Patent Grant
8249400
1. Field of the Invention
This application relates generally to sensor systems, and more particularly to optical-fiber-compatible sensor systems.
2. Description of the Related Art
Fiber-optic acoustic sensors have been extensively researched since their first emergence in the 1970s (see, e.g., J. H. Cole, R. L. Johnson, and P. G. Bhuta, “Fiber-optic detection of sound,” J. Acoust. Soc. Am. Vol. 62, 1136 (1977); J. A. Bucaro, H. D. Dardy, and E. F. Carome, “Fiber-optic hydrophone,” J. Acoust. Soc. Am. Vol. 62, 1302 (1977)), largely because of their multiple advantages over conventional acoustic sensors. These advantages include small size and weight, ability to operate in chemically and/or electrically harsh environments, ease of multiplexing large numbers of sensors together, and compatibility with fiber-optic networks for data transport and processing.
Various forms of biological, chemical, and mechanical sensors (such as acoustic or pressure sensors) that can be addressed optically at the end of an optical fiber can be useful for medical and security applications. The very small size (e.g., 125 μm diameter) of these sensors, can be used for example to penetrate tissue or veins, or to deploy in places where small size is crucial. Also, optical devices such as filters, mirrors, and polarizers at a fiber end can be very useful in fiber communication applications.
In several key applications, such as undersea oil exploration and smart wells, the demand for more sensitive and more compact fiber sensors has been a strong drive behind recent research efforts. A current limitation of acoustic fiber sensors is that in order to be highly sensitive, they require a long length of fiber, which makes them bulky and poorly to non-responsive to frequencies above a few hundred Hz.
Various fiber optic sensor systems have been previously disclosed that provide acoustic pressure measurements based on the relative displacements of the two mirrors of a Fabry-Perot interferometric cavity. See, e.g., M. Yu et al., “Acoustic Measurements Using a Fiber Optic Sensor System,” J. Intelligent Mat'l Systems and Structures, vol. 14, pages 409-414 (July 2003); K. Totsu et al., “Ultra-Miniature Fiber-Optic Pressure Sensor Using White Light Interferometry,” J. Micromech. Microeng., vol. 15, pages 71-75 (2005); W. B. Spillman, Jr. et al., “Moving Fiber-Optic Hydrophone,” Optics Lett., vol. 5, no. 1, pages 30-31 (January 1980); K. Kardirvel et al., “Design and Characterization of MEMS Optical Microphone for Aeroacoustic Measurement,” 42nd AIAA Aerospace Sciences Meeting and Exhibit, 5-8 Jan. 2004, Reno, Nev.; J. A. Bucaro et al., “Miniature, High Performance, Low-Cost Fiber Optic Microphone,” J. Acoust. Soc. Am., vol. 118, no. 3, part 1, pages 1406-1413 (September 2005); T. K. Gangopadhyay et al., “Modeling and Analysis of an Extrinsic Fabry-Perot Interferometer Cavity,” Appl. Optics, vol. 44, no. 16, pages 312-3196 (1 Jun. 2005); and P. J. Kuzmenko, “Experimental Performance of a Miniature Fabry-Perot Fiber Optic Hydrophone,” Proceedings of 8th Optical Fiber Sensors Conference, Monterey, Calif., Jan. 29-31, 1992, pages 354-357.
Photonic crystal slabs (PCSs) are photonic crystal structures having a spatially periodically varying refractive index. A PCS exhibits guided resonance optical modes that are strongly confined within the PCS, but are coupled to incident radiation through a phase matching mechanism due to the periodically varying refractive index. These guided resonance modes are typically manifest in transmission or reflection spectra as sharp Fano lineshapes superimposed on a smoothly varying background. See, e.g., M. Kanskar et al., “Observation of leaky slab modes in an air-bridged semiconductor waveguide with a two-dimensional photonic lattice,” Appl. Phys. Lett., vol. 70, page 1438 (1997); V. N. Astratov et al., “Resonant coupling of near-infrared radiation to photonic band structure waveguides,” J. Lightwave Technol., vol. 17, page 2050 (1999); and S. Fan and J. D. Joannopoulos, “Analysis of guided resonances in photonic crystal slabs,” Phys. Rev. B, vol. 65, page 235112 (2002). Such guided resonance modes have been used previously as optical filters or mirrors in light emitting diodes and lasers.
In certain embodiments, an acoustic sensor is provided. The acoustic sensor comprises at least one photonic crystal structure. The acoustic sensor further comprises an optical fiber having an end optically coupled to the at least one photonic crystal structure. The acoustic sensor further comprises a structural portion mechanically coupled to the at least one photonic crystal structure and to the optical fiber. The at least one photonic crystal structure, the optical fiber, and the structural portion substantially bound a region having a volume such that a frequency response of the acoustic sensor is generally flat in a range of acoustic frequencies.
In certain embodiments, a method of fabricating an acoustic sensor is provided. The method comprises providing at least one photonic crystal structure. The method further comprises optically coupling an end of an optical fiber to the at least one photonic crystal structure. The method further comprises providing a structural portion and mechanically coupling the structural portion to the at least one photonic crystal structure and to the optical fiber. The at least one photonic crystal structure, the optical fiber, and the structural portion substantially bound a region having a volume such that a frequency response of the acoustic sensor is generally flat within a range of acoustic frequencies.
In certain embodiments, an acoustic sensor is provided. The acoustic sensor comprises at least one photonic crystal structure. The acoustic sensor further comprises an optical fiber having an end optically coupled to the at least one photonic crystal structure. The acoustic sensor further comprises a structural portion mechanically coupled to the at least one photonic crystal structure and to the optical fiber. The at least one photonic crystal structure, the optical fiber, and the structural portion substantially bound a region having a volume such that a sensitivity of the acoustic sensor is better than about 100 μPa/Hz1/2 in a range of acoustic frequencies below 20 kHz.
In certain embodiments, a method of fabricating an acoustic sensor is provided. The method comprises providing at least one photonic crystal structure. The method further comprises optically coupling an end of an optical fiber to the at least one photonic crystal structure. The method further comprises providing a structural portion and mechanically coupling the structural portion to the at least one photonic crystal structure and to the optical fiber. The at least one photonic crystal structure, the optical fiber, and the structural portion substantially bound a region having a volume such that a sensitivity of the acoustic sensor is better than about 100 μPa/Hz1/2 in a range of acoustic frequencies below 20 kHz.
In certain embodiments, a method of fabricating an acoustic sensor is provided. The method comprises providing at least one photonic crystal structure. The method further comprises optically coupling an end of an optical fiber to the at least one photonic crystal structure. The method further comprises providing a structural portion and mechanically coupling the structural portion to the at least one photonic crystal structure and to the optical fiber. A distance between the at least one photonic crystal structure and the end of the optical fiber is selected to improve a temperature stability of the acoustic sensor.
In certain embodiments, a method of fabricating an optical structure on an optical fiber is provided. The method comprises preparing an end of an optical fiber. The method further comprises forming a layer onto the fiber end. The method further comprises patterning the layer to form the optical structure.
