling.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"16","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":667,"height":1178,"image_format":"jpeg","image_sha256":"d2c76494d148bce90e51cf13e87663c814534ce0cb6d252e1dbdbb2f542ef7dd","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_20_figure_10.jpg","caption":"Figure 10. Etch rate as measured by change in integrated absorbance of the  $\\mathrm{Ti - O}$  vibrational modes from 400 to  $960~\\mathrm{cm^{-2}}$  for each HF exposure. Temperature is varied from 80 to  $300^{\\circ}C$","id":"train/atomic-layer-etching/simulation-usecase/20/figure_10","sample_id":"atomic-layer-etching/simulation-usecase/20/figure_10","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This line chart  plots the etch rate (measured by the loss of integrated Ti-O IR absorbance per HF exposure) against temperature (80-300°C) for a titanium-based film, showing that the etch rate increases with temperature.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Temperature (°C) | Δ Int. Abs. Loss / HF Exposure |\\n|-----------------:|--------------------------------:|\\n| 80                | 0.00                            |\\n| 100              | 0.02                            |\\n| 120              | 0.05                            |\\n| 140              | 0.07                            |\\n| 160              | 0.25                            |\\n| 180              | 1.00                            |\\n| 200              | 1.10                            |\\n| 220              | 1.30                            |\\n| 240              | 1.90                            |\\n| 260              | 2.10                            |\\n| 280              | 2.50                            |\\n| 300              | 3.80                            |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The cut-off temperature is approximately 150°C. Below this point, the curve is effectively flat (close to zero on the Y-axis), indicating that the reaction kinetics are insufficient to remove material.Minimum Reactor Setting: To ensure the film is successfully etched, the user must set the reactor to a minimum of 150°C (or ideally higher, such as 175°C, for a practical rate). Any setting below 150°C results in negligible absorbance loss, meaning the process essentially fails to etch the film.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The low-temperature region is flat (low sensitivity), while the high-temperature region is steep (high sensitivity).At 300 °C, the etch rate is high, enabling fast throughput, but the process is highly sensitive to temperature variations; a ±5 °C deviation causes large fluctuations in etch rate. At 100 °C, the etch rate is stable but negligible. Thus, while intermediate temperatures may offer a trade-off, operation at 300 °C requires stringent thermal control.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Since infrared absorbance follows the Beer–Lambert law and is proportional to the number of bonds, a fourfold increase in absorbance loss indicates that approximately four times more TiO2  is removed per HF exposure at 300 °C than at 175 °C. This strong temperature dependence underscores the need for precise thermal control to ensure a stable etch rate.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"At 250 °C, the etch rate shows strong temperature dependence; a 10 °C gradient (250 °C at the top vs. 240 °C at the bottom) would therefore cause slower etching at the bottom, leading to a tapered (V-shaped) profile. Achieving vertical sidewalls would require operation in a temperature-insensitive (saturated) regime, which is not observed here.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":3,"width":647,"height":651}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/20/images/figure_10.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/20/images/figure_10.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/20/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/20/Prediction and Validation of the Process Window for Atomic Layer Etching HF Exposure on TiO2.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"20","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":650,"height":658,"image_format":"jpeg","image_sha256":"402a0c3918af6a9ffc87775a8b91679c4b76868aa82a49425ef568ce55cc23bc","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_20_figure_11.jpg","caption":"Figure 11. Arrhenius plot obtained from etch rates versus temperature presented in Figure 10. Slope of the plot yields an activation barrier of  $E_{\\mathrm{a}} = 42\\mathrm{kJ / mol}$ .","id":"train/atomic-layer-etching/simulation-usecase/20/figure_11","sample_id":"atomic-layer-etching/simulation-usecase/20/figure_11","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"This figure is an Arrhenius plot (logarithm of etch rate vs. inverse temperature) derived from the data in Figure 10. The linear fit (a dashed line) yields an activation energy (Eₐ) of 42 kJ/mol for the HF etching reaction of the Ti-based film.