Dynamic Bending Stress Sensing Using NV Centers in a Single-Crystal Diamond MEMS Cantilever
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Functional Diamond |
| Published | 2026-08-28 |
| Authors | Yuta OCHIAI, Vu Xuan Tung Duong, Zilong Zhang, Takahito Ono, Meiyong Liao |
| DOI | 10.1080/26941112.2026.2721758 |
A single-crystal diamond MEMS cantilever produced an ODMR stress sensitivity of 19.8 MHz/GPa during cantilever vibration. Tensile bending shifted the selected NV resonance toward higher frequency, while compressive bending shifted it toward lower frequency. The shift magnitude also varied with vibration phase, showing that the implanted NV layer can act as an integrated bending-stress sensor.
Sensor architecture
Section titled “Sensor architecture”The sensing element was a freestanding cantilever fabricated from a (100)-oriented single-crystal diamond substrate. NV centers formed near an ion-implantation-induced defective layer within the released beam. This arrangement put the sensing region directly under mechanical load and avoided attaching a separate diamond sensing element to the mechanical structure.
The authors applied a static magnetic field of approximately 3.29 mT. Its direction was adjusted to align approximately with one of the four NV crystallographic axes. The aligned orientation had the largest magnetic-field projection, which produced widely separated outer ODMR resonances. The authors used these outer resonances to evaluate stress-dependent shifts.
Measurements were performed in a temperature-controlled laboratory at approximately 25 °C. The magnetic-field configuration remained fixed, and no noticeable ambient temperature variation occurred during the experiments.
Cantilever fabrication and NV formation
Section titled “Cantilever fabrication and NV formation”Carbon ions at 180 keV formed a buried graphite layer approximately 200 nm thick at a depth of around 250 nm. The implantation also produced a graded defective region extending toward the diamond surface. During annealing, vacancies in this region combined with substitutional nitrogen atoms in the diamond lattice. This formed the NV-center layer above the graphite layer.
An additional single-crystal diamond layer approximately 0.4 µm thick was grown by plasma chemical vapor deposition. Oxygen plasma etching with a vapor-deposited aluminum mask defined the cantilever. A wet etch then removed the buried graphite and released the beam.
SEM measurements gave a cantilever length of 0.14 mm, a width of 0.01 mm, and a thickness of approximately 450 nm. Raman-photoluminescence measurements used 532 nm excitation after microfabrication. The measured spectra retained the characteristic NV-related optical emission after implantation, annealing, graphite removal, and MEMS processing.
Mechanical actuation and stress calibration
Section titled “Mechanical actuation and stress calibration”A laser Doppler vibrometer identified the cantilever resonance at 28.3 kHz. The device was mounted on a piezoelectric actuator for vibration measurements. Tip displacement varied linearly with actuator voltage up to 0.4 V.
The cantilever contacted the substrate above 0.4 V. Its maximum vibration amplitude was therefore limited by the clearance between the beam and substrate. The authors observed an initial upward deflection of approximately 4.5 µm.
A rectangular-beam calculation used an elastic modulus of 1100 GPa, a thickness of 0.45 µm, and a length of 140 µm. This gave a maximum surface stress of approximately 38 MPa per micrometer of tip displacement. A finite-element calibration then converted the measured tip displacement into maximum surface stress.
Finite-element analysis of the NV-containing region gave an average bending-stress magnitude of 201.12 MPa. The corresponding average shear-stress magnitude was 85.54 kPa, giving a shear-to-bending ratio of 0.04%. The mechanical response was therefore dominated by longitudinal tensile and compressive bending stress.
Pulsed ODMR measurement
Section titled “Pulsed ODMR measurement”The confocal setup focused 532 nm laser light onto the diamond surface. An RF field near 2.87 GHz was applied from below the sample. Fluorescence near 637 nm passed through optical filters and reached a photon counter.
Each measurement began with a 10 µs laser pulse. The timing selected the phase at which tensile stress on the back of the cantilever reached its maximum. Fluorescence was counted during the same 10 µs interval. Period A recorded fluorescence without RF excitation, while period B recorded fluorescence with RF excitation. The intervals were separated by 1 ms.
The normalized pulsed ODMR signal compared RF-induced fluorescence with the reference count. This reduced the influence of common fluctuations in the detected fluorescence. Preliminary optimization selected a 300 µs RF pulse because it produced the highest ODMR contrast and a stable fluorescence response.
Each spectrum covered 2.8 to 2.94 GHz in 1 MHz steps. Data were accumulated for 20 s and smoothed with five-point averaging. The same procedure measured the spectrum at the phase of maximum compressive stress. The actuator was driven at 0.2 Vpp and 0.4 Vpp during these measurements.
