Skip to content

Single crystalline diamond metasurface with large elasticity and tunable structural colors

FieldValue
JournalDiamond and Related Materials
Published2026-08-25
AuthorsTiqing Zhao, Pak San Yip, Ruihan Xu, Ran An, Juzheng Chen
DOI10.1016/j.diamond.2026.114100

Single-crystal diamond nanopillars fabricated in ordered arrays returned to their original shape after unloading, while finite element analysis placed their base strain above 9%. The process also produced maskless nanoneedles about 30 nm in diameter and orange, green, and blue structural colors. These results place reversible nanoscale deformation and visible-light diffraction on the same single-crystal diamond platform.

Commercial optical-grade 〈100〉-oriented CVD single-crystal diamond, measuring 7 mm × 7 mm × 0.25 mm, served as the starting substrate. The surface received sequential piranha, acetone, ethanol, and deionized water cleaning for about 5 min. PMMA 950K A4 photoresist was spun at 4000 rpm for 50 s and baked at 180 °C for 1.5 min. The resulting film was about 200 nm thick.

Electron-beam lithography used a 30 kV accelerating voltage and a 280 μC/cm² exposure dose. A 130 nm titanium sacrificial layer was deposited by PVD, followed by lift-off in NMP. ICP-RIE used oxygen at 30 sccm, 700 W ICP power, 100 W table power, and 10 mTorr chamber pressure. Masked etching lasted 5 to 15 min. Hydrofluoric acid then removed the titanium.

EBSD confirmed the (100) face. The Raman spectrum had the single-crystal diamond peak at 1332 cm⁻¹, with a narrow full width at half maximum. Photoluminescence under 532 nm excitation identified nitrogen-vacancy centers. TEM and SAED confirmed the single-crystal structure and [100] growth direction of a representative pillar.

The measured shapes showed systematic mask erosion rather than straight cylindrical pillars. The authors measured nine pillars for each of two designed diameters.

Designed diameterMean height (nm)Mean bottom diameter (nm)Mean top diameter (nm)Side angle (degrees)
300 nm752 ± 8365 ± 6240 ± 785.27 ± 0.43
200 nm769 ± 17223 ± 10101 ± 985.47 ± 0.39

The 100 nm design produced a pillar with an aspect ratio of about 6. Taper was attributed to progressive titanium consumption and edge shrinkage under high-power ion bombardment. AFM gave a height of about 770 nm, consistent with the SEM value of 780 nm. Surface roughness changed little. Rq was 1.294 nm before etching and 1.180 nm after 5 min of RIE.

The tapered profile matters for metasurface design. The top diameter sets the exposed pillar width, while mask consumption determines that width. The arrays remained deterministic because EBL defined each site before etching.

The maskless route etched polished diamond directly in O2 plasma without lithography. Tilting the stage at 10°, 45°, or 60° combined dynamic micro-masking with directional ion bombardment. Etching time extended to 30 min.

At 10° tilt and 15 min, the array had needle diameters of approximately 30 nm and depths of 350 nm. At 60° and 15 min, needles were about 40 nm in diameter and 1050 nm deep. Their aspect ratio exceeded 20. The authors also report a geometry with about 30 nm diameter and an aspect ratio of about 24.

The tips had near-90° verticality, and the array surface remained flat. This route needs only one O2 plasma etch. It avoids mixed gases, particle masks, hydrogen, and complex FIB milling. Its limitation is position control. Needle positions were stochastic, unlike the EBL-defined lattice.

Elastic bending and finite element analysis

Section titled “Elastic bending and finite element analysis”

A 200 nm diameter pillar was bent in situ with a flat diamond indenter while SEM recorded the shape. Its tip moved 123 nm horizontally at a maximum indentation displacement of 165 nm. That deflection was about 16% of the pillar length. Three independent bending experiments produced an oblique fracture bevel at the base when the pillar fractured.

The fracture face sat about 54° to the base crystal face. The authors related the crack path to stress resolved along 〈111〉 and the weak bonding of the {111} face. This is their crystallographic interpretation of the post-test geometry.

Before catastrophic fracture, unloading restored the pillar to its original shape. The authors did not apply an Au coating during these mechanical tests.

ABAQUS models used the measured pillar geometry under displacement control. The indenter and pillar were elastic, with a contact friction coefficient of 1.0 and a diamond Young’s modulus of 1100 GPa. The simulated base strain exceeded 9%, which the authors describe as approaching diamond’s theoretical elastic limit. The simulated bent profile matched the SEM sequence. The authors interpret this agreement and high strain as evidence that the etched array retained pristine mechanical properties.

Structural color was measured under white LED illumination at 50° incidence. A camera viewed the diffracted light along the substrate normal through a 2× objective. First-order diffraction followed λ = d sin θ. The duty cycle was designed as 50%.

ColorMeasured diameter (nm)Measured pitch (nm)Calculated pitch (nm)
Orange500740790
Green475680690
Blue385580610

The calculated pitches agreed with the measured arrays. The blue array used gaps of about 65 nm between adjacent pillars. Combining different pitches in one pattern produced a wider visible-spectrum color image.

Linear-polarizer images were recorded from 0° to 180°. The authors saw no discernible change in perceived hue. They attributed the polarization-insensitive response to the circular pillar geometry and square lattice symmetry. They also fabricated a QR code with pillar dimensions spanning the nanometer and micrometer scales.

The authors noted that large pitches and small duty cycles limit color intensity. Their 50% duty-cycle design addressed this constraint. The color response was measured under one oblique illumination geometry, not under strain-driven tuning.

The paper positions the masked route as a scalable alternative to laser ablation and FIB milling. The authors describe those methods as costly, susceptible to ion implantation damage, and inefficient for ordered structures. EBL and RIE provide site-specific control, but the measured taper shows that mask design must account for titanium loss during etching.

Prior work had shown elasticity in isolated nanoneedles and FIB-milled diamond bridges, but the authors argue that those routes were difficult to reproduce or scale. The maskless method reaches about 30 nm diameter and an aspect ratio of about 24 in one O2 etch. It trades deterministic placement for smaller diameters. The paper does not give a numerical benchmark for the earlier attempts.

The study does not report fabrication yield, writing time, wafer-scale dimensional uniformity, or device-level throughput. The pillar statistics use nine samples per diameter. No cyclic fatigue test establishes long-term recovery. The FEM friction and elastic assumptions remain model inputs.

Engineers can choose between two fabrication routes. Masked EBL and RIE suit ordered metasurfaces with specified pitch and diameter. Maskless angled etching suits dense, ultra-thin needle arrays when stochastic placement is acceptable. The results support local elastic deformation above 9% in 〈100〉 single-crystal diamond, but base fracture, mask-induced taper, and untested fatigue remain important constraints.

For dynamic optical use, the authors propose embedding the metasurface in a flexible PDMS substrate. Applied strain could change pillar spacing and tilt, shifting the diffracted wavelengths. That strain-tunable response remains a proposed future design, not a measured result.

The paper requires an optical-grade 〈100〉-oriented single-crystal diamond plate for EBL and RIE fabrication of the metasurface. 6CCVD offers MPCVD-grown single-crystal diamond plates and wafers up to 125 mm. Its thickness range of 0.1 to 500 micrometers includes the paper’s 0.25 mm plate, as shown in the diamond materials catalog. The required orientation, polished surface finish, and process integration can be reviewed through engineering support.

View Original Abstract

Scalable, high-precision fabrication of single crystalline diamond (SCD) nanostructures remains a key challenge for next-generation optoelectronic and nanomechanical devices. Here, we present a highly controllable fabrication approach using electron-beam lithography (EBL) and optimized vertical reactive ion etching (RIE) to create large-area SCD nanopillar arrays. By precisely tuning the fabrication parameters, we realized high-aspect-ratio (up to 6:1) nanopillars with diameters ranging from 100 to 300 nm. To push the dimensional limits further, we additionally introduced a maskless angled etching method, yielding ultra-thin nanoneedles (down to ~30 nm diameter) with extreme aspect ratios (>20). Using in situ scanning electron microscopy (SEM) bending tests and finite element method (FEM) simulations, we reveal that these high-aspect-ratio SCD nanopillars exhibit remarkable mechanical robustness, enduring large elastic strain (>9%) and demonstrating full shape recovery upon unloading. Furthermore, exploiting the nanoscale resolution and precise dimensional control of our arrays, we engineered diamond-based metasurface that generates diameter- and pitch-dependent structural colors. This work establishes a scalable platform for fabricating strain-tunable diamond nanostructures with combined mechanical reliability and optical versatility, paving the way toward advanced robust micro-electromechanical systems (MEMS), photonics and flexible display technologies.