Fig. 1. CD of integrated circuit layout

Fig. 2. (a) Sketch of imaging process and (b) diagram of structure for DUV immersion lithography tool^[1]

Fig. 3. Schematic diagram of OPC. (a) Imaging process before and after OPC optimization^[2]; (b) masks and their image results before and after OPC optimization^[15]

Fig. 4. Schematic diagram of imaging processes using two different types of masks. (a) Binary mask; (b) Alt-PSM^[2]

Fig. 5. Conventional illumination and common OAIs^［2］. (a) Conventional circular illumination; (b) (c) dipole illuminations; (d) (e) quadrapole illuminations; (f) annular illumination

Fig. 6. Schematic diagram of imaging principle of lithography system^[52]

Fig. 7. Comparison between thin-mask model and thick-mask model. (a) Kirchhoff approximation of photomask; (b) diffraction near-field model of thick mask^[53]

Fig. 8. Schematic diagram of PE and EPE

Fig. 9. Schematic diagram of process windows^[82]. (a) Rectangular process window; (b) elliptical process window

Fig. 10. Transformation and solution process of computational lithography problem^[83]

Fig. 11. Different types of OPC methods. (a) Target pattern; (b) RBOPC result; (c) EBOPC result; (d) PBOPC result^[109]

Fig. 12. Examples of RBOPC optimization results and correction rule tables^[110]

Fig. 13. EBOPC method. (a) Optimization process of EBOPC; (b) image result after optimization^[113]

Fig. 14. PBOPC simulation results based on SD algorithm. (a) Target pattern; (b) optimized gray mask; (c) optimized binary mask; (d)-(f) imaging results corresponding to mask patterns of first row^[30]

Fig. 15. Comparison of ILT methods based on SD and CG algorithms. (a) Target pattern; (b) mask optimization result obtained by SD algorithm; (c) mask optimization result obtained by CG algorithm; (d)-(f) imaging results corresponding to mask patterns of first row ^[43]

Fig. 16. Comparison of PBOPC methods based on SIAOS and SGD. (a) Optimization results of mask and print image; (b) convergence curves^[140]

Fig. 17. Comparison of ILT optimization results based on Adam and SGD algorithms. (a) Sarget pattern; (b) mask optimization result based on Adam algorithm; (c) mask optimization result based on SGD algorithm; (d) convergence curves of these two algorithms; (e) print image of Adam algorithm; (f) print image of SGD algorithm^[157]

Fig. 18. Comparison of optimization results between PBSO and parametric SO. (a) Mask pattern; (b) dipole illumination obtained by parametric SO; (c) source optimization result obtained by PBSO; (d) cross-sections of aerial images; (e) process windows^[163]

Fig. 19. SMO technique and its improvement on process window^[25]

Fig. 20. SMO workflow for full-chip layout optimization^[4]

Fig. 21. Optimization results and convergence curves of the SISMO, SESMO, and HSMO algorithms ^[52]

Fig. 22. Optimization results and convergence curve of PBSMO algorithm^[42]

Fig. 23. Comparison of SMO optimization results between ALM algorithm and CG algorithm. (a)-(f) Optimized sources, masks, and print images obtained by ALM algorithm and CG algorithm, respectively; (g) average process windows obtained by two algorithms^[44]

Fig. 24. Comparison of SMO simulation results among SD algorithm, conventional level-set algorithm, and narrow-band level-set algorithm. (a)-(l) Optimization results of sources, masks, and print images based on three algorithms; (m) convergence curves of these algorithms^[192]

Fig. 25. Comparison of simulation results between traditional SMO and robust hybrid SMO algorithms (considering the influence of source blur and flare). (a) (f) Source optimization results; (b) (g) mask optimization results; (c) (h) print images under ideal parameters; (d) (i) print images influenced by the source blur (standard deviation of source blur σ_J=0.05); (e) (j) print images influenced by the flare (flare ratio S=2%)^[197]

Fig. 26. Compressive measurement process and reconstruction process of the linear CS^[202]

Fig. 27. Comparison of fast SO techniques based on diffraction subspace method and ACS method (L is the function dimensionality after compression). (a)-(h) Source optimization results and imaging results; (i) (j) process windows^[210]

Fig. 28. Comparison of the fast SO techniques based on the LCS, ACS, CG, and PSO algorithms^[174]

Fig. 29. Comparison of simulation results for two nonlinear CS-OPC methods (K is the compression ratio). (a)-(h) Mask optimization results and print image results for different methods; (i) convergence curves of the cost functions; (j) process windows^[216]

Fig. 30. Typical optimization examples of the CS-SMO algorithms. (a)-(d) Source patterns; (e)-(h) mask patterns; (i)-(l) print images^[218]

Fig. 31. Simulation results of the fast OPC method based on adaptive kernel regression at 45 nm metal layer. (a) Corrected mask pattern; (b) optimized lithography image^[49]

Fig. 32. Comparison of the fast EUV lithography mask diffraction near-field calculation methods (from left to right, it uses the rigorous electromagnetic field method, non-parametric kernel regression method, Kirchhoff approximation method, and DDM method). (a) Mask diffraction matrices obtained by different methods; (b) error maps of mask diffraction matrices^[65]

Fig. 33. Prediction method and results of the etch bias based on neural network. (a) Features of mask pattern are extracted and inputted into the neural network to predict the etching bias; (b) comparison between measured etch bias and predicted etch bias^[232]

Fig. 34. Fast mask optimization method based on the GAN-OPC model, where OPC layout is outputted by generator^[105]

Fig. 35. Sketch of fast mask diffraction near-field calculation method based on FCN^[67]

Fig. 36. Layout hotspot detection based on the CNN model^[104]

Fig. 37. Designed masks and their print images obtained by MCNN^[106]

Fig. 38. Comparison of the simulation results between SMO algorithm and SMPO algorithm. (a)-(d) Simulation result before optimization; (e)-(p) simulation results of SMO, SISMPO, and SESMPO algorithms^[254]

Fig. 39. Workflow of DTCO^[267]

Fig. 40. Expected development situation of the pitch sizes of some key devices on the chip^[271]

Fig. 41. Sketch of EUV lithography system^[274]

Fig. 42. Mask correction results and algorithm convergence curves for defects at different positions (mask defects locate at the center, edge, and corner of the pattern, respectively) ^[95]

Fig. 43. Simulation results of the EUV-SMO methods based on pixelated source model. (a)-(c) Source, mask, and print image obtained by the SMO algorithm; (d)-(f) source, mask, and print image obtained by the “SMO+Retargeting” method^[275]