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| 1000/T (K⁻¹) | ln[Δ Int. Abs. Loss / HF Exposure] |\\n|-------------:|------------------------------------:|\\n| 1.75         | 2.65                               |\\n| 1.80         | 2.50                               |\\n| 1.85         | 2.45                               |\\n| 2.00         | 2.05                               |\\n| 2.20         | 2.00                               |\\n| 2.35         | 1.30                               |\\n| 2.45         | 0.70                               |\\n| 2.55         | 0.75                               |\\n| 2.65         | 0.10                               |\\n| 2.80         | 0.25                               |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The key step is taking the natural logarithm of the etch rate (ln(k)) and plotting it against the inverse absolute temperature (1/T). This transformation linearizes the Arrhenius equation, allowing the activation energy (Eₐ) to be determined directly from the slope of the best-fit line (-Eₐ/R).\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"Thermal HF Wet Etch: Pro: Selective, low damage. Con: Isotropic (etches in all directions).\\n\\nPlasma-Enhanced Dry Etch: Pro: Can be highly anisotropic via ion bombardment.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"An Eₐ of 42 kJ/mol (a relatively low value, ~0.44 eV) indicates the rate-limiting step is a surface chemical reaction with a moderate thermal barrier. It is not diffusion-limited (which would be lower, ~10-20 kJ/mol) nor a very high-energy bond-breaking process, suggesting a specific, activated chemisorption or ligand-exchange step controls the etch rate.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It enables predictable process tuning and scaling. Engineers can precisely calculate the temperature change needed to achieve a target etch rate (e.g., to meet throughput specs) or to maintain a constant rate despite equipment variations, ensuring reproducible results during high-volume manufacturing.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":5,"width":646,"height":644}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/20/images/figure_11.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/20/images/figure_11.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/20/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/20/Prediction and Validation of the Process Window for Atomic Layer Etching HF Exposure on TiO2.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"20","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":652,"height":655,"image_format":"jpeg","image_sha256":"4b899a56c6a03b4afe99259e1d74518cca115b4ce8c8e06437c224426399c4e5","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_27_fig_6.jpg","caption":"FIG. 6. Synergy plot for GaN using  $\\mathrm{SF}_6$  plasma and TMA at  $300^{\\circ}\\mathrm{C}$ . The first 25 pulses only involve dosing  $\\mathrm{SF}_6$  plasma, the next 25 pulses only TMA, and the last 75 cycles are full ALE cycles with both  $\\mathrm{SF}_6$  plasma and TMA dosed. The inset shows the transition between only  $\\mathrm{SF}_6$  pulses and only TMA pulses.","id":"train/atomic-layer-etching/simulation-usecase/27/fig_6","sample_id":"atomic-layer-etching/simulation-usecase/27/fig_6","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"Synergy plot demonstrating self-limiting behavior of the GaN ALE process at 300°C. During SF6 plasma-only pulses (0-25), the film thickness remains constant after an initial ~0.7 nm drop. During TMA-only pulses (25-50), thickness is also constant. Full ALE cycles (50-125) show linear etching with EPC of 0.41 nm/cycle initially (oxidized layer) decreasing to 0.31 nm/cycle (bulk GaN).\"},{\"panel_id\":\"b\",\"text\":\"Magnified view of the transition from SF6-only to TMA-only pulses, showing a 0.33 nm thickness decrease when TMA is first introduced, indicating that etching occurs during the TMA step.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Pulses/Cycles | Film Thickness (nm) | Phase / Rate |\\n|---|---|---|\\n| 0 | 43.5 | Half-cycle A |\\n| 10 | 43.4 | Half-cycle A |\\n| 20 | 43.3 | Half-cycle A |\\n| 25 | 43.2 | Half-cycle A (End) |\\n| 26 | 42.9 | Half-cycle B (Start, -0.33 nm drop) |\\n| 40 | 42.7 | Half-cycle B |\\n| 50 | 42.5 | Half-cycle B (End) |\\n| 60 | 38.5 | ALE (~0.41 nm/cycle) |\\n| 70 | 34.5 | ALE |\\n| 80 | 31.5 | ALE (~0.31 nm/cycle) |\\n| 90 | 28.5 | ALE |\\n| 100 | 25.5 | ALE |\\n| 110 | 22.5 | ALE |\\n| 120 | 19.5 | ALE |\\n| 125 | 18.0 | ALE (End) |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"The higher initial EPC corresponds to etching through the oxidized surface layer, while the lower EPC reflects etching of the more stoichiometric bulk GaN underneath.