betaFeedback

WorksPhysics and Astronomy

Angular displacement readout of a mechanical oscillator with a guided mode resonance

Abstract

Measuring the angular displacement of a mechanical oscillator is a ubiquitous task; however, the multimode nature of angular optomechanical coupling makes coherent signal enhancement challenging. Here we demonstrate coherently enhanced angular displacement readout with an integrated guided mode resonance (GMR) structure, applying it to precision readout of a nanomechanical oscillator. Specifically, we fabricate subwavelength gratings into 100-nm-thick Si3_3N4_4 membranes and record their vibration by direct transmission measurements. The narrow linewidth ≈2.5  mrad\approx 2.5\;\text{mrad} of the GMR enables a shot-noise-limited displacement imprecision of 10−9  rad/Hz 10^{-9}\;\text{rad}/\sqrt{\text{Hz}} with nanowatts of optical power, sufficient to resolve the thermal motion of a Q≈106Q\approx 10^6 torsion mode with a signal-to-noise ratio of 47 dB. Control experiments based on polarization and wavelength detuning confirm that the measured signal arises from GMR-mediated transduction. These results establish guided-mode resonance as an on-chip approach to angular displacement readout in quantum optomechanical sensors.

main.pdf

Angular displacement readout of a mechanical oscillator with a guided mode resonance

Abstract

Precise angular displacement readout of a mechanical oscillator is a common task in science and engineering, from atomic force microscopy [1] to equivalence principle tests [2]. The traditional optical lever method [1–3] offers high sensitivity and quantum efficiency [4], but relies on free-space beam propagation over extended path lengths. Next-generation optomechanical sensors demand both sensitivity and compactness, motivating investigation of photonic structures sensitive to angular displacement. Guided mode resonance (GMR) structures [5–8] present a solution that combines sharp angular response and compatibility with nanofabrication methods used in quantum optomechanics [9–11]; however, to date GMR-based angle sensing has been limited to static [12] and low frequency [13] applications limited by classical noise.

Here we demonstrate use of GMR for shot-noise-limited angular displacement readout of a nanomechanical oscillator. We highlight the coherently-enhanced nature of the readout—a challenge for free-space approaches due the multimode nature of the optomechanical coupling [14]—which manifests in a conventional cavity optomechanical framework [15] as a shot-noise-limited angular displacement imprecision of

Sθimp≳κ28Gθ2hcλP=Δθ28N(1)S_{\theta}^{\mathrm{imp}} \gtrsim\frac{\kappa^2}{8G_{\theta}^2}\frac{hc}{\lambda P} = \frac{\Delta\theta^2}{8N} \tag*{(1)}

where κ\kappa and Gθ=∂ω/∂θG_{\theta} = \partial\omega/\partial\theta are the linewidth and optomechanical coupling to the resonance frequency ω0=c/λ0\omega_0 = c/\lambda_0, respectively, Δθ=κ/Gθ\Delta\theta= \kappa/G_{\theta} is the angular resonance linewidth, and PP (N=λ0P/hcN = \lambda_0P/hc) is the detected optical power (photon flux). By patterning a 25-nm-thick grating into a 100-nm-thick Si3_3N4_4 membrane, we realize a λ0≈1\lambda_0 \approx1 μm GMR with κ/(2π)∼1\kappa/(2\pi) \sim1 THz, G/(2π)∼100G/(2\pi) \sim100 GHz/mrad, and Δθ∼1\Delta\theta\sim1 mrad, enabling readout of a higher-order membrane mode with an imprecision of Sθ∼(1 rad/Hz)2S_{\theta} \sim(1\ \mathrm{rad}/\sqrt{\mathrm{Hz}})^2 using P∼100P \sim100 nW (N∼1011 s−1N \sim10^{11}\ \mathrm{s}^{-1}). We conclude by comparing the quantum limits of optical lever and GMR approaches and demonstrating thermal-noise-limited readout of a high-QQ torsion oscillator.