In certain embodiments, a method of fabricating a photonic crystal structure is provided. The method comprises forming a first dielectric layer on an end of an optical fiber. The method further comprises forming a sacrificial layer over at least a portion of the first dielectric layer. The method further comprises forming a second dielectric layer over the sacrificial layer. The method further comprises patterning the first dielectric layer to form a first photonic crystal slab. The method further comprises patterning the second dielectric layer to form a second photonic crystal slab. The method further comprises removing the sacrificial layer.
Present-day optical resonators which have sufficient quality factors to achieve sensitivities comparable to those of piezoelectric transducers are typically large and impractical to fabricate, install, align, and operate. In contrast, certain embodiments described herein comprise an acoustic sensor based on optical resonators formed by photonic crystal slab (PCS) structures with apertures which are orders of magnitude smaller than those of traditional optical cavities. The small size of certain such embodiments provides a sensitivity comparable to that of piezoelectric and capacitive displacement sensors for frequencies larger than about 10 kHz. Photonic crystal structures comprising a pair of PCSs can be used to provide notch and bandpass transmission and reflection filters, and such structures can be utilized in acoustic sensor systems compatible with various applications (e.g., oil exploration, undersea acoustic wave detection).
Certain embodiments described herein provide a novel fiber acoustic sensor based on an external fiber Fabry-Perot (FP) interferometer formed by placing a photonic-crystal slab (PCS) in proximity to the metal- or dielectric-coated reflective end of a single-mode fiber. Certain such embodiments provide an acoustic sensor that is extremely small (e.g., about 0.5 millimeter×about 0.5 millimeter×about 0.1 millimeter), operates up to high frequencies (e.g., 100 kHz and higher), and has a very high sensitivity. The fiber FP acoustic sensor is interrogated in certain embodiments with a single-frequency laser. When an acoustic wave is incident on the PCS, the PCS deflects, which modulates the spacing between the PCS and the mirrored fiber end, at the acoustic frequency. This spacing modulation is detected as a change in the laser power reflected by the FP interferometer. Employing a PCS as the external mirror in certain embodiments provides a lightweight membrane that moves easily in response to an incident acoustic wave, either in a gas (e.g., air) or in a liquid (e.g., water). Certain embodiments also provide the high reflectivity advantageously used for high sensitivity, as detailed herein. Certain such a fiber FP acoustic sensor can be advantageously more sensitive than other reported low-finesse miniature fiber FP sensors (see, e.g., K. Totsu, Y. Haga, and M. Esashi, “Ultra-miniature fiber-optic pressure sensor using white light interferometry,” J. Micromech. Microeng. Vol. 15, 71-75 (2005)). Compared to FP sensors based on other reflecting media, certain embodiments utilizing a PCS combine the advantages of being extremely thin and mass producible, as well as offering a high reflectivity (e.g., greater than 99%) and hence high sensitivity.
PCS structures have been used previously as filters and mirrors, in the same way as multi-layer dielectric stacks. However, PCS structures have several advantages over multi-layer mirrors, including but not limited to, being a single dielectric layer, being compatible with microelectromechanical systems (MEMS), and having unique properties that are difficult or impossible to achieve with multilayer stacks and that can be controlled through geometrical parameters. For example, PCS structures can have a high reflectivity over a broad range of wavelengths (e.g., an observed extinction in transmission of over 99% in a range of wavelengths greater than about 30 nanometers), and can be used as efficient filters at telecom wavelengths (e.g., 1540 nanometers) with sharp resonances observed to have Q of about 5000. In addition, a PCS structure can be used as a circular polarization beam-splitter separating plane-polarized light into its spin-polarized components. Also, though an introduction of a small form birefringence, a PCS structure can act as a dual quarter-wave retarder-based polarizing beam splitter, which separates an incoming wave equally into two orthogonal polarizations through reflection and transmission.
Single PCS Structures
In certain embodiments, the at least one photonic crystal structure 20 comprises a PCS 70, an example of which is schematically illustrated by
In certain embodiments, the first material 72 comprises a solid dielectric material, examples of which include but are not limited to, silicon, silica, silicon nitride, ceramics, and plastics. In certain embodiments, the first material 72 comprises a solid semiconductor material, examples of which include but are not limited to, silicon, germanium, indium phosphide, gallium arsenide, or other III-V semiconductor materials. In certain embodiments, the second material 76 comprises a gas (e.g., air). In certain embodiments, the second material 76 comprises a fluid, examples of which include but are not limited to, water, isopropanol, ethanol, methanol, and other alcohols.
In certain embodiments, the thickness T of the PCS 70 is in a range between about 100 nanometers and about 1000 nanometers. In certain embodiments, the PCS 70 has a substantially square shape, while in other embodiments, the PCS 70 has a substantially circular, rectangular, hexagonal, elliptical, or other shape.
In certain embodiments, the regions 74 have a maximum width along a direction substantially parallel to the PCS 70 in a range between about 100 nanometers and about 1500 nanometers. In certain embodiments, the regions 74 have a substantially circular shape, while in certain other embodiments, the regions 74 have a substantially elliptical, oval, square, rectangular, triangular, pentagonal, hexagonal, semicircular, or other shape.
In certain embodiments, the array of regions 74 has a substantially two-dimensionally-periodic distribution. The periodicities of the distribution in two different directions generally parallel to the PCS 70 are substantially the same in certain embodiments, while in certain other embodiments, the periodicities are different. In certain embodiments, the center-to-center distance between nearest-neighboring regions 74 is in a range between about 100 nanometers and about 1550 nanometers. In certain embodiments, the substantially two-dimensionally-periodic distribution of the array of regions 74 is square, while in certain other embodiments, the substantially two-dimensionally-periodic distribution is rectangular, triangular, square, rhombic, oblique, or hexagonal. Other substantially two-dimensionally-periodic distributions are also compatible with certain embodiments described herein.
In certain embodiments, the regions 74 comprise a plurality of holes extending at least partially through the thickness of the PCS 70, containing the second material 76, and having a substantially two-dimensionally-periodic distribution within the PCS 70. For example,
As schematically illustrated by
In certain other embodiments, as schematically illustrated by
In certain embodiments, the at least one photonic crystal structure 20 comprises a single PCS 70 that exhibits at least one optical resonance having a resonance frequency and a resonance lineshape.
Similar behavior was observed for a PCS 70 in the experimental apparatus schematically illustrated by
Dual PCS Structures
In certain embodiments, the photonic crystal structure 20 comprises a first PCS 100 and a second PCS 102 substantially parallel to the first PCS 100, as schematically illustrated by
In certain embodiments, there is no physical contact between the first PCS 100 and the second PCS 102. The first and second PCSs 100, 102 can undergo displacements relative to one another in response to incident acoustic waves 40. In certain embodiments, light is incident on the first and second PCSs 100, 102 in a direction substantially perpendicular to the PCSs 100, 102. In certain embodiments, the light is provided by an optical fiber 50, as schematically illustrated by
The resonance frequency and the resonance lineshape of the pair of PCSs are both dependent on changes of the perpendicular distance between the two PCSs and on changes of the lateral relative positions of the two PCSs. The two PCSs exhibit optical behavior similar to that of a single PCS, and through the relative displacements, the geometry and optical properties of the photonic crystal structure can be tuned. U.S. Patent Application Publication No. US 2004/0080726 A1, which is incorporated in its entirety by reference herein, discloses calculations (e.g., temporal coupled-mode theory calculations and finite-difference time-domain simulations) of the transmission spectrum for a pair of PCSs as functions of the frequency of the incident light and of the displacement between the two PCSs. These calculations replicate the behavior shown in
In certain embodiments, the two PCSs are brought sufficiently close to one another that they are optically coupled in the near-field to one another (referred to herein as a near-field configuration). In certain embodiments, the two PCSs are spaced apart from one another such that the PCSs are not optically coupled to one another, but form a cavity (referred to herein as a Fabry-Perot configuration). In either the Fabry-Perot configuration or the near-field configuration, the optical resonances shift in frequency (or wavelength) with changing displacement between the two PCSs. Thus, the amount of displacement between the two PCSs can be detected by measuring the transmitted power (or the reflected power) at a predetermined frequency (or wavelength). In general, the near-field configuration generates a larger shift of frequency (or wavelength) than does the Fabry-Perot configuration, such that the near-field configuration has a higher sensitivity to displacements than does the Fabry-Perot configuration.
