Article Cite This: ACS Photonics XXXX, XXX, XXX−XXX
Optomechanical Cavities for All-Optical Photothermal Sensing Marcel W. Pruessner,* Doewon Park, Todd H. Stievater, Dmitry A. Kozak, and William S. Rabinovich U.S. Naval Research Laboratory, 4555 Overlook Avenue SW, Washington, D.C. 20375-0001, United States
Downloaded via NAGOYA UNIV on June 26, 2018 at 18:49:16 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
S Supporting Information *
ABSTRACT: Cavity optomechanics enables strong coupling of optics and mechanics. Although remarkable progress has been made, practical applications of cavity optomechanics are only recently being realized. In this work we propose an alloptical sensing technique enabling the measurement of photothermally induced strains with ultrahigh-resolution. We demonstrate an optomechanical sensor consisting of a silicon nitride (Si3N4) microring cavity that is evanescently coupled to a suspended SiNx micromechanical (MEMS) oscillator. Experiments show that MEMS resonances are excited purely via cavity-enhanced gradient optical forces. However, small levels of absorption in the oscillator result in photothermally induced strains that shift the mechanical resonance frequencies. By measuring absorption-induced frequency shifts our technique enables high-resolution with nanostrain sensitivity corresponding to fJ-levels of absorption. As a demonstration, we perform absorption spectroscopy on the MEMS device and measure the known Si−H absorption feature of deposited silicon nitride. The unprecedented sensitivity enabled by absorption-induced frequency shifts enables entirely new sensors in fields ranging from materials and chemical sensing to bolometers and imaging arrays. KEYWORDS: optomechanics, micromechanical systems (MEMS), optical forces, microring cavities, absorption spectroscopy, photothermal sensors
C
avity optomechanics,1−4 in which micromechanical structures (MEMS) are coupled to optical cavities, is an interdisciplinary field that is finding application in many exciting areas ranging from fundamental studies to practical devices. Initial demonstrations showed that the mutual interaction between the optical field in a cavity and the motion of a mechanical oscillator can result in an increase or decrease in the effective spring constant.2 This can result in cooling of the MEMS oscillator3,5 down to its quantum ground state6 as well as driven (coherent) oscillations using a CW laser pump.1,3,4,7 Further studies have shown that optical forces can also result in frequency tuning of the MEMS oscillator.2,5,8 More recently, there has been an increased interest in applications of optical forces in optomechanical systems. These include optically reconfigurable filters,9 high-resolution optical accelerometers,10 fully integrated cavity optomechanical atomic force microscope (AFM) probes,11 and all-optical chemical sensors based on mass-loading.12 Besides cavity optomechanics, the broader MEMS field has also made extraordinary advances in sensing. Of particular interest are sensors that can detect small temperature shifts. Such sensors are versatile and can find application in chemical detection (i.e., absorption spectroscopy) as well as in imaging (i.e., bolometry). While sensitive temperature shifts can be measured via the absorption-induced loss in nanophotonic waveguides,13,14 such approaches are limited by the spectral bandwidth of the waveguide material as well as the nanophotonic architecture used (e.g., microrings). To circum© XXXX American Chemical Society
vent this bandwidth limitation, MEMS structures, whose mechanical dynamics are modified by temperature, can be used.15,16 In this manner, the absorber is separated from the readout device. The MEMS structure consists of a broadband absorber in the form of a metal-coated dielectric cantilever serving as a bimorph actuator. The MEMS structure converts incident radiation into heat, followed by a thermal strain that induces a deflection that can be measured optically. The displacement measurement can use position-sensitive photodetectors15 or microcavity interferometry16 to enhance the displacement resolution. Such micromechanical photothermal absorption spectroscopy has proven to be highly sensitive demonstrating pJ heat sensing resolution (with fJ-levels predicted).15 At the same time, the approach can lead to highly specific chemical sensing capabilities since the measured absorption spectra uniquely identify individual chemicals.16,17 To date, however, the absorption spectrum readout mechanism has focused primarily on static deflection of MEMS cantilevers15−18 in which the displacement resolution is limited by thermal-mechanical displacement noise. Potentially higher sensing resolution can be obtained by considering absorption-induced changes in the mechanical resonance. In this manner, the MEMS oscillator can be driven to large displacement amplitudes, far greater than the thermalmechanical displacement noise, thereby enabling precise Received: April 9, 2018
A
DOI: 10.1021/acsphotonics.8b00452 ACS Photonics XXXX, XXX, XXX−XXX
Article
ACS Photonics
power can originate externally to the device (Figure 1a), for example, as in bolometers, or internally (Figure 1b), as might occur in absorption spectroscopy. We demonstrate the utility of our approach by measuring the Si−H absorption peak of the silicon nitride (SiNx) MEMS structure and show that the approach enables the measurement of ultrasmall heat with potential for fJ-level radiation detection. This new type of absorption spectroscopy technique and sensor platform is waveguide-integrated and can be arrayed on-chip (each high-Q cavity requires minimal power), is broadband (the spectral bandwidth is determined by the MEMS, not by the micoring cavity), does not require on-chip power (all-optical actuation and measurement), and may find application in a variety of areas ranging from chemical sensing based on absorption spectroscopy16 to bolometry20 and imaging.