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"100%.\"}]},{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The synergy plot shows that neither SF6 plasma alone nor TMA alone causes continuous etching. The 0.33 nm thickness drop when switching from SF6-only to TMA-only pulses indicates that etching occurs during the TMA ligand-exchange step, which removes the fluorinated surface layer. Since no continuous etching is observed for either half-cycle individually, the process exhibits 100% synergy.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Removal of adventitious carbon, the heavily oxidized GaN surface layer, or changes in optical properties due to surface fluorination.\"}]}]","bbox":[{"panel_id":"a","x":2,"y":2,"width":669,"height":507},{"panel_id":"b","x":2,"y":1,"width":309,"height":465}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/27/images/fig_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/27/images/fig_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/27/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/27/Isotropic atomic layer etching of GaN using SF6 plasma and Al(CH3)3.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"27","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":675,"height":511,"image_format":"jpeg","image_sha256":"70cbbc1d040332c998b1f83d703cf4869dcce8e0f3b3e92ee8e499447d0d13be","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_27_fig_9.jpg","caption":"FIG. 9. RMS roughness as a function of etched thickness, plotted alongside the fitted change in RMS roughness as obtained using the curvature-dependent model from Ref. 85.","id":"train/atomic-layer-etching/simulation-usecase/27/fig_9","sample_id":"atomic-layer-etching/simulation-usecase/27/fig_9","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"RMS roughness evolution during GaN ALE at 300°C, comparing experimental data with a curvature-dependent smoothing model. Roughness decreases from ~2.6 nm to ~1.9 nm after 25 nm of etching. The rate of smoothing slows with increasing etch depth, in agreement with the model using a fitted diffusion parameter ε = 0.05×10⁻⁹ m.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Etched thickness (nm) | RMS roughness (nm) |\\n|---|---|\\n| 0 | 2.8 |\\n| 5 | 2.4 |\\n| 10 | 2.2 |\\n| 15 | 2.0 |\\n| 20 | 1.9 |\\n| 25 | 1.8 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The rate of smoothing decreases as the etched thickness increases. Initially, roughness drops rapidly from ~2.6 nm to ~2.2 nm within the first 10 nm of etching. Beyond this, the smoothing rate slows and begins to plateau. This behavior is captured by the curvature-dependent model, where high-curvature surface features etch faster initially, but as the surface becomes smoother, further improvement becomes more gradual.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"It represents a measure of the fluorine diffusion rate into the GaN surface during the modification step.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"GaN has a lower ε (0.05×10⁻⁹ m) compared to amorphous Al2O3 (1.50×10⁻⁹ m), indicating GaN is more resistant to fluorination.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Approximately 1.9 nm.\"}]}]","bbox":[{"panel_id":"a","x":2,"y":4,"width":664,"height":527}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/27/images/fig_9.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/27/images/fig_9.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/27/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/27/Isotropic atomic layer etching of GaN using SF6 plasma and Al(CH3)3.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"27","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":669,"height":536,"image_format":"jpeg","image_sha256":"94b5c28753734707a4c68352c23fa9037d890ef43a18001e349c054bcb95ca03","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_33_54aef25b16ec4f51b16e9d84a9a930c89371e345050a8d628002da041a377e5d.jpg","caption":"","id":"train/atomic-layer-etching/simulation-usecase/33/54aef25b16ec4f51b16e9d84a9a930c89371e345050a8d628002da041a377e5d","sample_id":"atomic-layer-etching/simulation-usecase/33/54aef25b16ec4f51b16e9d84a9a930c89371e345050a8d628002da041a377e5d","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot displays the variation in the nitrogen-to-silicon atomic ratio (N_N/N_Si) as a function of HF dose, expressed in HF/nm². Initially, the ratio begins around 1.3, then decreases to a minimum near 1.05 at approximately 80–100 HF/nm², before gradually increasing again toward the initial value. A fitted curve highlights this non-linear trend, suggesting that HF exposure initially disrupts nitrogen content before stabilizing or reversing the effect at higher doses. A horizontal dashed line serves as a reference level, possibly corresponding to the stoichiometric or ideal N_N/N_Si ratio prior to treatment.