In certain embodiments in which the two PCSs are optically coupled together in the near-field configuration, the optical resonances are split into two resonances. The amount of splitting varies with the displacement between the two PCSs which, in certain embodiments, provides a measure of the displacement.
In certain embodiments in which the two PCSs are coupled in the near-field configuration, additional resonances appear in the transmission spectra when the PCSs are laterally displaced relative to one other in a direction substantially parallel to the PCSs, as schematically illustrated by
Fiber Compatibility
The sharp resonances of typical optical resonators or filters are sensitive to the incident angle of the light. Typically, to avoid this sensitivity to the incident angle, the incident light is collimated so as to approximate a plane wave. When using an optical fiber as the light source, the light emitted by the optical fiber possesses a certain angular distribution which is typically collimated for present-day optical resonators using additional collimation optics and additional fiber-to-fiber coupling hardware.
In contrast, certain embodiments described herein have one or more resonances which are substantially independent of the incidence angle of the optical beam over a range of incidence angles. In certain such embodiments, the light emitted by the optical fiber has an angular distribution such that a substantial fraction (e.g., more than 50%) of the light incident on the PCS is within the range of incidence angles for which the resonance frequency of such resonances does not change. For such resonances, the linewidth of the resonance is also essentially independent of the incidence angle. Such an angular insensitivity implies that the resonances do not have to be excited by a collimated beam (e.g., by light which approximates a plane wave).
In certain embodiments in which the resonance is insensitive to the incidence angle, the various angular components of the light emitted by the optical fiber are all affected by the PCS structure in the same way, so the acoustic sensor behaves in much the same way as if the light was collimated. In certain such embodiments, since the resonance is insensitive to the incidence angle, the light from the optical fiber directly impinges the PCS structure without intervening collimation optics between the optical fiber and the PCS structure. Certain such embodiments advantageously avoid using complicated collimation or coupling components, thereby simplifying integration and packaging and lowering cost.
The fiber-compatibility of the PCS structure advantageously permits certain embodiments described herein to be easily incorporated into already-present and widely-used fiber-based acoustic sensor systems. In addition, the angular insensitivity of the PCS structure advantageously facilitates incorporating several types of filters into fiber-based optical communication networks.
In an example embodiment, a silicon-nitride PCS illuminated by transverse-electric (TE) polarized light has a resonance mode with a wavelength of about 695 nanometers.
In certain embodiments in which the acoustic sensor 10 further comprises an optical fiber 50 optically coupled to the at least one photonic crystal structure 20 (e.g., as schematically illustrated by
Tailoring the Optical Resonance
Certain eigenmodes in a PCS possess infinite lifetimes, hence are uncoupled to outside radiation at normal incidence. Therefore, in present-day optical resonator systems utilizing photonic crystals, it is generally not possible to couple to certain resonances (referred to herein as non-degenerate resonances) with normally-incident plane waves due to a symmetry mismatch between the resonance mode and the incident wave. This effect was observed experimentally by Pacradouni et al., “Photonic band structure of dielectric membranes periodically textured in two dimensions,” Phys. Rev. B, vol. 62, page 4204 (2000), and discussed theoretically by Paddon and Young, “Two-dimensional vector-coupled-mode theory for textured planar waveguides,” Phys. Rev. B, vol. 61, page 2090 (2000). Using group theoretical arguments, Ochiai and Sakoda, in “Dispersion relation and optical transmittance of a hexagonal photonic crystal slab,” Phys. Rev. B, vol. 63, page 125107 (2001), showed that these resonances are uncoupled due to a symmetry mismatch with outside radiation.
However, measurements and group theory calculations show that it is possible to couple to these non-degenerate resonances in a PCS lacking mirror symmetry. As described more fully below, simulations and experimental results show that such non-degenerate resonances can indeed be excited by breaking the mirror symmetry of the PCS structure, either by breaking the periodicity of the lattice array or by breaking the mirror symmetry of the unit cells (e.g., in a square lattice array). In addition, it is possible to control the sharpness (e.g., linewidth, quality factor) of such resonances by adjusting the degree of asymmetry (e.g., the size of the non-symmetric region of the holes of the PCS structure). In certain embodiments, the quality factor of these resonances can be tuned from a finite minimum to infinity. Resonances sharper than, the spectral linewidth of the source are generally practically useless, so in certain embodiments, the tuning is, done from a finite minimum to a finite maximum (as determined by the linewidth of the incident light).
Such PCS structures are expected to have applications for mode selection and linewidth control in lasers, and will find use in acoustic sensor applications by advantageously improving and controlling the sensitivity of the acoustic sensor system. Certain embodiments described herein advantageously improve the sensitivity of the acoustic sensor system up to a limit imposed by other factors, such that the PCS structure is not the limiting element. In certain embodiments in which a lower sensitivity is desirable (e.g., to improve the dynamic range), the sensitivity of the acoustic sensor system is lowered such that the PCS structure is the limiting element. In certain embodiments, the lack of mirror symmetry is implemented for a PCS structure with a triangular lattice array or any other lattice array geometry, or in general, for any kind of an optical resonator system.
In certain embodiments, the non-degenerate resonances of a PCS with a symmetric structure that are uncoupled to normally-incident plane waves are excited in a mirror-symmetry-lacking PCS structure. In certain embodiments, one or more of the mirror symmetries of the PCS structure is advantageously broken or removed to allow coupling to the non-degenerate resonances. In certain embodiments, the coupling to these non-degenerate resonances is advantageously controlled by selecting the degree of asymmetry. In certain embodiments, the at least one photonic crystal structure has a symmetry axis and the light incident normal to the at least one photonic crystal structure is polarized in a direction substantially perpendicular to the symmetry axis. In certain other embodiments, the normally-incident light is polarized in a direction substantially parallel to the symmetry axis.
In certain embodiments, the asymmetry of the PCS structure is generated by an asymmetry in the substantially periodic distribution of holes.
In certain embodiments, one or more of the mirror symmetries of the PCS structure is broken or removed. In certain such embodiments, one or more of the mirror symmetries of the unit cell of the periodic array of holes in the PCS is removed.
In certain embodiments, the coupling to the non-degenerate resonances can be controlled by advantageously selecting the degree of asymmetry of the hole.