■
EXPERIMENTAL RESULTS AND DISCUSSION The optomechanical cavity sensor shown in Figure 2a enables all-optical actuation and readout of a MEMS oscillator whose resonant frequency is modified by small absorption of incident radiation. By focusing on MEMS resonant frequency shifts as opposed to MEMS deflection,15,16 broadband mechanical frequency noise can be minimized. The device consists of a microring optical cavity fabricated in 175 nm thick silicon nitride (Si3N4) on 5 μm thermal oxide (SiO2) with air top cladding. The rib waveguides and fabrication process are similar to previous work23 with hrib = 105 nm (i.e., 105 nm etch) and wrib = 2.5 μm (waveguide width). Suspended above a straight segment of the microring is a SiNx MEMS microbridge oscillator with length lMEMS = 30−140 μm, thickness tMEMS = 250 nm, and width wMEMS = 3 μm (Figure 2b). The separation between the MEMS and waveguide is set by the sacrificial oxide layer (tSiO2 = 465 nm) and ensures strong evanescent field interaction (Figure 2a, inset), resulting in dispersive coupling and large dneff/dgap (i.e., large change in the optical mode’s effective index, neff, with changes in separation or gap), while ensuring that the MEMS perturbation does not introduce substantial optical loss in the microring.23 By locating the MEMS structure above the microring, we can control the optomechanical coupling by adjusting the MEMSwaveguide separation via the deposited SiO2 sacrificial layer.23 We measured the optomechanical cavity spectrum for an lMEMS = 120 μm device (Figure 2c) and extract a peak Qoptical = 80000 and finesse F = 35 for transverse electric (TE) polarized light (Figure 2d). Other measured devices showed similar performance over wavelengths λ = 1370−1570 nm. Next, we performed pulsed pump−probe measurements to excite and measure the MEMS resonances (Figure 3a). The sample was placed in a custom vacuum cell that is pumped down to modest pressures ( 1520 nm the frequency shifts back toward higher frequencies. We plot the extracted mechanical resonance frequency versus λpump in Figure 6b and also plot the measured PECVD-SiNx microring loss spectrum for comparison in Figure 6c. The optomechanical cavity pump−probe resonance frequency spectrum clearly shows the λ = 1520 nm absorption peak, suggesting that photothermally induced frequency shifts can serve as a sensitive transduction mechanism for absorption spectroscopy in optomechanical cavities. Besides the localized absorption feature at λpump = 1520 nm the frequency spectra also show a linear decrease toward shorter wavelengths. We attribute this to the combination of broadband absorption in the PECVD-SiNx material and the higher finesse at shorter wavelengths (see Figure 2d). The linear frequency shift versus λpump follows the cavity finesse indicating that the broadband frequency tuning (i.e., linear tuning component over λpump = 1470−1570 nm) results from the variation in microring circulating optical power due to the wavelength-dependent cavity finesse. Looking at the λpump = 1500 nm measurement (red arrow, Figure 6b) we obtain a frequency shift Δf = 150 Hz, similar to the frequency shift measured previously in Figure 5a (Ppump = 1 mW). From the Δf we extract an absorbed power of 45 nW (see Figure 5b) and an absorption coefficient α = 2.1 dB/cm, in general agreement with the PECVD loss and absorption measurement (Δα) in Figure 6c. We attribute any discrepancy to a variety of factors including variation in power coupling during the measurement compared to previous power calibration, uncertainty in the modal overlap with the MEMS structure, and variation in material property coefficients (especially the coefficient of thermal expansion, cTE,SiNx) in