Guided-mode resonance - A brief overview of guided mode resonance (GMR) is germane to understanding our approach. GMRs arise in subwavelength grating slabs that support leaky waveguide modes [5–7]. When incident light is near the

resonance condition ω≈ω0(θ)\omega\approx\omega_0(\theta), it couples into the mode and interferes with the directly transmitted (or reflected) wave, resulting in a Fano-shaped transmission spectrum (Figure 1a inset)

Figure 1. Optomechanical coupling of a Si3_3N4_4 membrane to a guided mode resonance (GMR). (a) Membrane displacement modifies the local angle of incidence θ\theta of a laser beam incident on an embedded grating slab structure, altering the transmittivity TT of a GMR. Inset: Orientation of the grating relative to the mechanical mode. (b) Grating geometry: period Λ\Lambda, duty cycle DD, grating thickness dgd_g, waveguide thickness dwgd_{wg}. (c) Photo of 2.5 mm Si3_3N4_4 membrane with embedded grating. Inset: scanning electron micrograph of the grating.

T(ω,θ)≈T0∣1+q1+2i(ω−ω0[θ])/κ∣2(2)T(\omega,\theta) \approx T_0\left|1+\frac{q}{1+2i(\omega-\omega_0[\theta])/\kappa}\right|^2 \tag*{(2)}

where T0T_0 is the background transmittivity of the grating and qq is a (generally complex) Fano parameter. The GMR linewidth κ\kappa is governed by structural parameters such as the grating period, waveguide and grating thickness, and refractive index contrast, all of which can be finely tuned through simulation [16] and nanofabrication. These features have made GMRs attractive for a diversity of photonic-integrated sensing applications [5] from biology [16–18] to structural engineering [13].

Displacement readout of a mechanical oscillator is enabled by patterning a grating structure onto its surface and monitoring the transmitted (or reflected) field near a GMR resonance. In the limit of small displacement, the GMR can be modeled as sensitive to the local angular displacement

θ(x,y)=h^⋅∇u(x,y)(3)\theta(x,y) = \hat{h}\cdot\nabla u(x,y) \tag*{(3)}

where u(x,y)u(x,y) is the transverse displacement profile of the oscillator and h^\hat{h} is a unit vector normal to the grating. Monitoring a fraction of the transmitted power then yields a shot-noise-limited angular displacement sensitivity

Sθimp=2N(T∂T/∂θ)2=Δθ2T8Nβ(4)S_{\theta}^{\mathrm{imp}} = \frac{2}{N}\left(\frac{T}{\partial T/\partial\theta}\right)^2 = \frac{\Delta\theta^2 T}{8N\beta} \tag*{(4)}

where β=(Tθ′Δθ/4)2/T≤1\beta= (T'_{\theta}\Delta\theta/4)^2/T \le1 expresses the measurement efficiency per incident photon N/TN/T, which depends on the Fano lineshape and detuning (e.g., β≈0.42T0\beta\approx0.42T_0 and β/T≈1.7\beta/T \approx1.7 for a Lorentzian q=−1q=-1 detuned to maximize Tθ′T'_{\theta}).

To explore these concepts, we fabricated a 290×290 μm2290 \times290\ \mu\mathrm{m}^2 grating onto a 100-nm-thick, L=2.5L=2.5 mm wide square Si3N4\mathrm{Si}_3\mathrm{N}_4 membrane, following a standard e-beam lithography and wet-release method [19, 20]. Grating period Λ≈700\Lambda\approx700 nm, duty cycle D≈5/7D \approx5/7, and thickness dg≈25d_g \approx25 nm were chosen according to Rigorous Coupled-Wave Analysis simulations [21] to realize a GMR near normal incidence and λ0≈890\lambda_0 \approx890 nm. The grating was positioned near the membrane center (x0,y0)=(L/2,L/2)(x_0,y_0)=(L/2,L/2), enabling sensitivity to the antisymmetric second-order flexural mode u(x,y)=Asin⁡(πx/L)sin⁡(2πy/L)u(x,y)=A\sin(\pi x/L)\sin(2\pi y/L) with local angular displacement θ(x0,y0)≈2Aπ/L\theta(x_0,y_0)\approx2A\pi/L.