The simulations were done for a dielectric constant of 12, corresponding roughly to the dielectric constant of Si or GaAs at optical frequencies. The PCS thickness was chosen to be 0.75a, where a is the lattice constant of the periodic structure. The radius of the circular portion of the hole was chosen to be 0.4a and the width of the square-shaped regions was chosen to be 0.025a. As can be seen in
In certain embodiments, the magnitude of the asymmetry of the holes is selected to provide a desired amount of coupling to normally-incident plane waves.
To demonstrate that the results of the analysis and simulations can be observed in a real structure, the three PCS structures generally corresponding to
The measured sharp resonances shown in
To illustrate that the non-degenerate resonance appears only in the mirror-asymmetric PCS structure (corresponding to
Acoustic Sensor Systems
In certain embodiments, the acoustic sensor system 200 is compatible with operation in a liquid (e.g., seawater) or other media. As schematically illustrated in
In the embodiment schematically illustrated by
In certain embodiments, the PCSs 70a, 70b undergo movement relative to one another (e.g., one movable PCS 70b moves relative to a non-moving PCS 70a) in response to forces applied to the at least one photonic crystal structure 20. In the embodiment schematically illustrated by
In certain embodiments, the housing 30 comprises a structure 210 comprising one or more supports 212 and a movable portion 220. The housing 30 further comprises a coupler 230 configured to be coupled to the optical fiber 50. The movable portion 220 is mechanically coupled to the coupler 230 by the one or more supports 212. The optical fiber 50 of certain embodiments passes through an opening in the coupler 230 and the fiber core 52 is in proximity to and is optically coupled with the photonic crystal structure 20.
Example materials for the structure 210, the movable portion 220, and the supports 212 include, but are not limited to, crystalline silicon, polysilicon, silica, silicon nitride, ceramics, plastics, amorphous diamond, germanium, indium phosphide, gallium arsenide, and metals and metal alloys. Example materials for the coupler 230 include, but are not limited to, crystalline silicon, Pyrex glass, quartz, polysilicon, silica, silicon nitride, ceramics, plastics, amorphous diamond, germanium, indium phosphide, gallium arsenide, and metals and metal alloys.
In certain embodiments, the coupler 230 comprises an optically transmissive portion 232 (e.g., a hole, a window, an optically transmissive membrane) through which the optical fiber 50 emits light to irradiate the photonic crystal structure 20. The optically transmissive portion 232 allows light emitted by the fiber core 52 to irradiate the photonic crystal structure 20, and allows light reflected by the photonic crystal structure 20 to be received by the fiber core 52.
The movable portion 220 is configured to move (e.g., as denoted by the double-headed arrow in
In certain embodiments, the movement of the PCS 70b has a component in a direction substantially perpendicular to the PCS 70a, wherein the movement changes a distance between the PCSs 70a, 70b. In the embodiment schematically illustrated by
In certain embodiments, the sensitivity (e.g., the change of the detected reflected power per unit of incident acoustic pressure) of the photonic crystal structure 20 is advantageously increased by utilizing a signal having a frequency (or wavelength) offset slightly from one of the resonance frequencies (or wavelengths) of the double-PCS photonic crystal structure 20. In certain embodiments utilizing extremely high sensitivities, the PCSs 70a, 70b are designed to have extremely sharp resonances, e.g., by breaking a mirror symmetry of at least one of the PCSs 70a, 70b, as described herein.
In certain embodiments, the mechanical properties of the acoustic sensor structure 200 (e.g., mechanical resonance frequency, spring constant) are dependent on both the movable portion 220 of the structure 210 and the one or more supports 212. In certain embodiments, the movable portion 220 serves as the mechanical spring by providing a restoring force in response to displacements of the movable portion 220 by acoustic waves 40. In certain other embodiments, the supports 212 serve as the mechanical spring by providing the restoring force in response to displacements of the movable portion 220 by acoustic waves 40. Other embodiments utilizing other spring designs for the structure 210 or the supports 212 are also compatible with embodiments described herein.
In certain embodiments, the acoustic sensor system 200 is insensitive to static pressure variations in the medium (e.g., seawater) in which it operates. As an example, the acoustic sensor system 200 of certain embodiments is operable close to the surface of seawater, or several feet below the surface of seawater. In certain embodiments, the housing 30 comprises at least one pressure conduit 240 between an inner region 250 within the housing 30 and an outer region 260 outside the housing 30. In certain embodiments, the at least one pressure conduit 240 comprises the movable portion 220 of the housing 30. In certain such embodiments, the movable portion 220 comprises an elastic membrane that is responsive to a pressure differential between the inner region 250 and the outer region 260 by moving to remove the pressure differential. In certain embodiments, the supports 210 provide the restoring force to the movable portion 220 and are responsive to a pressure differential across the movable portion by moving the movable portion 220 to reduce the pressure differential. The at least one pressure conduit 240 of certain embodiments serves as low-pass filters that equalize the static pressure between the inner region 250 and the outer region 260.
In certain embodiments, the at least one pressure conduit 240 comprises a hole through the housing 30, with the hole fluidly coupling the inner region 250 with the outer region 260. In certain such embodiments, the inner region 250 is filled with the same medium (e.g., seawater) of the acoustic waves 40 as is the outer region 260, and the medium is free to flow between the inner region 250 and the outer region 260. In certain embodiments, the at least one pressure conduit 240 comprises a hole through the housing 30 and an elastic membrane that seals the at least one pressure conduit 240 to fluidly isolate the inner region 250 from the outer region 260. The membrane of certain embodiments is responsive to a pressure differential between the inner region 250 and the outer region 260 by moving to reduce the pressure differential, thereby still acting as a low-pass filter equalizing the pressure inside and outside the acoustic sensor system 200, while keeping the medium (e.g., seawater) from entering the acoustic sensor system 200. In certain such embodiments in which it is desirable to not expose the photonic crystal structure 20 or other internal components of the acoustic sensor system 200 to the medium (e.g., seawater) which can be corrosive and dirty, the membrane advantageously keeps the medium of the acoustic waves 40 from entering the inner region 250 within the housing 30. Example materials for the membrane include, but are not limited to, silicon nitride or rubber.
In certain embodiments, the acoustic sensor system 200 includes other structural components for better performance and reliability. These other structural components are not crucial for the operation of the acoustic sensor system 200. In certain embodiments, the acoustic sensor system 200 comprises one or more spacers 270 positioned to avoid contact between the two PCSs 70a, 70b in response to a large-magnitude pressure wave incident on the acoustic sensor system 200, thereby advantageously avoiding stiction between the two PCSs 70a, 70b. The spacers 270 of certain embodiments serve as safety structures which define a minimum separation between the two PCSs 70a, 70b, thereby preventing the two PCSs 70a, 70b from contacting and sticking to each other. Example materials for the spacers 270 include, but are not limited to, crystalline silicon, polysilicon, silicon nitride, silicon oxide, amorphous diamond, ceramics, plastics, germanium, indium phosphide, gallium arsenide, and metals and metal alloys. In certain embodiments, amorphous diamond is used because it is hydrophobic which facilitates the prevention of sticking of the two PCSs 70a, 70b.