Figure 5. Frequency tuning: (a) Measured mechanical resonance frequency shift (Δf) vs pump power with λpump ≈ 1500 nm and λprobe ≈ 1400 nm for the lMEMS = 120 μm optomechancial cavity (see measurement in Figure 4); inset: detail of resonance vs Ppump near f 0 = 1.298 MHz. (b) Calculated mechanical resonance frequency shift due to gradient optical forces (black line) and photothermal absorption (red line); inset: simulated fundamental mechanical resonance frequency for an lMEMS = 120 μm structure vs intrinsic film stress (σ0); for σ0 = 300 MPa the stress-dependent resonance frequency shift is Δf/Δσ0 = 2186 Hz/MPa.
0.17 J/g·K, the MEMS mass is m, the thermal time constant is τthermal = 100 μs (see Supporting Information), and Pabsorbed is the absorbed optical power. We estimate the absorbed optical power assuming an absorption coefficient α = 1.5 dB/cm, which is in general agreement with optical loss measurements performed on PECVD deposited SiNx waveguides (see Supporting Information) and the on-resonance circulating power enhancement in the microring. Using this approach, we calculate the frequency shift versus optical power in the microring. The calculated frequency tuning (Figure 5b) shows good agreement with measurements (Figure 5a), as evidenced by the frequency tuning slope as a function of pump power (experiment: −112 Hz/mW, photothermal model: −106 Hz/ mW). Importantly, photothermal absorption results in compressive strain (i.e., Δε > 0) that reduces the total tensile strain (i.e., εtotal < 0) in the MEMS structure resulting in a E
DOI: 10.1021/acsphotonics.8b00452 ACS Photonics XXXX, XXX, XXX−XXX
Article
ACS Photonics
on MEMS cantilever deflection.15 The increased sensitivity is attributed to our dynamic measurement approach, in which changes in MEMS frequency are measured as opposed to absolute displacement. Although Allan deviation measurements show that the frequency resolution is 1 Hz at long measurement integration times (Supporting Information), future improvements including actuation to larger displacement amplitudes and the potential use of feedback28 to minimize thermal-mechanical noise may improve the heat sensing resolution down to the 100 fJ-level with short integration times of the order of the thermal time constant. We emphasize that the MEMS structures have not been optimized for strain sensing, and a number of improvements can be made. The MEMS SiNx material, while exhibiting a localized absorption peak near λ = 1520 nm is generally transparent and low-loss in the near-IR. Higher absorption can be attained using, for example, silicon-rich SiNx or by depositing a thin absorptive layer on the MEMS surface. Alternatively, plasmonic structures24 can be used to enable greater than 90% absorption,29 thereby enhancing thermalmechanical transduction. Finally, the addition of metal to the SiNx surface leads to bimorph structures that enhance the thermally induced MEMS displacement, similar to previous structures.15,16 These improvements in both optical (plasmonic absorbers) and mechanical (bimorph actuators) domains suggest a two-orders of magnitude (or higher) improvement is possible enabling fJ-level absorption measurement (pW-sensitivity) with optomechanical cavities. The cavity optomechanical sensor presented here embodies a new approach for photothermal sensing. It is all-optical and relies on optical forces for device actuation and on interferometric (dispersive) readout. As such, our sensor does not require any electrical power on-chip and may find application where stand-off sensors are required. The nanophotonic waveguide architecture lends itself to large-scale integration such that many sensors can be arrayed and interrogated, either using wavelength-multiplexing (microring) or frequency-multiplexing (MEMS). Since the sensor relies on measuring frequency shifts as opposed to absolute frequency, small variations in temperature are not an issue and the sensor resonant frequency can be self-calibrated. Finally, the MEMS structure can be optimized for broadband absorption and maximum thermal-mechanical transduction without affecting the microring cavity optical performance. We expect this optomechanical cavity sensor to find application in a variety of areas ranging from chemical sensing to bolometers and imaging arrays.