Optical characterization of the GMR device is shown in Figure 2. We first created a broadband hyperspectral map of the transmittance T(λ,θ)T(\lambda,\theta) using a collimated supercontinuum source and spectrometer, yielding qualitative agreement with model over visible wavelengths λ≈[500 nm,1000 nm]\lambda\approx[500\ \mathrm{nm},1000\ \mathrm{nm}] and incidence angles θ=[−4∘,4∘]\theta=[-4^\circ,4^\circ] (Figure 2b,Figure 2c). Higher resolution cross-sections were generated using a wavelength tunable diode laser, revealing a Fano resonance [22] with κ/(2π)≈0.41\kappa/(2\pi)\approx0.41 THz (0.95 nm) and Δθ≈2.5\Delta\theta\approx2.5 mrad at (λ0,θ0)≈(835 nm,6∘)(\lambda_0,\theta_0)\approx(835\ \mathrm{nm},6^\circ) (Figure 2d,Figure 2e). For these measurements, the laser beam was expanded to a diameter much larger than the grating, and the transmission of a patch of the grating was selected using a CCD camera as shown in the Figure 2e insets.

Figure 2. Optical characterization of guided mode resonance (GMR) device. (a) Experimental setup for measuring wavelength λ\lambda and incidence-angle θ\theta dependent membrane transmittance T(λ,θ)T(\lambda,\theta) using broadband (supercontinuum) and wavelength-tunable narrowband (external cavity diode) laser light. A half-wave plate (HWP) controls the laser polarization. Lens L1 forms a 4f system for spatial filtering with a pinhole. Lenses L2-L4 reimage the laser onto a CCD camera, spectrometer (Spec), or avalanche photodiode (APD). (b) Broadband measurement of T(λ,θ)T(\lambda,\theta) using the spectrometer, revealing several GMRs and avoided crossings. (c) Simulated and measured T(λ,θ)T(\lambda,\theta) at normal incidence, θ=0\theta= 0, for different polarizations. Insets: simulated GMR field profiles. (d) Narrowband measurement of T(λ,θ)T(\lambda,\theta) at fixed incidence angle using wavelength-scanned diode laser and CCD for readout. (e) Narrowband measurement of T(λ,θ)T(\lambda,\theta) at fixed wavelength using a gimbal mount to tilt the device and CCD for readout. Insets: CCD images at highlighted points. Red boxes indicated pinhole-apertured regions.

Displacement measurements were performed with the sample housed in a high-vacuum chamber (∼10−8\sim10^{-8} mbar) and the CCD replaced with a high-speed avalanche photodetector (Thorlabs APD210) and a 100-μ\mum-wide pinhole, enabling a local patch of the membrane to be monitored. While Q∼106Q\sim10^6 silicon nitride membranes with embedded crystal structures have been realized and studied [10, 23, 24], the device for this study exhibited Q≈5000Q\approx5000 for the targeted antisymmetric (2,1)(2,1) mode at f≈250f\approx250 kHz (Figure 1a inset), making thermal noise difficult to observe. We therefore performed driven measurements with an external piezo affixed to the vacuum chamber. Driven displacement measurements at frequencies around the near-degenerate (2,1)(2,1) and (1,2)(1,2) membrane modes are shown in Figure 3a, using a digital lock-in amplifier (Zurich Instruments MFLI) to record the photodetector output VV. To maximize sensitivity, the chip was tilted to near a resonance (θ0,λ0)(\theta_0,\lambda_0) at fixed wavelength; the wavelength was then tuned to the corresponding maximum slope ∂V/∂λ\partial V/\partial\lambda on Figure 2d, which is equivalent to the maximum slope ∂V/∂θ=∂V/∂λ⋅∂λ0/∂θ0\partial V/\partial\theta= \partial V/\partial\lambda\cdot\partial\lambda_0/\partial\theta_0 in Figure 2e, where ∂λ0/∂θ0=0.39\partial\lambda_0/\partial\theta_0 = 0.39 nm/mrad [25] is the slope of the resonance contour in Figure 2b. The driven response was converted to PSD units SVS_V by normalized by lock-in resolution bandwidth, and expressed in displacement units using Sθ=(∂V/∂θ)−2SV=(∂V/∂λ⋅∂λ0/∂θ0)−2SVS_\theta= (\partial V/\partial\theta)^{-2}S_V = (\partial V/\partial\lambda\cdot\partial\lambda_0/\partial\theta_0)^{-2}S_V.