Due to the sensitivity of the optical properties of the photonic crystal structure 20 on the medium surrounding the PCSs 70a, 70b, in certain embodiments, the medium in which the acoustic sensor system 200 is placed (e.g., water) is advantageously restricted from the region 280 within the acoustic sensor system 200. In certain such embodiments, the PCSs 70a, 70b of the photonic crystal structure 20 operate within a gas (e.g., air). In certain embodiments, the housing 30 defines a region (e.g., inner region 250) comprising a liquid and external to the at least one photonic crystal structure 20 and defines the region 280 containing the at least one photonic crystal structure 20 and that is substantially free of the liquid. While liquid may be able to intrude into the region 280 through the opening under the spacers 270, in certain embodiments, both the pressure of the gas inside the region 280 and the small size of the openings under the spacers 270 are selected to advantageously prevent intrusion of the liquid into the region 280, which could otherwise degrade the operation of the acoustic sensor system 200. Certain embodiments advantageously improve the liquid expulsion out of the region 280 further by providing at least a portion of the photonic crystal structure 20 with a hydrophobic surface configured to restrict the liquid from the region 280.
In certain embodiments, the secondary housing 310 is sufficiently elastic to equalize the pressure outside and inside the secondary housing 310 such that pressure modulations due to the incident acoustic wave 40 are translated into the medium (e.g., gas or fluid) within the secondary housing 310. In certain such embodiments, the secondary housing 310 comprises a balloon. In certain other embodiments, the secondary housing 310 comprises a rigid portion and an elastic membrane.
Certain embodiments of the acoustic sensor system 200 described herein provide various advantages over standard fiber-based sensor systems. In certain embodiments, the acoustic sensor system 200 advantageously achieves higher frequency operation due to the flexibility provided by MEMS fabrication technology. In certain such embodiments, the acoustic sensor system 200 is designed to operate at frequencies larger than 10 kHz, a range that is inaccessible for present-day acoustic fiber sensor systems, and in certain embodiments, can operate at frequencies up to about 50 kHz. In certain embodiments, the PCS-based acoustic sensor system described herein is advantageously more sensitive at higher frequencies than are present-day acoustic fiber sensor systems. In certain embodiments, the acoustic sensor system 200 advantageously provides high sensitivity (e.g., sensitive to less than 30 micropascals/Hz1/2). In certain embodiments, the acoustic sensor system 200 comprises a photonic crystal structure 20 that can be fabricated on substrates (e.g., chips) using lithography techniques (as described more fully below), thereby facilitating mass production and low cost, and that is fiber-compatible. In certain embodiments, utilizing MEMS fabrication technology to fabricate the acoustic sensor system 200 advantageously results in acoustic sensor systems that are small in size, light, and compact. In certain embodiments, the compactness of the PCS-based acoustic sensor systems described herein advantageously facilitates their deployment. In certain embodiments, the PCS-based acoustic sensor systems described herein can be advantageously designed to be insensitive to the polarization of the incident light, thereby eliminating the need for compensation for polarization-induced signal fading.
Fabrication of an Acoustic Sensor
In certain embodiments, surface micromachining techniques and bulk micromachining techniques are used in the fabrication process flow to form various components of the acoustic sensor system 200. Lithography techniques compatible with embodiments described herein include, but are not limited to, optical lithography, electron-beam lithography, nano-imprinting techniques, and other techniques generally compatible with microelectromechanical system (MEMS) fabrication. Surface micromachining techniques compatible with embodiments described herein include, but are not limited to, film deposition, dry etching, wet etching, epitaxial growth, wafer bonding, and sacrificial releasing. Bulk micromachining techniques compatible with embodiments described herein include, but are not limited to, anisotropic or isotropic deep reactive ion etching, anisotropic wet etching using KOH (potassium hydroxide) or TMAH (tetramethylammonium hydroxide), and isotropic wet etching.
As schematically illustrated by
As schematically illustrated by
As schematically illustrated by
As schematically illustrated by
As schematically illustrated by
As schematically illustrated by
As schematically illustrated by
As schematically illustrated by
In certain embodiments, to create 100 micron×100 micron free-standing silicon PCSs 70a, 70b, 808-micron-wide square apertures 650 are formed on the back of the SOI wafer 500 using anisotropic etching to etch through the 500-micron-thick substrate 510. Using an anisotropic etchant of 30% KOH in water with 1% isopropyl alcohol to reduce surface tension, well-defined structures with smooth etched surfaces can be achieved.
Analysis of the Mechanics of a Diaphragm
The mechanics of the movable portion 220 and of the elastic portions (e.g., the secondary housing 310) of the acoustic sensor system 200 affect the performance of various embodiments described herein. These mechanics are analyzed below for various configurations of the acoustic sensor system 200. While the calculations below provide some insight into the operation of various embodiments described herein, but are not intended to be limiting.
A. Free Vibration of a Diaphragm
The equation of motion for the transverse displacement u of a stretched diaphragm with thickness h, and density ρ can be expressed as:
(See, e.g., I. Ladabaum et al., “Surface micromachined capacitive ultrasonic transducers,” Ultrasonics, Ferroelectrics and Frequency Control, IEEE Transactions, vol. 45, issue 3, pages 678-690 (May 1998); and M. Yu, “Fiber-Optic Sensor Systems for Acoustic Measurements,” Ph.D. Dissertation, University of Maryland, College Park, Md.) Here σ is the residual stress and D is the flexural rigidity, defined as:
where E is the Young's modulus, and ν is Poisson's ratio. It should be noted that equation (1) is only applicable for small transverse displacements. When the deflection is large, the equation becomes non-linear.
For a clamped circular diaphragm with radius a, assuming a solution u(r,θ,t)=u(r,θ)ejωt, equation (1) becomes:
D∇4u−hσ∇2u=hρω2u (3)
which has a solution of the form:
u(r,θ)=[AJm(αr)+BIm(βr)] cos(mθ) (4)
where Jm( ) is the Bessel function of the first kind of order m, and Im( ) is the modified Bessel function of the first kind of order m, with
The boundary conditions state that u(a, θ)=0, and
These conditions reduce to the eigenvalue equation:
that can be solved together with equations (5), which can be summarized as:
(βa)2−(αa)2=κ2 (7)
where κ is the useful “tension parameter” defined as κ=a√{square root over (hσ/D)}.
The solutions to equations (6) and (7) for each m=0, 1, 2, . . . can be denoted as αmn and βmn, where n=1, 2, 3 . . . denotes the nth root. The boundary conditions give the eigenmodes as:
where the eigenfrequency of a mode is found through equations (5) as:
B. Forced Oscillation of a Diaphragm
For a forced and damped diaphragm, the equation of motion becomes:
where b is the damping coefficient, and P is the pressure applied on the diaphragm surface. In the case of a miniature microphone, where λacoustic>>a, the pressure wave will be a plane wave, hence P(r,θ,t)=P(t)=P0ejωt.
Due to the similarity to the free vibration problem, we expect a solution of the form:
where umn are the modes from the free vibration problem, and Amn are modal participation factors. Putting equation (11) into equation (10) provides the following:
The second term on the left-hand-side is given by equation (3). Hence, equation (12) becomes:
To solve this equation, the orthogonality of the eigenmodes can be exploited, which is:
Using the orthogonality, the left-hand-side in equation (13) becomes:
while the right-hand-side becomes:
Since
the incident pressure wave only couples to modes with m=0, the modes that have only radial nodes (no polar nodes). Therefore, the index m can be dropped, so that only the index n is used.