Figure 6. Resonance cavity optomechanical absorption spectroscopy: (a) Measured mechanical resonance spectra vs pump wavelength over the range λpump ≈ 1470−1570 nm and λprobe ≈ 1400 nm for the lMEMS = 120 μm optomechancial cavity; the black line is a guide to emphasize the resonance peak at specific λpump. (b) Extracted frequency shift vs pump wavelength showing a clear absorption feature near λpump ≈ 1520 nm. (c) Loss spectrum from a microring cavity fabricated in PECVD silicon nitride (no MEMS) showing a clear peak near λ ≈1520 nm due to Si−H absorption.
the fabricated structures compared to those assumed in our photothermal model. Nonetheless, the results in Figure 6 confirm that optomechanical cavities can serve as sensitive alloptical transducers of absorbed radiation and can also enable absorption spectroscopy for a variety of applications. Simulations (Figure 5b) showed that the strain sensitivity is Δε/Δf ≈ 1.8 × 10−9/Hz, that is, nanostrain per Hz frequency shift. The Allan deviation in our frequency shift measurement is around 1 Hz (see Supporting Information), so the strain sensing resolution is Δε/Δf = 2.3 × 10−9/Hz per √Hz (1601 points with 1 kHz measurement bandwidth). This is significantly better than the microstrain sensing resolution typical in commercial strain sensors. The corresponding temperature increase is ΔT ≈ 2 mK. We extract the corresponding theoretical heat sensing resolution as δQheat = ΔT × cp × m ≈ 100 fJ (i.e., 1 nW absorbed power for a 100 μs thermal time constant), which exceeds the pJ-levels of previously demonstrated photothermal spectroscopy based
■
METHODS Thermal-Mechanical Displacement Noise Measurement and Calibration. The driven (i.e., pump−probe) displacement amplitude was calibrated using equipartition. We use the setup in Figure 3a; however, the network analyzer (NA) is replaced with an electrical spectrum analyzer (ESA) and only the CW probe laser is used. The thermal-mechanical displacement noise spectrum is measured in the ESA (at room temperature, same as the pump−probe measurements). The thermal-mechanical displacement noise in the absence of a driving pump laser is calculated30 as 4k T Δf ω0 znoise 2 = m bQ 2 2 2 2 . The MEMS effective eff
mech
(ω0 − ω ) + (ω ω0 / Q mech )
mass depends on the mechanical mode and is meff = 0.73m0 for the fundamental mode;31 we note that the meff and hence F
DOI: 10.1021/acsphotonics.8b00452 ACS Photonics XXXX, XXX, XXX−XXX
Article
ACS Photonics
(10) Krause, A. G.; Winger, M.; Blasius, T. D.; Lin, Q.; Painter, O. A high-resolution microchip optomechanical accelerometer. Nat. Photonics 2012, 6 (11), 768−772. (11) Chae, J.; An, S.; Ramer, G.; Stavila, V.; Holland, G.; Yoon, Y.; Talin, A. A.; Allendorf, M.; Aksyuk, V. A.; Centrone, A. Nanophotonic Atomic Force Microscope Transducers Enable Chemical Composition and Thermal Conductivity Measurements at the Nanoscale. Nano Lett. 2017, 17 (9), 5587−5594. (12) Stievater, T. H.; Rabinovich, W. S.; Ferraro, M. S.; Papanicolaou, N. A.; Bass, R.; Boos, J. B.; Stepnowski, J. L.; McGill, R. A. Photonic microharp chemical sensors. Opt. Express 2008, 16 (4), 2423−2430. (13) Stievater, T. H.; Pruessner, M. W.; Park, D.; McGill, R. R. A.; Kozak, D. A.; Furstenberg, R.; Holmstrom, S. A.; Khurgin, J. B. Trace gas absorption spectroscopy