Image not extracted

Figure 3. Angular displacement measurements and validation of optomechanical coupling. (a) Driven angular displacement spectra for various detected powers PP. The two peaks corresponding to degenerate mechanical modes of the square membrane. (b) Noise floor in raw voltage units versus PP, showing shot-noise scaling for P≳100P \gtrsim100 nW. (c) Driven angular displacement spectra for different linear polarizaion orientations of the incident field, varied with a rotatable half waveplate (HWP). (d) RMS displacement versus HWP angle. (e) Detector optical power (black) and RMS displacement versus probe laser wavelength. (f) Spectrograph of displacement spectra (panel a) versus probe laser wavelength.

For detected powers from 9 nW and 380 nW, we infer an angular displacement imprecision from 30 nrad/Hz\sqrt{\mathrm{Hz}} to 3 nrad/Hz\sqrt{\mathrm{Hz}}, respectively. The scaling of the raw noise floor (after subtracting detector noise) is consistent with shot-noise (Figure 3b). Comparing to Eq. (4) with a measured linewidth Δθ≈2.5\Delta\theta\approx2.5 mrad, we infer a measurement efficiency β≈0.05T\beta\approx0.05T, ∼3\sim3 times smaller than predicted by the lineshape in Figure 4e.

Image not extracted

Figure 4. GMR readout of a nanomechanical torsion oscillator. (a) Microscope image of the device: a 100-nm-thick, 0.4-mm-wide, 7-mm-long Si3_3N4_4 nanoribbon [30] with embedded grating similar to that in Figure 1. (b) Thermal-noise-limited readout of the fundamental torsion mode via GMR transmission and optical lever (OL) methods with varying detected power PP. (c) Detected power and raw signal variance (area under thermal noise peak) versus wavelength, used to optimize the signal-to-noise ratio in (b). (d) Simulated nanoribbon mechanical modeshapes. (e) Broadband measurements for the data in (b). Inset: Ringdown measurement of torsion mode.

As further confirmation of GMR optomechanical coupling, we repeated displacement measurements as a function of polarization (Figure 3c,Figure 3d) and wavelength (Figure 3e,Figure 3f) for the nominal operating point in Figure 3a. As anticipated, sensitivity was found to depend sinusoidally on input linear polarization angle, vanishing at an angle normal to the grating, where the GMR contrast is minimal. Tuning the laser wavelength likewise samples different local gradients of the 2D GMR resonance, T(λ,θ)T(\lambda,\theta). For a fixed θ\theta, the sensitivity Tθ′T'_\theta is expected to vanish near maxima of the wavelength gradient Tλ′=0T'_\lambda= 0. This

is corroborated by cross-sections of the driven displacement spectrograph Sθ(ω,λ)S_\theta(\omega,\lambda) shown in Figure 3e,Figure 3f.

Comparison to an optical lever- Our results show that a GMR structure can serve as an integrated cavity-enhanced angular displacement sensor. It is interesting to compare its performance to that of an optical lever (OL), whose shot-noise-limited displacement sensitivity can be expressed as [4]

Sθimp≥θD28N1βOL(5)S_\theta^{\mathrm{imp}} \geq\frac{\theta_D^2}{8N}\frac{1}{\beta_{\mathrm{OL}}} \tag*{(5)}

where θD=λ/(πw0)\theta_D = \lambda/(\pi w_0) is the diffraction angle of the reflected laser and βOL≤1\beta_{\mathrm{OL}} \leq1 is the measurement efficiency (for a split photodiode, βOL≤2/π\beta_{\mathrm{OL}} \leq2/\pi [4]). Comparing to Eq. (4) at equal incident power and background reflectivity R0=1−T0R_0 = 1-T_0, the GMR sensitivity can be recast as an effective OL spot size

w0=λπΔθββOLR0≈λ4Δθβ/T00.422/πβOLR0(6)w_0 = \frac{\lambda}{\pi\Delta\theta}\sqrt{\frac{\beta}{\beta_{\mathrm{OL}}R_0}} \approx\frac{\lambda}{4\Delta\theta}\sqrt{\frac{\beta/T_0}{0.42}\frac{2/\pi}{\beta_{\mathrm{OL}}R_0}} \tag*{(6)}

which for Δθ≈2.5\Delta\theta\approx2.5 mrad corresponds to λ/(4Δθ)≈85\lambda/(4\Delta\theta) \approx85 μm. Recent studies of quantum-limited OL measurements on nanomechanical oscillators [4, 26, 27] have employed similar spot sizes and realized similar measurement efficiencies βOL≈0.5\beta_{\mathrm{OL}} \approx0.5. A distinction of the GMR is that it is a local displacement probe. This may have significant advantages at low frequencies where apparatus and beam pointing noise are known to limit the sensitivity of OL measurements [27]. It can also enable robust free-space readout of chipscale sensors, by replacing sensitive alignment with wavelength tuning.