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Dose (HF/nm²) | N_N/N_Si |\\n|---------------|----------|\\n| 0             | 1.3      |\\n| 20            | 1.2      |\\n| 40            | 1.15     |\\n| 60            | 1.10     |\\n| 80            | 1.05     |\\n| 100           | 1.07     |\\n| 120           | 1.12     |\\n| 140           | 1.18     |\\n| 160           | 1.22     |\\n| 180           | 1.26     |\\n| 200           | 1.3      |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"As the HF dose increases from 0 to around 100 HF/nm², the N_N/N_Si ratio decreases, reaching a minimum, indicating a loss or displacement of nitrogen relative to silicon. Beyond this point, further HF exposure leads to a recovery in the N_N/N_Si ratio, suggesting re-equilibration or re-incorporation of nitrogen species. The resulting U-shaped trend suggests dose-dependent etching or modification behaviour.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 80–100 HF/nm².\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Initial HF doses may deplete nitrogen content, a threshold dose exists where nitrogen loss peaks, higher doses may help restore or stabilise nitrogen levels\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":275,"height":222}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/33/images/54aef25b16ec4f51b16e9d84a9a930c89371e345050a8d628002da041a377e5d.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/33/images/54aef25b16ec4f51b16e9d84a9a930c89371e345050a8d628002da041a377e5d.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/33/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/33/Atomistic Simulation of HF Etching Process of Amorphous.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"33","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"not_found"},"width":275,"height":222,"image_format":"jpeg","image_sha256":"981a8017008e530c28b052623f068b3e2ad13dbadd19db601ad3485e378e236b","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_33_8a80cb301b636349b065c782a8596739ca56704250fa04b4481f225791f3f0a3.jpg","caption":"(b)","id":"train/atomic-layer-etching/simulation-usecase/33/8a80cb301b636349b065c782a8596739ca56704250fa04b4481f225791f3f0a3","sample_id":"atomic-layer-etching/simulation-usecase/33/8a80cb301b636349b065c782a8596739ca56704250fa04b4481f225791f3f0a3","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The scatter plot displays the etching yield of silicon nitride (Si₃N₄) as a function of the electronic input energy (Eₙ) during HF-based processing. As the input energy increases from 0 to 50 eV, the etching yield steadily rises in a nonlinear fashion. The data points are fit with a dashed curve, showing an accelerating increase, suggesting that higher Eₙ values significantly enhance the reaction efficiency. The inclusion of error bars indicates variability in the measurements, particularly at higher energies.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| $E_{\\\\mathrm{in}}$ (eV) | Etching yield ($\\\\mathrm{Si}_3\\\\mathrm{N}_4$/HF) |\\n|---|---|\\n| 0.00 | 0.00 |\\n| 10.00 | 0.01 |\\n| 20.00 | 0.04 |\\n| 30.00 | 0.06 |\\n| 40.00 | 0.07 |\\n| 50.00 | 0.08 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"As the input energy increases from 0 to 50 eV, the etching yield of Si₃N₄ increases in a nonlinear manner. At low energies, the increase in yield is modest, but the trend becomes steeper at higher energies, indicating enhanced reaction kinetics or more efficient bond breaking with increasing Eₙ. This suggests that a threshold energy is required to initiate etching effectively, and further increases in Eₙ significantly boost the reaction rate.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 40 eV.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Higher Eₙ values result in greater etching yields, threshold below which etching is inefficient, Nonlinear increases suggest diminishing returns at very high energies.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":309,"height":244}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/33/images/8a80cb301b636349b065c782a8596739ca56704250fa04b4481f225791f3f0a3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/33/images/8a80cb301b636349b065c782a8596739ca56704250fa04b4481f225791f3f0a3.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/33/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/33/Atomistic Simulation of HF Etching Process of Amorphous.