using functionalized microring resonators. Opt. Lett. 2014, 39 (4), 969−972. (14) Nitkowski, A.; Chen, L.; Lipson, M. Cavity-enhanced on-chip absorption spectroscopy using microring resonators. Opt. Express 2008, 16 (16), 11930−11936. (15) Barnes, J. R.; Stephenson, R. J.; Welland, M. E.; Gerber, C.; Gimzewski, J. K. Photothermal Spectroscopy with Femtojoule Sensitivity Using a Micromechanical Device. Nature 1994, 372 (6501), 79−81. (16) Stievater, T. H.; Papanicolaou, N. A.; Bass, R.; Rabinovich, W. S.; McGill, R. A. Micromechanical photothermal spectroscopy of trace gases using functionalized polymers. Opt. Lett. 2012, 37 (12), 2328− 2330. (17) Lee, D.; Chae, I.; Kwon, O.; Lee, K. H.; Kim, C.; Kim, S.; Thundat, T. Plasmonic absorbers with optical cavity for the enhancement of photothermal/opto-calorimetric infrared spectroscopy. Appl. Phys. Lett. 2017, 110 (1), 011901. (18) Belacel, C.; Todorov, Y.; Barbieri, S.; Gacemi, D.; Favero, I.; Sirtori, C. Optomechanical terahertz detection with single meta-atom resonator. Nat. Commun. 2017, 8 (1), na. (19) Chen, G. Y.; Thundat, T.; Wachter, E. A.; Warmack, R. J. Adsorption-Induced Surface Stress and Its Effects on Resonance Frequency of Microcantilevers. J. Appl. Phys. 1995, 77 (8), 3618− 3622. (20) Laurent, L.; Yon, J.-J.; Moulet, J.-S.; Roukes, M.; Duraffourg, L. 12-μm-Pitch Electromechanical Resonator for Thermal Sensing. Phys. Rev. Appl. 2018, 9 (2), 024016. (21) Povinelli, M. L.; Loncar, M.; Ibanescu, M.; Smythe, E. J.; Johnson, S. G.; Capasso, F.; Joannopoulos, J. D. Evanescent-wave bonding between optical waveguides. Opt. Lett. 2005, 30 (22), 3042− 3044. (22) Li, M.; Pernice, W. H. P.; Xiong, C.; Baehr-Jones, T.; Hochberg, M.; Tang, H. X. Harnessing optical forces in integrated photonic circuits. Nature 2008, 456 (7221), 480−U28. (23) Pruessner, M. W.; Park, D.; Stievater, T. H.; Kozak, D. A.; Rabinovich, W. S. Broadband opto-electro-mechanical effective refractive index tuning on a chip. Opt. Express 2016, 24 (13), 13917−13930. (24) Roxworthy, B. J.; Aksyuk, V. A. Electrically tunable plasmomechanical oscillators for localized modulation, transduction, and amplification. Optica 2018, 5 (1), 71−79. (25) Albrecht, T. R.; Grutter, P.; Horne, D.; Rugar, D. FrequencyModulation Detection Using High-Q Cantilevers for Enhanced Force Microscope Sensitivity. J. Appl. Phys. 1991, 69 (2), 668−673. (26) Thijssen, R.; Kippenberg, T. J.; Polman, A.; Verhagen, E. Parallel Transduction of Nanomechanical Motion Using Plasmonic Resonators. ACS Photonics 2014, 1 (11), 1181−1188. (27) Henry, C. H.; Kazarinov, R. F.; Lee, H. J.; Orlowsky, K. J.; Katz, L. E. Low-Loss Si3n4-Sio2 Optical Wave-Guides on Si. Appl. Opt. 1987, 26 (13), 2621−2624. (28) Gavartin, E.; Verlot, P.; Kippenberg, T. J. Stabilization of a linear nanomechanical oscillator to its thermodynamic limit. Nat. Commun. 2013, 4 (1), na. (29) Zhou, L.; Tan, Y. L.; Ji, D. X.; Zhu, B.; Zhang, P.; Xu, J.; Gan, Q. Q.; Yu, Z. F.; Zhu, J. Self-assembly of highly efficient, broadband
thermal-mechanical displacement noise model is limited by the uncertainty in the assumed mass density. By comparing the measured noise spectrum (μV) and calculated znoise (pm) we can convert our measured driven MEMS oscillations to an absolute displacement amplitude (Figure 4a).