Displacement sensitivity limits- In our experiment, the achieved displacement imprecision Sθ≈(3 nrad/Hz)2S_\theta\approx(3\ \mathrm{nrad}/\sqrt{\mathrm{Hz}})^2 was limited by the throughput of our plane-wave imaging scheme, which allowed only η∼0.01%\eta\sim0.01\% of the optical power incident on the membrane (∼1 mW\sim1\ \mathrm{mW}) to reach the photodetector. In principle η∼1\eta\sim1 is achievable with an aperture-filling laser beam 2w0≈LGMR2w_0 \approx L_{\mathrm{GMR}} [28], implying that Sθimp<(0.1 nrad/Hz)2S_\theta^{\mathrm{imp}} < (0.1\ \mathrm{nrad}/\sqrt{\mathrm{Hz}})^2 is accessible with our current GMR linewidth Δθ≈2.5 mrad\Delta\theta\approx2.5\ \mathrm{mrad} and probe power. Achieving greater precision requires decreasing GMR linewidth and, or, increasing power. A practical lower bound is given by the diffraction limit Δθ≳λ/LGMR\Delta\theta\gtrsim\lambda/L_{\mathrm{GMR}} [28, 29], implying a LGMR=1 mmL_{\mathrm{GMR}} = 1\ \mathrm{mm} grating probed with P=10 mWP = 10\ \mathrm{mW} should enable Sθimp<(0.01 nrad/Hz)2S_\theta^{\mathrm{imp}} < (0.01\ \mathrm{nrad}/\sqrt{\mathrm{Hz}})^2, on par with the state-of-the-art OL measurements [4, 26, 27].

Nanomechanical torque sensing- A promising application of GMR-based angular displacement sensing is readout of a nanomechanical torsion oscillator for precision torque sensing [31, 32]. Recently, centimeter-scale Si3_3N4_4 nanoribbons have been reported with ultrahigh quality factors Q∼108Q \sim10^8 at acoustic frequencies f0∼100 kHzf_0 \sim100\ \mathrm{kHz}, yielding thermal torque sensitivities of Sτth∼(10 zNm/Hz)2S_\tau^{\mathrm{th}} \sim(10\ \mathrm{zNm}/\sqrt{\mathrm{Hz}})^2 and resonant thermal displacements of Sθth(f0)∼(μrad/Hz)2S_\theta^{\mathrm{th}}(f_0) \sim(\mu\mathrm{rad}/\sqrt{\mathrm{Hz}})^2 [30, 33]. Embedding a GMR into such a nanoribbon could enable zNm/Hz\mathrm{zNm}/\sqrt{\mathrm{Hz}} torque sensitivity with picowatts of optical power. We present progress towards this goal in Figure 4, by fabricating a grating similar to that in Figure 1 into a w=400 μmw = 400\ \mu\mathrm{m}-wide, h=100 nmh = 100\ \mathrm{nm}-thick, 7-mm-long Si3_3N4_4 nanoribbon [30]. Broadband displacement measurements shown in Figure 4b were made by probing the side of a GMR with P≈0.6 μWP \approx0.6\ \mu\mathrm{W}, revealing a series of thermal noise peaks, including a Q≈1.5×106Q \approx1.5 \times10^6 noise peak at f0=52.5 kHzf_0 = 52.5\ \mathrm{kHz}, corresponding to the fundamental torsion mode. Bootstrapping a Lorentzian fit to the simulated peak thermal displacement Sθth(f0)≈(0.37 μrad/Hz)2S_\theta^{\mathrm{th}}(f_0) \approx(0.37\ \mu\mathrm{rad}/\sqrt{\mathrm{Hz}})^2, we infer a readout imprecision of Sθimp≈(1.6 nrad/Hz)2S_\theta^{\mathrm{imp}} \approx(1.6\ \mathrm{nrad}/\sqrt{\mathrm{Hz}})^2, qualitatively consistent with the direct calibration in Figure 3a. Independent of calibration, the high signal-to-noise ratio Sθth[f0]/Sθimp≈47 dBS_\theta^{\mathrm{th}}[f_0]/S_\theta^{\mathrm{imp}} \approx47\ \mathrm{dB} corresponds to realizing a thermal torque sensitivity of Sτth≈(0.1 aNm/Hz)2S_\tau^{\mathrm{th}} \approx(0.1\ \mathrm{aNm}/\sqrt{\mathrm{Hz}})^2 [30] over a bandwidth of Δf=(f0/Q)(Sθth[f0]/Sθimp)1/2≈8 Hz\Delta f = (f_0/Q)(S_\theta^{\mathrm{th}}[f_0]/S_\theta^{\mathrm{imp}})^{1/2} \approx8\ \mathrm{Hz}.