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"33","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":309,"height":244,"image_format":"jpeg","image_sha256":"6e30bd4f4ee4c3cce541f0d5d2c6cd9498872b26a42c879d6a6c810febe16af7","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_33_figure_4.jpg","caption":"Figure 4. Comparison of reaction energies during etching MD calculated using DFT  $(E_{\\mathrm{rxn}}^{\\mathrm{DFT}})$  and NNP  $(E_{\\mathrm{rxn}}^{\\mathrm{NNP}})$ . Each point refers to individual reactions. The density of data points on the color bar indicates the local density of reaction events in the vicinity of each point, accounting for the superposition of many data points.","id":"train/atomic-layer-etching/simulation-usecase/33/figure_4","sample_id":"atomic-layer-etching/simulation-usecase/33/figure_4","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure presents a parity plot comparing reaction energies computed via Density Functional Theory (DFT) and a Neural Network Potential (NNP) model. Each point represents a single reaction’s predicted energy from both methods, with the color intensity indicating the density of data points at that value. The data clusters tightly along the diagonal parity line (dashed), suggesting strong agreement between DFT and NNP predictions across a broad energy range (from -5 eV to 10 eV). This indicates that the NNP model is well-trained and capable of accurately reproducing DFT-level energetics for the evaluated reactions.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| $E_{\\\\mathrm{rxn}}^{\\\\mathrm{DFT}}$ (eV) | $E_{\\\\mathrm{rxn}}^{\\\\mathrm{NNP}}$ (eV) |\\n|---------------------------------------|----------------------------------------|\\n| -5                                    | -5                                     |\\n| -3                                    | -2.9                                   |\\n| 0                                     | 0.1                                    |\\n| 2                                     | 2.1                                    |\\n| 5                                     | 5.2                                    |\\n| 8                                     | 8.3                                    |\\n| 10                                    | 10                                     |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The plot indicates a strong linear correlation between the reaction energies predicted by the neural network potential (NNP) and those calculated via density functional theory (DFT). Most points cluster along the diagonal, which represents perfect agreement. This suggests that the NNP model closely replicates DFT-calculated values across a wide range of reaction energies, validating its reliability as a surrogate model for high-throughput or large-scale simulations.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 0 eV.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"It can replace DFT in large-scale simulations with high accuracy, It provides significant computational speed-up while preserving fidelity.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":0,"width":569,"height":439}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/33/images/figure_4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/33/images/figure_4.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/33/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/33/Atomistic Simulation of HF Etching Process of Amorphous.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"33","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":569,"height":439,"image_format":"jpeg","image_sha256":"3d803554aca5d724f4ea612dcd65278f854e5268636ec86a06d4d71956bae100","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_35_FIG4.jpg","caption":"Figure 4. Scatter plot showing the adsorption energy versus oxygen coverage of the minima shown in Figure 3. All of the points were fit to a parabola for eye-guiding purposes. The highlighted points are cross-referenced to the configurations shown in Figure 3. Brown, white, and red spheres represent carbon, hydrogen, and oxygen atoms, respectively.","id":"train/atomic-layer-etching/simulation-usecase/35/FIG4","sample_id":"atomic-layer-etching/simulation-usecase/35/FIG4","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows adsorption energy as oxygen coverage increases. At low coverage, adsorption is strongly favorable. Stable species such as H₂O are formed. At higher coverage, adsorption becomes less stable. New bonds form and the polymer structure begins to break.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Coverage [O/nm²] | Adsorption Energy [eV/nm²] | Annotated chemical event |\\n|------------------|----------------------------|--------------------------|\\n| 0.5              | −2.02 (global minimum)     | Initial adsorption |\\n| 3.0              | −1.00 (local minimum)      | H₂O formation (O₅) |\\n| 6.8              | +0.10 (local maximum)      | O₂H compound formed (O₁₁) |\\n| 8.0              | +0.09                      | C–O–OH unstable state (O₁₃) |\\n| 9.0              | −0.21                      | Benzene ring rupture (O₁₄–O₁₅) |\\n| 9.5              | −0.20                      | C–O–C bridge bond formed (O₁₆) |\\n| 12.5             | −0.20                      | O₃ formation (O₁₉) |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"At low oxygen coverage, reactions favor stable adsorption and low-energy products such as water. As coverage increases, the surface begins to form OH and peroxide-like species. At even higher coverage, the chemistry becomes destructive, leading to bond bridging, ozone formation, and ring rupture. This shows a transition from surface functionalization to surface damage.