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsphotonics.8b00452. Additional discussion of device fabrication, insertion loss, pump−probe measurement, resonant frequency measurement resolution and Allan variance, equipartition and thermal-mechanical displacement noise, thermal time constant, SiNx absorption, signal-to-noise ratio (SNR), and additional discussion of photothermal forces (PDF).
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. ORCID
Marcel W. Pruessner: 0000-0002-6552-5057 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors acknowledge support for this work by a U.S. Naval Research Laboratory (USNRL) base 6.1 program. They also thank L. D. Epp for machining the vacuum cell and other components, R. Mahon for assistance with the experimental setup, and B. J. Roxworthy for helpful discussions regarding this work.
■
REFERENCES
(1) Zalalutdinov, M.; Zehnder, A.; Olkhovets, A.; Turner, S.; Sekaric, L.; Ilic, B.; Czaplewski, D.; Parpia, J. M.; Craighead, H. G. Autoparametric optical drive for micromechanical oscillators. Appl. Phys. Lett. 2001, 79 (5), 695−697. (2) Vogel, M.; Mooser, C.; Karrai, K.; Warburton, R. J. Optically tunable mechanics of microlevers. Appl. Phys. Lett. 2003, 83 (7), 1337−1339. (3) Metzger, C. H.; Karrai, K. Cavity cooling of a microlever. Nature 2004, 432 (7020), 1002−1005. (4) Kippenberg, T. J.; Rokhsari, H.; Carmon, T.; Scherer, A.; Vahala, K. J. Analysis of radiation-pressure induced mechanical oscillation of an optical microcavity. Phys. Rev. Lett. 2005, 95 (3), na. (5) Arcizet, O.; Cohadon, P. F.; Briant, T.; Pinard, M.; Heidmann, A. Radiation-pressure cooling and optomechanical instability of a micromirror. Nature 2006, 444 (7115), 71−74. (6) Chan, J.; Alegre, T. P. M.; Safavi-Naeini, A. H.; Hill, J. T.; Krause, A.; Groblacher, S.; Aspelmeyer, M.; Painter, O. Laser cooling of a nanomechanical oscillator into its quantum ground state. Nature 2011, 478 (7367), 89−92. (7) Khurgin, J. B.; Pruessner, M. W.; Stievater, T. H.; Rabinovich, W. S. Laser-Rate-Equation Description of Optomechanical Oscillators. Phys. Rev. Lett. 2012, 108 (22), na. (8) Pruessner, M. W.; Stievater, T. H.; Khurgin, J. B.; Rabinovich, W. S. Integrated waveguide-DBR microcavity opto-mechanical system. Opt. Express 2011, 19 (22), 21904−21918. (9) Deotare, P. B.; Bulu, I.; Frank, I. W.; Quan, Q. M.; Zhang, Y. N.; Ilic, R.; Loncar, M. All optical reconfiguration of optomechanical filters. Nat. Commun. 2012, 3, na. G
DOI: 10.1021/acsphotonics.8b00452 ACS Photonics XXXX, XXX, XXX−XXX
Article
ACS Photonics plasmonic absorbers for solar steam generation. Sci. Adv. 2016, 2 (4), e1501227. (30) Stievater, T. H.; Rabinovich, W. S.; Papanicolaou, N. A.; Bass, R.; Boos, J. B. Measured limits of detection based on thermalmechanical frequency noise in micromechanical sensors. Appl. Phys. Lett. 2007, 90 (5), 051114. (31) Ekinci, K. L.; Roukes, M. L. Nanoelectromechanical systems. Rev. Sci. Instrum. 2005, 76 (6), 061101.
H
DOI: 10.1021/acsphotonics.8b00452 ACS Photonics XXXX, XXX, XXX−XXX