In this case, the eigenvalue equation (6) reduces to:
And the eigenmodes in equation (8) become:
The equation for the modal participation factor An becomes then:
where Qn=hρωn/b is the quality factor of the nth mode, and the constant Un is:
Assuming a solution of the form An(t)=Anej(ωt+φ
Hence, we get the displacement as:
This is the general solution for any frequency. For low frequencies, such that ω<<ωn:
This is a general solution for the transverse displacement of a stretched diaphragm that is vibrated by a pressure plane wave at a frequency below resonance.
C. Solutions for the Special Cases of Membrane and Plate
Two different kinds of structures, which are diaphragms made of silicon-nitride and crystalline-silicon, are of interest. Due to the mechanical properties of these two materials, the diaphragm solutions have closed forms as is discussed below.
C.1 Membrane Solution
A membrane is a diaphragm where the residual stress is dominant, e.g. κ→∞. The membrane structure is a good approximation for κ>20, which is the case for a silicon-nitride diaphragm that usually has a high residual stress. In this case, since βn→κ/a→∞, the eigenvalue equation (15) becomes simply J0(αna)=0. For notational simplicity, αna=zn, where zn denotes the nth zero of J0(x).
Also, the eigenmodes in equation (16) become un(r)=CJ0(znr/a), so that:
The eigenfrequencies in equation (9), on the other hand, become:
Using these in the general solution of equation (20):
To find a closed form of this expression, two different damping conditions, which are b=0, and b→∞ will be considered.
C.1.a Membrane Solution—Negligible Damping Case
For b=0, the displacement in equation (21) becomes:
which can be recognized as a Fourier-Bessel series. A function in the interval x=(0,1) can be expanded in a Fourier-Bessel series as:
where the coefficients Cn are given as:
Considering the integral
the displacement for negligible damping in a closed form can be expressed as:
This solution is also consistent with other reports. See, e.g., W. P. Eaton et al., “A new analytical solution for diaphragm deflection and its application to a surface micromachined pressure sensor,” Int'l Conf. on Modeling and Simulation of Microsystems, 1999. Note that equation (22) is an exact solution applicable to the whole range of r=(0,a).
C.1.b Membrane Solution—Strong Damping Case
For b→∞, the displacement in equation (21) becomes:
Considering the integral
the displacement for strong damping in a closed form can be expressed as:
Therefore, when the damping is very strong, the membrane tends to move as a whole without a noticeable bending.
C.2 Plate Solution
A plate is a diaphragm where the bending stiffness is dominant, e.g. κ=0. The plate structure is a good approximation for κ<2, which is the case for a crystalline-silicon diaphragm that usually has very low residual stress.
In this case, since βn=αn, the eigenvalue equation (15) becomes:
For notational simplicity, αna=zn, where zn denotes the nth zero of the function ℑ0(x) that is defined as:
Whence, the eigenmodes in equation (16) become un(r)=Cℑ0(znr/a), so that:
The eigenfrequencies in equation (9), on the other hand, become:
Using these in the general solution of equation (20):
To find a closed form of this expression, two different damping conditions, which are b=0, and b→∞ are considered.
C.2.a Plate Solution—Negligible Damping Case
For b=0, the displacement in equation (24) becomes:
It is possible to define a generalized Fourier-Bessel series for the function ℑ0(x), using the orthogonality of ℑ0(znx), which is:
Using this orthogonality, a function in the interval x=(0,1) can be expanded as:
Where the coefficients Cn are given in this case as:
Calculation shows that
Hence, the displacement for negligible damping in a closed form can be expressed as:
Note that this is an exact solution applicable to the whole range of r=(0,a). This solution is also consistent with other reports. See, e.g., W. P. Eaton et al., “A new analytical solution for diaphragm deflection and its application to a surface micromachined pressure sensor,” Int'l Conf. on Modeling and Simulation of Microsystems, 1999. Also note that the decay from r=0 to r=a is more rapid compared to the membrane case.
C.2.b Plate Solution—Strong Damping Case
For b→∞, the displacement in equation (21) becomes:
Calculation shows that
Hence, the displacement for strong damping in a closed form can be expressed as:
Therefore, as in the membrane case, when the damping is very strong, the plate tends to move as a whole without a noticeable bending.
D. Mechanical Impedance Effects of the Surrounding Medium
Calculations of mechanical impedances can facilitate understanding what effect the surrounding medium (such as air or water), and the damping, will have on the displacement of the diaphragm. The mechanical impedance Z is defined as the ratio of pressure to speed, namely Z=P/υ. In the case discussed here, υ(r)=jωu(r). To calculate the impedance of the diaphragm, the lumped speed is used, which is:
so that,
Using equation (22), the impedance of a diaphragm that can be approximated by a membrane is then:
And similarly, using equation (25), the impedance of a diaphragm that can be approximated by a plate is then:
The impedance due to the damping will be just Zb=b, as can be verified using equation (23) or equation (26). The fact that the damping impedance is a real number means that it is responsible for dissipation loss in the system.
To calculate the impedance of the surrounding medium, the expression for the displacement of particles in an acoustic wave can be used:
where ρa is the density of the medium, and υn is the speed of the acoustic wave (not to be confused with the speed of the particles that are displaced in the medium). Using equation (29), the impedance of the surrounding medium can be expressed as:
The total impedance of the system will then be the sum of the impedance of the diaphragm, the damping impedance, and the impedance of the surrounding medium:
Ztotal=Zd+Zb+Zs
The total displacement of the diaphragm will depend on the value of this total impedance. If one of the impedances is much larger than the others, the diaphragm displacement will be dominated by it. For example, if the membrane impedance is dominant, i.e. Zd>>Zb,Zs, the displacement will be just as in equation (22) or equation (25), the diaphragm displacements under negligible damping. If the damping impedance is dominant, i.e. Zb>>Zd,Zs, the displacement will be just as in equation (23) or equation (26), the diaphragm displacements under large damping conditions. And, if the impedance of the surrounding medium is dominant, i.e. Zs>>Zd,Zb, the displacement will be just as in equation (29), which is the displacement of the particles in the acoustic wave.
E. Numerical Values for the Impedances
As an example system, a circular diaphragm made of either silicon-nitride or crystalline-silicon has the radius of a typical SMF-28 singlemode fiber (e.g., 62.5 microns), and is separated by a distance d from the end of the fiber. Table 1 gives values of various parameters and constants to be used in the calculations.
Other than the formulas given in the previous sections, an expression can be used to calculate the damping. The calculation of damping is usually complex, and has also a strong dependence on the overall geometry. However, an estimate of the damping can still be made. Because of the close proximity of the diaphragm and the fiber end, the dominant damping will be the squeeze film damping which can estimated as:
where μ is the dynamic viscosity of the medium, A=πa2 is the area of the plates, and d is the gap length (see, e.g., M. Andrews et al., “A comparison of squeeze-film theory with measurements on a microstructure, Sensors and Actuators A, vol. 36, pages 79-87 (1993)).
Using the values in Table 1:
bair=9.38×104 kg/m2/s, damping in air
bwater=5.04×106 kg/m2/s, damping in water
Similarly, using the values in Table 1 in the impedance formulas equations (27), (28), and (30):
|ZSiN|=1.62×106 kg/m2/s, impedance of a silicon-nitride membrane
|ZSi|=1.09×105 kg/m2/s, impedance of a silicon plate
|Zair|=415 kg/m2/s, impedance of air
|Zwater|=1.48×106 kg/m2/s, impedance of water
|Zsea-water|=1.56×106 kg/m2/s, impedance of sea-water
These results show that for the given dimensions, the impedances, of the membranes are comparable to the damping and water impedances.
These results show that the initial assumptions about the displacement of the diaphragm in air and water were inaccurate, and these calculations can be used to make a more optimal sensor design, either for air or water operation.
Fiber Fabry-Perot-Based Acoustic Sensor Designs
The expressions previously derived can be used to find optimal parameters for the acoustic sensor systems.
As a first step, the limitations or ranges of these parameters can be calculated. The membrane radius a is preferably equal to at least the radius of the fiber, which is about 62.5 microns, so that the impedance of the membrane does not becomes excessively large so that it limits the pressure sensitivity of the sensor. The size of the membrane is preferably sufficiently small to provide a compact and robust sensor. Therefore, the membrane diameter advantageously does not exceed about 300 microns, above which it becomes large and fragile.
For reasons shown below, the reflectivity of the membrane mirror Rm is preferably as large as possible (e.g., Rm˜1), which is achieved in certain embodiments with a photonic crystal mirror. The reflectivity of the fiber end (Rf) is preferably as small as possible. The reasons for this are discussed more fully below. Also, a small reflectivity on the fiber end is preferably used, since it is usually difficult to deposit a highly reflective mirror, either metal or dielectric, on the end of a fiber. Also, if a fiber Bragg mirror is used, its length is preferably reduced by requiring a small reflectivity, as it was explained in previous texts.
To calculate limitations on the cavity length L, several factors are considered. From the mechanical point of view, it is advantageous to have a large length, since this results in a smaller damping. However, when the optical properties are considered, there are certain limitations on the cavity length, as calculated below.
The contrast of the resonance tends to decrease with increasing mirror reflectivities, so that for very high reflectivities, it appears that there is almost no visible resonance. This effect can be avoided by balancing the mirrors of the Fabry-Perot. In fact, as calculated, full contrast can be restored by having:
Rf=Rm2−L/z
where z0=πw02nc/λ=kw02/2 is the Rayleigh range, a characteristic length over which the beam does not diverge significantly.
Also, the maximum sensitivity to displacement occurs at the point where the overall reflection is RP=PrPi=½, on the steeper side of the asymmetric resonance. At that point, the sensitivity is equal to the sensitivity of a regular Fabry-Perot that has an effective reflectivity of:
Reff=√{square root over (RfRm)}=Rm2−L/2z
The sensitivity to displacement σ=∂RP/∂L of a regular Fabry-Perot at the point RP=½ is calculated as:
where K=4Reff/(1−Reff)2.
The above relations can be used to calculate the maximum L. This calculated value is dependent on the minimum reflectivity Reff that is used to achieve the required sensitivity. The minimum required reflectivity for the best case scenario corresponds to the noise level being in the shot-noise limit, and the impedance of water being dominant, so that the membrane displaces with the water particles.
The relations between pressure and water displacement is expressed as:
Using the values υwater=1482 m/s, ρwater=998 kg/m3, and the numbers wanted for the sensor ω=2π×30 kHz, and P=30 μPa/√{square root over (Hz)}:
ΔL=1.08×10−7 nm/√{square root over (Hz)}
When the noise level is at the shot-noise limit, then the minimum detectable displacement is:
Substituting ΔLmin with the above number, and using the values, Pi=1 mW, λ=1550 nm, η=0.9, and solving for Reff:
Reff=0.954
This is the minimum effective reflectivity to achieve the desired sensitivity under the best conditions. This value can be used to calculate the maximum cavity length. Using the above expression Reff=Rm2−L/2z
This is the maximum cavity length in water, a limitation imposed by the optical properties of the fiber Fabry-Perot. This number is smaller (7.21 microns) in air, due to the difference in the Rayleigh range, calculated for a regular SMF-28 fiber.
The other constraint on the cavity length is that it is a multiple of half the operation wavelength, which in certain embodiments is λ=1550 nm.
With this knowledge of the limitations for the main parameters: the cavity length (L), the radius of the membrane (a), and the reflectivities of the fiber end (Rf) and the membrane mirror (Rm), the values can be optimized.
To optimize these parameters, the mechanical properties of the device are considered. In the mechanics calculations, the following impedance values were found for the parameters a=62.5 μm (=radius of an SMF-28 fiber) and L=1 μm:
Zwater=1.48×106 kg/m2/s, impedance of water
bwater=5.04×106 kg/m2/s, damping in water
ZSi=1.09×105 kg/m2/s, impedance of a silicon plate
The three impedances are on the same order, which means that the displacement of the membrane will be reduced by a factor f with respect to the displacement of water particles, where:
With these impedance values, the membrane will displace only by about 22% of the displacement of the water particles. This number is advantageously closer to about 90% for a sensitive device. To achieve this result, the damping in water, and also possibly, the impedance of the silicon plate are advantageously reduced to have:
bwater+ZSi≈1.64×105 kg/m2s
The expressions we need to consider are:
To reduce the overall impedance, L can be increased without changing a, since bwater, has a larger contribution. In such a case, ZSi will remain unchanged, so that advantageously:
bwater≈5.50×104 kg/m2/s
Hence, the length is advantageously:
Since the cavity length is a multiple of half the operation wavelength, the closest number to this is 6×1.55 μm/2=4.65 μm, which is still within the range of Lmax=9.48 μm. Using the cavity length L=4.65 μm, the reduction factor is f=0.9=90%. Since a=62.5 μm remained unchanged in this calculation, the other two design parameters Rf and Rm remain to be optimized.
The displacement of the membrane will be:
ΔL≈f×ΔLwater=0.9×1.08×10−7 nm/√{square root over (Hz)}=9.72×10−8 nm/√{square root over (Hz)}
which results in Reff=0.958 and Rm=Reff2L/z
Therefore, a set of parameters for an example design that provides a sensitivity of 30 μPa/√{square root over (Hz)} at 30 kHz are: a=62.5 microns, L=4.65 microns, Rm=0.980, and Rf=0.936. Other values are also compatible with certain embodiments described herein.
Fabry-Perot-Based Acoustic Sensor Compared to a Fiber Bragg Grating
A simple Fabry-Perot structure to be employed as an acoustic sensor can be constructed with a photonic crystal mirror and a fiber Bragg grating, (e.g., as shown in
The sensitivity σ to the displacement L will then be:
To find the maximum sensitivity, equation (32) is solved for dσ/dL=0, which is satisfied for φ=(3K)−1/2+mπ, keeping in mind that K>>1. This result is expected, stating that the highest sensitivity is at the side of a resonance. Using this value, the maximum sensitivity can be expressed as:
Therefore, the maximum sensitivity only depends on the reflectivity of the mirrors, provided that the linewidth of the laser is much smaller than the linewidth of the Fabry-Perot resonance. This condition is satisfied if L<<c/Δνlaserπ√{square root over (K)}, where Δνlaser is the linewidth of a single-mode laser (or the frequency spread of a multi-mode laser). Thus, for a sensitive measurement, the linewidth of the laser Δνlaser, is advantageously selected to be much smaller than the linewidth of the Fabry-Perot resonance ΔνF-P=c/Lπ√{square root over (K)}, which is dependent on the cavity length L. Thus, equation (33) for the maximum sensitivity imposes a limit on the maximum cavity length on the Fabry-Perot cavity depending on the laser linewidth.