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Adsorption is more energetically favorable at low oxygen coverage.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The appearance of C–O–C bridges and benzene ring rupture indicates that the polymer backbone is being altered. These reactions go beyond surface functionalization and involve structural damage. This suggests reduced mechanical and chemical stability at high oxygen loading.\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"In surface treatment processes, controlled functionalization is often desired without damaging the material. This figure shows that excessive oxygen leads to polymer degradation. Identifying the onset of ring rupture helps define safe processing limits. This is critical for plasma and oxidative treatments of polymers.\"}]}]","bbox":[{"panel_id":"a","x":3,"y":2,"width":668,"height":402}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/35/images/FIG4.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/35/images/FIG4.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/35/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/35/Density Functional Theory Study of Oxygen Adsorption on Polymer Surfaces for Atomic-Layer.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"35","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:image_caption"},"width":672,"height":406,"image_format":"jpeg","image_sha256":"f20d59391b19adae8f9ce813ef08ac731dfbccaee2e7fb2da02fef6d9a360754","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_6_fig_15.jpg","caption":"FIG. 15. Normalized atomic Cl counts as a function of  $\\mathsf{Ar}^+$  fluence. Subfigure (a) shows the results for OES experiments (where the ion energy is estimated to be about  $85\\mathrm{eV}$ ) and subfigure (b) shows results from MD ALE simulations where the ion energy is  $80\\mathrm{eV}$ . In both cases, the results are normalized using the value observed at the peak.","id":"train/atomic-layer-etching/simulation-usecase/6/fig_15","sample_id":"atomic-layer-etching/simulation-usecase/6/fig_15","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"box plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The experimental scatter plot shows a sharp decrease in Cl count with increasing ion dosage up to about 10–15 × 10¹⁵ Ar⁺/cm². After that, the count levels off around 0.5 arb. units, suggesting that most Cl atoms have been desorbed or sputtered and that the system has reached a steady state.\"},{\"panel_id\":\"b\",\"text\":\"The MD simulation results mirror this trend but show a steeper and more sustained decline. Initial Cl counts are slightly higher in simulation, but decrease more rapidly, falling below 0.2 arb. units by 50 × 10¹⁵ Ar⁺/cm². The more aggressive decline in simulations may reflect idealized surface models with fewer defect traps or re-adsorption pathways\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Ion Dosage (Ar⁺/cm²) × 10¹⁵ | Cl Count (arb. units)  |\\n|-----------------------------|------------------------|\\n| 0                           | 1.20                   |\\n| 2                           | 1.00                   |\\n| 4                           | 0.90                   |\\n| 6                           | 0.80                   |\\n| 8                           | 0.70                   |\\n| 10                          | 0.65                   |\\n| 12                          | 0.60                   |\\n| 14                          | 0.58                   |\\n| 16                          | 0.55                   |\\n| 18                          | 0.53                   |\\n| 20                          | 0.52                   |\\n| 25                          | 0.51                   |\\n| 30                          | 0.50                   |\\n| 35                          | 0.50                   |\\n| 40                          | 0.50                   |\\n| 45                          | 0.50                   |\\n| 50                          | 0.50                   |\"},{\"panel_id\":\"b\",\"text\":\"| Ion Dosage (Ar⁺/cm²) × 10¹⁵ | Cl Count (arb. units)  |\\n|-----------------------------|------------------------|\\n| 0                           | 1.60                   |\\n| 2                           | 1.20                   |\\n| 4                           | 1.00                   |\\n| 6                           | 0.80                   |\\n| 8                           | 0.65                   |\\n| 10                          | 