Summary & outlook - We have explored optomechanical coupling to a guided mode resonance (GMR) as a method for coherently-enhanced angular displacement readout of a mechanical oscillator [14], comparing its quantum limits to that of the optical lever [4], and conducting proof-of-principle experiments with Si3_3N4_4 membranes patterned with subwavelength gratings. Using direct transmission measurements, we realized nrad/Hz\mathrm{nrad}/\sqrt{\mathrm{Hz}} sensitivity with nanowatts of optical power, limited by a diffraction-limited angular GMR linewidth Δθ≈2.5 mrad\Delta\theta\approx2.5\ \mathrm{mrad}, and corresponding to a O(1)\mathcal{O}(1) quantum efficiency. We applied this displacement sensitivity to readout of a high-QQ Si3_3N4_4 torsion nanoribbon, resolving thermal noise with a 47 dB signal-to-noise ratio, corresponding to a 0.1 aNm/Hz\sqrt{\mathrm{Hz}} torque sensitivity over an 8 Hz bandwidth.

Looking forward, combining low loss and lithographic control over the optomechanical properties of suspended GMR structures offers diverse opportunities. Hybrid GMR-Fabry-Perot cavities have been widely explored [34, 35] and offer a route to enhanced optomechanical interactions [36, 37]. Our focus has been on direct GMR optomechanical coupling to angular displacement, reducing engineering complexity and avoiding constraints posed by the nondegeneracy of Fabry–Perot spatial modes [14]. A promising application is optomechanical inertial sensing [38], which requires integrated displacement readout methods that are both sensitive and robust to parasitic vibrations, and for which chip-scale torsion oscillators have emerged as an attractive platform [39]. GMR optomechanical coupling is also not limited to angular displacement [40]. Matching the local grating orientation to the gradient of the mechanical modeshape can enable selective coupling to specific mechanical modes, as well as tailored intra- and intermodal coupling between transverse spatial modes of the optical field [41]. In conjunction with high-QQ nanomechanics [42], this approach could provide a general framework for resonant spatial-mode optomechanics.

Acknowledgements

The authors thank Atkin Hyatt and Justin Murante for helpful discussions and experimental assistance. This work was supported by the National Science Foundation (NSF), award no. 2209473 and by the Office of Naval Research MURI, award no. N000142612102. D.T.B. acknowledges additional support from the Friends of Tucson Optics scholarship. O.E.A. acknowledges support from the Secretaría de Ciencia, Humanidades, Tecnología e Innovación (SECIHTI), Mexico (CVU 2023435). The reactive ion etcher used for this study was funded by an NSF MRI grant, ECCS-1725571.

References

Supplemental Information for “Angular displacement readout of a mechanical oscillator with a guided mode resonance”

D. Torres-Barajas; O. Angulo; A. R. Agrawal; M. ElKabbash; D. J. Wilson

SI.pdf

Here we provide details about the method used to a fabricate Si3N4\mathrm{Si_3N_4} nanoribbon with embedded subwavelength grating shown in Figure 4 of the main text.

Fabrication of Si3N4\mathrm{Si_3N_4} membranes with embedded gratings

A photograph of the device described in Figure 4 of the main text—a Si3N4\mathrm{Si_3N_4} nanoribbon with embedded subwavelength grating—is shown in Figure S1. The process flow for fabrication of this device is shown in Figure S2, and is identical to that used for the patterned membrane described in main text Figure 1–Figure 3. Details are provided below.

