Suppressing high-dimensional crystallographic defects for ultra-scaled DNA arrays

While DNA-directed nano-fabrication enables the high-resolution patterning for conventional electronic materials and devices, the intrinsic self-assembly defects of DNA structures present challenges for further scaling into sub-1 nm technology nodes. The high-dimensional crystallographic defects, including line dislocations and grain boundaries, typically lead to the pattern defects of the DNA lattices. Using periodic line arrays as model systems, we discover that the sequence periodicity mainly determines the formation of line defects, and the defect rate reaches 74% at 8.2-nm line pitch. To suppress high-dimensional defects rate, we develop an effective approach by assigning the orthogonal sequence sets into neighboring unit cells, reducing line defect rate by two orders of magnitude at 7.5-nm line pitch. We further demonstrate densely aligned metal nano-line arrays by depositing metal layers onto the assembled DNA templates. The ultra-scaled critical pitches in the defect-free DNA arrays may further promote the dimension-dependent properties of DNA-templated materials.


S3.2 Measuring the DNA line pitches and widths for different designs
The 4-nm thick Ag film was deposited by magnetron sputtering, and the 4-nm thick Au and Al film were deposited by thermal evaporation. The SEM images displayed cracked films and randomly formed nanoparticles on the surface of DNA templates, indicating that these metals could not conformally adhere to the DNA surface.
Supplementary Figure 17. SEM images of the 4-nm thick Pd (a-b) and Ni (c-d) deposited on the DNA templates with 16.8-nm line pitch.

S5.2 AFM image after different metal deposition on DNA pattern
Supplementary Figure 18. AFM images of 4-nm thick Pd (a-b) and Ni (c-d) deposited onto the DNA templates with 16.8-nm line pitches.

S5.3 Statistical evaluation about the reproducibility of the metal pattern
In order to statistically evaluate the reproducibility of the metal patterns, 2~3 nm thicknesses of Pd were deposited on the DNA template with 25.2-nm pitch or 16.8-nm pitch. We separately counted up to 100 lines before and after Pd deposition under AFM measurement. The statistical analysis indicated that ( Supplementary Fig.19), before and after metal depositions, the critical pattern dimensions (including the line width, line pitch, and the spacing) remained similar values. Figure 19. The ratio of the average critical dimensions before and after metal deposition for (a) 25.2-nm DNA pitch and (b) 16.8-nm DNA pitch. The values were obtained via AFM measurements. Up to 100 lines before and after Pd deposition were counted.

S5.4 The method for cross-sectional TEM characterization.
To understand the morphology of mental layer coating within the nanotrench, we prepared samples using FEI Helios G4 UX DualBeam focused ion beam (FIB)/SEM system. To protect the interested samples, 200-300 nm depth of carbon and 2 μm depth of platinum were deposited sequentially under electron and ion beams. Then the sample was milled about 12 μm depth using the Ga ion beam, forming the U-cut. Lamella was extracted from the substrate into copper TEM grid through lift-out method. Finally, the lamella was thinned down by ion beam at 30 kV and polished at 3 kV. The cross-sectional images were imaged by TEM (FEI Tecnai F20) at 200 kV

S5.5 Cross-section TEM for metal nano-line arrays
We used the DNA templates with 25.2-pitch as a model system for the FIB milling and the crosssectional analysis. We first coated the surface-deposited DNA templates with four different thicknesses of Pd layer (6 nm, 3.7 nm, 2.5 nm, and 2 nm), and evaluated the groove depths using AFM (up to 100 lines were counted). For subsequent FIB milling, we sequentially deposited 200 nm amorphous carbon and more than 2 microns Pd layer onto the metal-coated DNA templates to protect the beneath structures from the high-energy ion milling process, followed by cutting the sample along the x-y directions. Then we imaged the x-y cross-section morphologies via TEM (Supplementary Fig. 21a). Notably, to enhance the structural rigidity of DNA templates, we used Ni 2+ to treat the DNA templates prior to their surface deposition We observed that, before FIB milling, the DNA-templated Pd patterns exhibited similar groove depths (~2 nm) to that of pure DNA templates ( Supplementary Fig. 20), regardless of the metal thicknesses. Thin-layer Pd coating did not affect either the pattern integrity or the groove depths of DNA templates, which was in consistent with the previous reply to the reviewer.
However, after FIB milling, we observed a metal thickness-dependent variation of the crosssection morphologies ( Supplementary Fig. 21). When the metal thicknesses (6 nm and 3.7 nm) exceeded the typical depths of DNA grooves (~2 nm), metals layers on top of the DNA lines and within the DNA grooves connected from the side, producing the continuous ripple-like periodic films ( Supplementary Fig. 21 a, b and c). Notably, the lateral metal growth within the DNA grooves were confined by the sidewalls of DNA lines, displaying similar dimensions and periodicities. The sidewalls DNA lines also prevented the lateral metal growth from penetrating into the lattices of DNA lines. The faint shades around the metal layer were likely to be the residuals at the cut edge during FIB milling.
Decreasing the metal thicknesses from 6 nm to 3.7 nm led to smaller groove depths from 3 nm (at 6-nm metal thickness) to ~1.7 nm (at 3.7-nm metal thickness) ( Supplementary Fig. 20), indicating more stresses pressed onto the metal ripples. At thinner metal thicknesses (2.5 nm and 2 nm), the periodic 3D morphologies vanished ( Supplementary Fig. 21 a, d and e). The groove depths decreased from less than 1 nm at 2.5-nm metal thickness to zero nm at 2-nm metal thickness (which is the minimal value we could deposit with our deposition facilities), leaving only planar morphology.
Considering that such planar morphologies were not observed in the AFM images before FIB milling, we therefore ascribed these morphological changes mainly to the template destruction during FIB milling, rather than during DNA-templated metal patterning. FIB milling required the deposition of 200 nm amorphous carbon and more than 2 microns Pd layer onto the deposited metal patterns followed by high-energy processing, which introduced heavy weights and stresses to be pressed onto the DNA-templated thin metal layers. Eventually, once the DNA-templated metal pattern could not resist the external stresses, the 3D space-and-line morphology collapsed into the planar morphology, as being observed at 2.5 nm and 2 nm metal thicknesses.
The collapse of the 3D morphology correlated with weak side-connection strength of the DNAtemplated metal patterns. At the metal thicknesses of 2.5 nm and 2 nm, because metal thicknesses were not significantly higher than the groove depth (~2 nm), the side connection could barely form. The resulting weak connections, as indicated by the cracks and different contrasts in the metal layers (signs of film discontinuity in Supplementary Fig. 21 d and e), failed to resist the stresses during the FIB milling. As results, only planar morphology could be produced. In contrast, for thicker metal layers (6 nm and 3.7 nm), the metals growth exceeded the groove depth, and formed strong side connections with those on top of the DNA lines. As indicated in the cross-section TEM images ( Supplementary Fig. 21 b and c), continuous ripple-like metal films with intact side connections, were not fully planarized during FIB. Meanwhile, we still observed signatures of the stress-induced distortion. The cross-section shape of DNA lines deviated from the designed rectangular shape into the trapezoid shape, as well as the descent groove depths. Notably, because FIB milling deformed DNA templates, it remained challenging to use the FIB milling for exploring the metal growth dynamics at the DNA sidewalls. Figure 20. The groove depths measured by AFM and TEM at different Pd thicknesses. Notably, we did not perform FIB for DNA templates without metal deposition, because of their fragile nature of DNA structures under high-energy processing.