0.55                   |\\n| 12                          | 0.48                   |\\n| 14                          | 0.42                   |\\n| 16                          | 0.38                   |\\n| 18                          | 0.34                   |\\n| 20                          | 0.30                   |\\n| 25                          | 0.27                   |\\n| 30                          | 0.24                   |\\n| 35                          | 0.22                   |\\n| 40                          | 0.20                   |\\n| 45                          | 0.19                   |\\n| 50                          | 0.18                   |\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"List\",\"answer\":\"Cl desorption, Surface sputtering, Bond breakage, Cl depletion over time\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"In both the experimental and simulation results, Cl count decreases as ion dosage increases. The experimental data shows a rapid initial drop followed by a plateau near 0.5 arb. units, indicating surface saturation or depletion. In contrast, simulations exhibit a sharper and more sustained decline, reaching much lower final values. This suggests that Cl is more easily removed in the idealized simulation environment, while experimental surfaces may retain Cl due to surface defects or incomplete reactions.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"Around 15 × 10¹⁵ Ar⁺/cm²\"},{\"question_type\":\"Application/Performance\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":8,"y":11,"width":686,"height":478},{"panel_id":"b","x":720,"y":7,"width":690,"height":481}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/6/images/fig_15.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/6/images/fig_15.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/6/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/6/Dynamics of plasma atomic layer etching Molecular.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"6","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":1411,"height":489,"image_format":"jpeg","image_sha256":"1fac63a969f142d7b775dfa54390c2dbba3660a16dfdb53394f215e7a46c5cbf","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_8_fig_6.jpg","caption":"FIG. 6. Synergy plot for GaN using  $\\mathrm{SF}_6$  plasma and TMA at  $300^{\\circ}\\mathrm{C}$ . The first 25 pulses only involve dosing  $\\mathrm{SF}_6$  plasma, the next 25 pulses only TMA, and the last 75 cycles are full ALE cycles with both  $\\mathrm{SF}_6$  plasma and TMA dosed. The inset shows the transition between only  $\\mathrm{SF}_6$  pulses and only TMA pulses.","id":"train/atomic-layer-etching/simulation-usecase/8/fig_6","sample_id":"atomic-layer-etching/simulation-usecase/8/fig_6","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart illustrates the relationship between film thickness and pulses/cycles, showing a decreasing trend if both reactants are dosed.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Pulses/Cycles | Film thickness (nm) |\\n|---|---|\\n| 0 | 43 |\\n| 25 | 43 |\\n| 50 | 42 |\\n| 75 | 33 |\\n| 100 | 25 |\\n| 125 | 18 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"At 25 pulses/cycles the SF6 plasma dosing is switched with TMA dosing, here a drop of 0.33 nm is observed. So based on this part of the graph the EPC is 0.33 nm.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It could be that there are temperature differences as an higher temperature can main a higher EPC.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It could be that the flurination step increases the thickness slightly with the incorporation of fluorine.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"It requires roughly 30 cycles.\"}]}]","bbox":[{"panel_id":"a","x":2,"y":1,"width":669,"height":510}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/8/images/fig_6.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/8/images/fig_6.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/8/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/8/Isotropic atomic layer etching of GaN using SF6 plasma and AI(CH3)3.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"8","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":675,"height":511,"image_format":"jpeg","image_sha256":"70cbbc1d040332c998b1f83d703cf4869dcce8e0f3b3e92ee8e499447d0d13be","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_8_fig_9.jpg","caption":"FIG. 9. RMS roughness as a function of etched thickness, plotted alongside the fitted change in RMS roughness as obtained using the curvature-dependent model from Ref. 85.","id":"train/atomic-layer-etching/simulation-usecase/8/fig_9","sample_id":"atomic-layer-etching/simulation-usecase/8/fig_9","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The line chart shows the relationship between etched thickness and RMS roughness, with experimental data points and a curvature-dependent model curve.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Etched thickness (nm) | RMS roughness (nm) |\\n|---|---|\\n| 0 | 2.65 |\\n| 5 | 2.4 |\\n| 10 | 2.2 |\\n| 15 | 2.0 |\\n| 20 | 2.0 |\\n| 25 | 1.9 |\"}]","vqa":"[{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"It could be that there is dust incorporated in the film or the deposition method with which the GaN is made is not uniform.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"This is not likely, even for infinite etch thickness, a material can never be perfectly flat.\"},{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"At least 20 nm must be etched away.\"},{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Yes/No\",\"answer\":\"Yes\"}]}]","bbox":[{"panel_id":"a","x":3,"y":6,"width":666,"height":527}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/8/images/fig_9.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/8/images/fig_9.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/8/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/8/Isotropic atomic layer etching of GaN using SF6 plasma and AI(CH3)3.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"8","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":669,"height":536,"image_format":"jpeg","image_sha256":"94b5c28753734707a4c68352c23fa9037d890ef43a18001e349c054bcb95ca03","metadata_license":"CC BY 4.0","image_license":"source_publisher_rights_reserved","image_reuse_status":"non_commercial_research_use_only","schema_version":"1.0.0"}
{"file_name":"images/train_atomic-layer-etching_simulation-usecase_9_fig_3.jpg","caption":"Fig. 3. (a) Potential energy curve for  $\\mathrm{SiCl_2}$  on the optimized Si(111) surface, here blue dots are the calculated data, and red line is the fitting function.  $Z = 0$  represents the equilibrium position, and insets are the local views of the desorbed species. (b) Potential energy curve for the Si-adatom (atom A1 in Fig. 2(a)) in the absence of chlorine, insets are the local views of the desorbed species.","id":"train/atomic-layer-etching/simulation-usecase/9/fig_3","sample_id":"atomic-layer-etching/simulation-usecase/9/fig_3","subset":"scatter-plot","split":"train","classification":[{"panel_id":"a","label":"scatter plot"},{"panel_id":"b","label":"scatter plot"}],"summarization":"[{\"panel_id\":\"a\",\"text\":\"The figure shows a potential energy curve as a function of surface coordinate Z, with calculated data points and a fitted line for the desorption of a SiCl2 unit.\"},{\"panel_id\":\"b\",\"text\":\"The figure shows a potential energy curve as a function of surface coordinate Z, with calculated data points and a fitted line for the desorption of a rest silicon atom.\"}]","data_extraction":"[{\"panel_id\":\"a\",\"text\":\"| Surface coordinate Z (Å) | Potential energy (eV) |\\n|---|---|\\n|-1 | 8 |\\n|-0.5 | 1.2 |\\n|0 | 0 |\\n|0.5 | 0.8 |\\n|1 | 1.5 |\\n|1.5 | 2.2 |\\n|2 | 2.4 |\\n|2.5| 2.5|\"},{\"panel_id\":\"b\",\"text\":\"| Surface coordinate Z (Å) | Potential energy (eV) |\\n|---|---|\\n|-1 | 8 |\\n|-0.5 | 1 |\\n|0 | 0 |\\n|0.5 | 0.5 |\\n|1 | 2 |\\n|1.5 | 3.2 |\\n|2 | 4.3 |\\n|2.5|5.2|\"}]","vqa":"[{\"panel_id\":\"b\",\"items\":[{\"question_type\":\"Process-Oriented\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"In principle can silicon be etched continuously, however an energy of 6.1832 eV is required per silicon atom to do this.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Factoid\",\"answer\":\"6.1832 eV\"}]},{\"panel_id\":\"a\",\"items\":[{\"question_type\":\"Structure-Property\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The SiCl2 unit has zero potential energy when it is in a standing up position on the surface.\"},{\"question_type\":\"Comparative/Trend\",\"questions\":\"\",\"answer_type\":\"Paragraph\",\"answer\":\"The desorption energy is 2.5159 eV which is provided by energetic ions from the argon plasma.\"}]}]","bbox":[{"panel_id":"a","x":0,"y":39,"width":672,"height":397},{"panel_id":"b","x":11,"y":525,"width":669,"height":388}],"source":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/9/images/fig_3.jpg","provenance":{"source_annotation":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/9/images/fig_3.json","source_content":"icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/9/content.json","source_pdf":["icdar2026-competition-data/train/atomic-layer-etching/simulation-usecase/9/Mechanism of photoassisted atomic layer etching of chlorinated Si(111) surfaces Insight.pdf"],"main_category":"atomic-layer-etching","sub_category":"simulation-usecase","paper_id":"9","first_classification_panel_id":"a","first_classification_label":"scatter plot","caption_source":"content.json:img_caption"},"width":680,"height":913,"image_format":"jpeg","image_sha256":"29fb684a0eb321552219fc1a347296adae14aad9ae5e8b26340f99d54007c769","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"}