Problem

The ability to assemble two-dimensional (2D) material layers into van der Waals (vdW) heterostructures is critical for exploring novel quantum phenomena and developing advanced devices. The dry-transfer technique, which uses flexible polydimethylsiloxane (PDMS) stamps, is a cornerstone of this assembly. While artificial intelligence (AI) and robotics are increasingly being integrated into 2D material research to automate the dry transfer process, achieving truly autonomous stacking requires highly standardized and uniform PDMS transfer templates. The thermomechanical behavior of these templates, including their geometry, surface smoothness, and temperature-dependent expansion, significantly impacts the reproducibility and reliability of the transfer process, posing a challenge for current methods.

Method

The researchers developed a hot-casted-droplet batch fabrication method for creating dome-shaped PDMS templates specifically for dry transfer of 2D materials. The process begins with a precise formulation of a PDMS precursor (Sylgard 184 mixed at a 10:1 ratio), followed by thorough stirring and a one-hour degassing period. A microinjector-assisted dispensing system, featuring a motorized mechanical controller and a temperature-controlled hotplate, is then used to dispense uniform droplets. Optimized extrusion parameters ensured clean droplet detachment and enabled batch fabrication of 2x5 template arrays. The curing temperature, maintained between 120-160℃, was identified as a critical parameter for controlling the templates' apex curvature. The fabricated templates underwent comprehensive characterization: Atomic Force Microscopy (AFM) was used to assess surface roughness, bright-field microscopy to determine geometric parameters (major/minor semi-axes, apex curvature radius), and a high-precision analytical balance measured template mass. Thermomechanical performance was evaluated by simulating a standard 2D material transfer process, precisely controlling substrate temperature, and using image processing to track contact area expansion.

Results

The batch fabrication method successfully produced PDMS templates with exceptionally smooth surfaces, achieving a root-mean-square (RMS) roughness as low as 0.3 nm at a curing temperature of 120℃. Surface roughness was found to increase with higher curing temperatures, attributed to accelerated crosslinking. The method also demonstrated excellent uniformity and reproducibility in surface smoothness, with very narrow statistical distributions. Geometrically, the apex curvature radius showed a strong inverse relationship with curing temperature, decreasing from approximately 9 mm at 120℃ to 3 mm at 160℃, indicating that higher temperatures yield sharper, more pronounced domes. Template mass remained highly consistent across all curing conditions, averaging about 14 mg, which highlights the homogeneity of the precursor mixture and stability of the cross-linking reaction. Thermomechanically, the PDMS templates exhibited temperature-dependent expansion: higher curing temperatures resulted in slower thermal expansion, with total displacements reducing from approximately 800 µm (120℃) to 600 µm (160℃). This effect was linked to the smaller apex radii at higher curing temperatures, which require greater compressive stress for conformal contact and lead to larger contact angles, suppressing expansion.

Implications

This research provides a scalable and parameterized fabrication protocol for creating uniform PDMS transfer stamps with ultra-smooth surfaces and precisely controlled thermomechanical properties. The ability to quantitatively correlate curing parameters with surface morphology, geometric dimensions, and thermal expansion characteristics establishes a robust and reproducible parametric database. This standardized and data-driven foundation is poised to play a pivotal role in the advancement of AI-driven robotic assembly of 2D material heterostructures. By allowing for the automated selection of suitable template geometries and the predictive control of contact dynamics through temperature tuning, this method significantly contributes to the development of intelligent, feedback-controlled 2D material transfer systems, moving closer to truly autonomous fabrication processes.