Image not extracted

Figure S1. Photograph of Si3N4\mathrm{Si_3N_4} nanoribbon with embedded subwavelength grating after KOH etching.

Image not extracted

Figure S2. Process flow. (a) Starting chip. (b) Spin-coated electron beam resist. (c) Written grating pattern on the resist. (c) Pattern is transferred to the Si3N4\mathrm{Si_3N_4} by dry etching. (e) Resist residue is cleaned. (f) Device is released by wet etching.

Patterning of subwavelength grating

The fabrication procedure starts with a 100-nm-thick Si3N4\mathrm{Si_3N_4} membrane or nanoribbon prepatterned on a 200-μm\mu\mathrm{m}-thick Si chip with a backwindow (Figure S2), following the procedure described in [1]. The grating is patterned using electron beam lithography (EBL, Elionix ELS-7000) and exposed using ZEP520A as a resist (Figure S2). For alignment of the grating pattern on the membrane, the stage coordinates of the corners of the chip were used (found using the SEM function of the EBL machine). Exposure parameters are summarized in Table 1.

Beam current300 pADots per field240000
Aperture60 μm\mu\mathrm{m}Dose390 μC/cm2\mu\mathrm{C}/\mathrm{cm}^2
Field size600 μm\mu\mathrm{m}Resist thickness400 nm

Table 1. Electron-beam exposure parameters.

Pattern transfer and release

After the exposure, the pattern was developed by immersion in Amyl Acetate for 3 minutes, followed by an IPA rinse (Figure S2). The pattern is transferred to the Si3N4\mathrm{Si_3N_4} using RIE with CHF3\mathrm{CHF_3}, in this case we etched 20 nm deep into the Si3N4\mathrm{Si_3N_4} film (Figure S2). The e-beam resist is then cleaned using 2 consecutive immersions in NMP at 80°C for 5 min each, followed by water and IPA rinses (Figure S2). The release is then performed in a 30% KOH solution at 80°C (Figure S2).

References

  1. [34]O. Černotík, A. Dantan, and C. Genes, Cavity quantum electrodynamics with frequency-dependent reflectors, Physical Review Letters 122, 243601 (2019).DOI
  2. [35]J. M. Fitzgerald, S. K. Manjeshwar, W. Wieczorek, and P. Tassin, Cavity optomechanics with photonic bound states in the continuum, Physical Review Research 3, 013131 (2021).DOI
  3. [36]J. Monsel, A. Ciers, S. K. Manjeshwar, W. Wieczorek, and J. Splendsteasser, Dissipative and dispersive cavity optomechanics with a frequency-dependent mirror, Physical Review A 109, 043532 (2024).DOI
  4. [37]T. Mitra, G. Singh, S. P. Madsen, and A. Dantan, Cavity optomechanics with a suspended resonant mirror, arXiv preprint arXiv:2608.02488 (2026).arxiv.org/abs/2608.02488
  5. [38]X. Liu, W. Liu, Z. Ren, Y. Ma, B. Dong, G. Zhou, and C. Lee, Progress of optomechanical micro/nano sensors: a review, International Journal of Optomechatronics 15, 120 (2021).DOI
  6. [39]C. Condos, J. Pratt, J. Manley, A. Agrawal, S. Schlaminger, C. Pluchar, and D. Wilson, Ultralow loss torsion micropendula for chipscale gravimetry, Phys. Rev. Lett. 134, 253602 (2025).DOI
  7. [40]S. Zanotto, G. Conte, L. Bellieres, A. Griol, D. Navarro-Urrios, A. Tredicucci, A. Martínez, and A. Pitanti, Optomechanical modulation spectroscopy of bound states in the continuum in a dielectric metasurface, Physical Review Applied 17, 044033 (2022).DOI
  8. [41]C. M. Pluchar, W. He, J. Manley, N. Deshler, S. Guha, and D. J. Wilson, Imaging-based quantum optomechanics, Physical Review Letters 135, 023601 (2025).DOI
  9. [42]N. J. Engelsen, A. Beccari, and T. J. Kippenberg, Ultrahigh-quality-factor micro- and nanomechanical resonators using dissipation dilution, Nature Nanotechnology 19, 725 (2024).DOI

Paper details

vlnc:W_27ybmahk
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents