Design and fabrication of guiding patterns for topography-based searching of 2D devices for scanning tunneling microscopy measurements

Technical Summary: Design and Fabrication of Guiding Patterns for Topography-Based Searching of 2D Devices in STM

Problem Statement
Scanning tunneling microscopy (STM) and spectroscopy (STM/S) are essential for probing atomic-scale structural and electronic properties of two-dimensional (2D) materials. However, a significant technical bottleneck exists in locating microscale 2D devices (typically 10–20 µm) within the limited field of view of an STM tip (1–1.5 µm). Traditional navigation methods face specific limitations:

• Optical Viewports: Many STM systems, particularly those operating under extreme conditions (cryogenic temperatures, high magnetic fields), lack optical access, making visual alignment impossible.
• Capacitive Sensing: While capacitance-based navigation is established, it requires specialized hardware (capacitance bridges) and specific device configurations (e.g., back-gate setups) to detect subtle capacitance changes at conductor-insulator boundaries.
• Direct Topographic Search: Relying solely on STM topography has been historically impractical due to the small field of view and the inability to distinguish features on common insulating substrates like SiO₂ or h-BN without prior knowledge of the tip's location.

Methodology
The authors propose a navigation strategy that utilizes custom-designed, sub-micron geometric guiding patterns etched directly into the SiO₂/Si substrate. This approach requires no hardware modifications to existing STM setups and is fully compatible with standard device fabrication workflows.

1. Design of Guiding Patterns:
   The guiding map spans a 1.8 mm × 1.8 mm area, divided into a 9 × 9 array of 200 µm × 200 µm square cells. The central unpatterned region is reserved for the 2D device. The navigation logic relies on three distinct geometric features:

   • Coordinate-Encoding Sectors: Circular sectors with angular spans from 10° to 90° encode numerical indices (1–9). Odd numbers are placed in the first quadrant, and even numbers in the fourth. A "b-variant" (e.g., 7b) is distinguished by an additional quarter-sector in the third quadrant. This allows the tip to determine its coarse position (~100 µm precision) by decoding the angular orientation of the pattern.
   • Boundary Markers: Linearly arranged half- and quarter-circles mark the edges and corners of the grid cells, enabling high-precision localization of specific boundaries.
   • Calibration Bars: A series of bar-shaped features (1 µm wide, 5 µm spaced) near the center allow for local calibration of the stage's x- and y-axis displacement and rotation, correcting for piezoelectric nonlinearities and mounting misalignments.

2. Fabrication and Integration:

   • Substrate Preparation: Patterns were written using 100 kV electron-beam lithography (PMMA resist) on SiO₂/Si wafers and transferred via CHF₃/O₂ reactive ion etching to a depth of ~40 nm.
   • Device Assembly: A monolayer graphene/h-BN heterostructure (~20 µm × 20 µm) was assembled using a dry transfer method involving PC and PPC stamps and a PDMS planar sheet. The stack was deterministically flipped and transferred onto the central unpatterned region of the guiding chip.
   • Electrical Contacts: To ensure the guiding patterns remain conductive (preventing tip crashes on insulators during approach), large-area Ti/Au electrodes were deposited via a shadow mask. One electrode contacts the device for bias, while the other serves as a gate contact.

3. Navigation Protocol:
   The process follows a decision-tree logic:

   • Coarse Localization: The tip performs a 1.5 µm × 1.5 µm scan to identify a sector pattern. By moving the stage in ~100 µm steps and scanning adjacent areas, the unique combination of indices (e.g., '7' → '4' → '4b') narrows the location to a specific 100 µm × 100 µm square.
   • Fine Positioning: The tip navigates toward the unpatterned device region by identifying boundary markers (half-circles) and determining the corner of the device area with ~1 µm accuracy.
   • Final Calibration: Before landing on the active device, the tip scans the calibration bars to correct for stage displacement offsets (found to be ~14% in x and ~10% in y) and rotation (<4°).

Key Results

• Successful Navigation: The authors demonstrated the ability to locate a sub-20 µm graphene/h-BN device solely through topographic imaging of the guiding patterns and in situ stage calibration, without optical viewports or capacitive sensing.
• Atomic Resolution: Once located, the STM achieved atomically resolved imaging of the graphene honeycomb lattice and the moiré superlattice formed by the graphene/h-BN heterostructure (periodicity ~0.2 nm, twist angle ~2.3°).
• Spectroscopy: Differential conductance (dI/dV) spectra were successfully acquired, showing clear shifts in the electronic density of states when back-gate voltages (0 V and 50 V) were applied.
• Efficiency: The entire navigation process was completed within half a day to a day, a timeframe compatible with multi-day STM measurement campaigns.

Significance and Claims
The paper claims that this topography-based strategy offers a practical and versatile alternative to existing navigation methods for STM studies of 2D systems.

• Hardware Independence: Unlike capacitance-based methods, this technique requires no additional electronics or specialized device configurations, making it applicable to any STM system equipped with xy-stage motion sensors.
• Compatibility: The fabrication process is fully compatible with standard 2D device workflows and functions effectively under extreme conditions (tested at 4.5 K, capable of 300 mK and 11 T).
• Scalability: The deterministic, decision-tree-based navigation logic is presented as a framework that could be automated via pattern recognition or machine learning.
• Broad Applicability: While demonstrated on graphene, the authors state the strategy is broadly applicable to other 2D materials (e.g., TMDs) and heterostructures with sub-10 µm active regions, provided the surface remains pristine.

The authors conclude that this approach removes a critical barrier to studying emerging quantum phenomena in miniaturized 2D devices, particularly in setups where optical or capacitive navigation is unavailable.