Stress response and sensitivity
Section titled “Stress response and sensitivity”Tensile bending shifted both selected outer resonances toward higher frequencies. Compressive bending shifted them toward lower frequencies. The measured shifts increased with actuator voltage.
| Drive condition | Bending state | ODMR peak shift | Frequency response |
|---|---|---|---|
| No vibration | Reference | 0 | Reference spectrum |
| 0.2 Vpp | Tensile | 2.0 ± 0.5 MHz | Higher frequency |
| 0.2 Vpp | Compressive | 2.0 ± 0.5 MHz | Lower frequency |
| 0.4 Vpp | Tensile | 4.0 ± 0.5 MHz | Higher frequency |
| 0.4 Vpp | Compressive | 4.0 ± 0.5 MHz | Lower frequency |
A linear fit yielded an ODMR stress sensitivity of 19.8 MHz/GPa. For comparison, Ivády et al. reported a theoretical pressure sensitivity of 10.3 MHz/GPa. Doherty et al. measured 14.58 MHz/GPa at room temperature under hydrostatic pressure.
The authors placed the 19.8 MHz/GPa result in the same general order as those pressure-based values. They did not treat the numbers as directly comparable because loading condition, crystal orientation, stress and strain tensors, and the NV coordinate system differ. The study also does not show superior sensitivity over previous NV-based mechanical sensing approaches.
Phase-resolved sensing
Section titled “Phase-resolved sensing”Phase-resolved measurements used the higher-frequency outer resonance because it produced a clearer dip and a better signal-to-noise ratio. The lower-frequency resonance was expected to show a similar stress response. The authors attributed the signal difference to the RF antenna and measurement system.
The cantilever vibration period at 28.3 kHz was approximately 36 µs. Photon counting used a window between 3.6 µs and 10 µs. Shifting that window within one vibration period sampled the ODMR response at different strain phases.
These spectra covered 2.97 to 3.02 GHz in 0.5 MHz steps. Data were accumulated for 20 s and smoothed with five-point averaging. The actuator was driven at 0.4 Vpp. The measured peak-shift magnitude varied periodically with the counting-window delay. The reported measurement error was 0.5 to 1.0 MHz.
Limits and device implications
Section titled “Limits and device implications”The authors identify four main limits. The ODMR signal-to-noise ratio was relatively low. The displacement-to-stress calibration and magnetic-field orientation introduced uncertainty. The frequency sweep also required a long acquisition time.
They recommend improved photon collection and quantitative resonance fitting. More accurate mechanical displacement calibration would reduce stress uncertainty. Better magnetic-field and temperature characterization would also improve the extracted shifts. The phase-resolved error of 0.5 to 1.0 MHz was substantial beside the 2.0 ± 0.5 MHz shift measured at 0.2 Vpp.
The experiment focused on beam-direction bending stress because it dominated the mechanical loading. The general spin-stress Hamiltonian included transverse stress terms, but the results did not evaluate separate shear-stress sensing. The demonstrated performance therefore supports feasibility under controlled laboratory conditions rather than a complete stress metrology system.
For diamond device selection, the result favors single-crystal diamond when the NV sensing region and the mechanical structure must occupy the same body. The key advantage is direct conversion of bending strain into an ODMR frequency shift. A practical instrument must still address resonance fitting, vibration phase, magnetic alignment, temperature control, and the time required to acquire each spectrum.
Materials and support
Section titled “Materials and support”The paper needs thin single-crystal diamond for a MEMS cantilever sensing application, with the demonstrated beam about 0.45 µm thick and built on a (100)-oriented substrate. 6CCVD’s diamond materials catalog lists single-crystal diamond grown by MPCVD with thickness from 0.1 to 500 µm, while implantation, annealing, graphitic release, and MEMS patterning remain separate process steps defined by the paper. Questions about those custom fabrication requirements can be sent to engineering support, and global shipping is available.
View Original Abstract
In this study, a single-crystal diamond (SCD) MEMS cantilever was fabricated using a buried graphite sacrificial-layer process. A region containing nitrogen-vacancy (NV) centers formed near the ion-implantation-induced defective layer was directly utilized as the stress-sensing region of the released cantilever, enabling dynamic stress sensing without the use of externally attached nanodiamonds. We investigated the relationship between the stress induced by the cantilever vibration and the peak shift of the fluorescence intensity of the optically detected magnetic resonance (ODMR) spectrum. By approximately aligning the static magnetic field with one of the four NV-center crystallographic axes, we were able to evaluate the dependence of the peak shift on stress. Our findings revealed that the peak shift can effectively detect uniaxial stress due to normal vibration. In addition, as the vibration amplitude increased, the magnitude of the ODMR peak shift also increased, showing a trend consistent with previously reported NV-center stress responses. Furthermore, the ODMR shift exhibited periodic modulation corresponding to the cantilever’s vibration phase. These results indicate that NV centers in micromachined SCD cantilevers can be used to detect vibration-induced dynamic bending stress.
Electronic Grade Intrinsic Diamond Epitaxial Wafer H Terminated Superior 100 Mosfet Ready Substrate
View Specs →Products
View Specs →