For a sufficiently large laser power such as 1 milliwatt, the dominant noise will be the photodiode shot current. The mean current in the photodiode measuring the reflected power will be I0=Preη/hν, where η is the quantum efficiency of the photodiode.
At the maximum sensitivity point, calculated from equation (31), Pr=Pi/4. This current gives rise to a shot noise current:
where Δf is the electronic system bandwidth.
For a small mirror displacement with peak amplitude ΔL, the signal current in the photodiode will be:
and ΔPr, can be calculated using equation (2) to be:
hence ΔPr=σPiΔL
Operating at maximum sensitivity given in equation (33), the power signal of equation (36) inside the signal current expression of equation (35):
From which the signal-to-noise ratio can be expressed as:
For a unity signal-to-noise ratio, the detection sensitivity of the system will then be:
Using the values, Pi=1mW, λ=1500 nm, η=0.9, and a modest reflectivity R=0.99, the value of ΔLmin=2.25·10−8 nm/√{square root over (Hz)} is obtained. The sensitivity can be increased through the reflectivity of the mirrors. For example, a reflectivity of R=0.999 provides about 10 times better sensitivity. Throughout the calculations below, the value of ΔLmin=10−6 nm/√{square root over (Hz)} is used, since the experimental values have previously been reported to be about an order of magnitude worse than the theoretical limit.
The sensitivity given in equation (33) is only dependent on the mirror reflectivity. It may be expected that the length of the cavity would play a crucial role in the sensitivity, so that a much smaller sensitivity would be provided by a long cavity. If equation (33) is written in terms of the cavity length L, and the resonance linewidth Δν1/2:
which is an expected result. The sensitivity drops with increasing length. Also, as expected, the sensitivity drops with increasing linewidth, since the resonances become less steep. However, in a Fabry-Perot cavity with high reflectivity mirrors, the resonance linewidth is dependent on L, so that the resonances become sharper for longer cavity lengths:
(See, e.g., P. Yeh, “Optical Waves in Layered Media,” (Wiley, New York, 1988).) Therefore, the dependence on L in the sensitivity equation (40) cancels out, so that it is the mirror reflectivity provides the dominant contribution (as long as it is high). In certain such embodiments, the important criterion is therefore that the laser linewidth should be much smaller than the Fabry-Perot resonance linewidth.
To calculate the dynamic range, the minimum detectable length is known, and therefore the maximum length shift is to be calculated. For a Fabry-Perot cavity, L=constant·λ, hence:
Therefore, the maximum shift is ΔLmax=(L/λ) Δλmax. The maximum wavelength shift one can detect is one-quarter the linewidth of the Fabry-Perot resonance. Therefore, the maximum detectable cavity length change is, using equation (41):
Hence, the dynamic range is:
which shows that the dynamic range is independent of the Fabry-Perot parameters such as the reflectivity or the cavity length. For the values used above, a dynamic range about 130 dB (20 log) results. Again, assuming an order of magnitude less sensitivity (10−6 nm/√{square root over (Hz)}) than that predicted, the dynamic range is then around 110 dB. Although this dynamic range is for the displacement measurements, it also applies for pressure, since the displacement is proportional to the pressure.
To compare these results to a single fiber Bragg grating, it is desirable to know if it is possible to get the same values by stretching a piece of such a fiber.
If the resonance wavelength and distance is proportional to each other, as in a Fabry-Perot cavity, so that equation (42) is valid, the sensitivity can be expressed as:
To verify this expression, the expressions for a Fabry-Perot cavity can be used, to get:
which is very close to equation (33), thereby verifying equations (45) and (46).
Having general expressions for the sensitivity, the sensitivity for a fiber Bragg grating can be calculated. The resonance wavelength of such a structure is:
where neff is the effective refractive index, L the total length of the grating, and N the number of layers. (See, e.g., Kersey et al., “Fiber grating sensors,” J. Lightwave Technol., vol. 15, no. 8, page 1442 (1997).) When such a structure is stretched by ΔL, the wavelength shifts by:
where the factor 0.78 comes from changes in the fiber index due to photo-elastic effects. Therefore:
which shows that equation (42) is valid to that order, meaning equation (46) is also valid to that order. Thus, the sensitivity of a Fabry-Perot cavity and a fiber Bragg grating are on the same order for a given wavelength, provided that L·Δλ1/2 of equation (46) is the same.
For example, a commercially available fiber Bragg gratings operating at around 1500 nanometers, a linewidth of 0.02 picometer for a grating around 5 centimeters long, the structure gives L·Δλ1/2=103 nm2. For a Fabry-Perot cavity on the other hand, using equation (11):
To get a similar number (e.g., L·Δλ1/2=103 nm2) for a Fabry-Perot cavity, a reflectivity of R≈0.998 would be used. Hence, such a commercially available fiber Bragg grating seems to have the same sensitivity of a Fabry-Perot cavity with high reflectivity mirrors.
In this analysis of the Fabry-Perot cavity, it was assumed that the linewidth of the laser is much smaller than the linewidth of the Fabry-Perot resonance. The linewidth of the laser should be one to two orders of magnitude smaller than the resonance linewidth, so that the laser does not average over less sensitive regions. When a small cavity length is used, the Fabry-Perot resonance linewidth is large, hence the laser does not have to be very narrow. When the cavity length is large however, the Fabry-Perot resonance becomes sharper, hence a narrower laser is used to achieve the same sensitivity achieved in a short Fabry-Perot cavity. The main problem arises at this point when the laser has to be extremely narrow.
Consider the above case for the 0.02 picometer linewidth, for example. To achieve the calculated sensitivity, a laser as narrow as 10−3 to 10−4 picometer would be used. When a laser is that narrow, other noise sources become dominant over the shot-noise. One of the most important noises for such narrow lasers is the free running frequency noise. In fact, by reducing this noise using a pre-stabilized laser, it was previously experimentally shown that a sensitivity of 10−5/√{square root over (Hz)} can be obtained for a greater than 25 millimeter long Fabry-Perot formed by two fiber Bragg gratings. (See, Chow et al., “Ultra resolution fiber sensor using a pre-stabilized diode laser,” page CPDA9, Post-deadline CLEO 2005 (2005).) This reported value is just about an order of magnitude worse than the fundamental shot-noise limited sensitivity for the same structure. Therefore, it is harder to get high sensitivities with long cavity lengths, since in that case a very good laser is used. However, these results should be encouraging for the fiber Bragg grating structure shown in
Fiber Fabry-Perot Interferometer Compared to Standard Fabry-Perot Interferometer
The reflectivity of a fiber FP interferometer is quite different due to the mode-selective properties of the fiber. As an example,
The detected field is the portion of the total field reflected by the fiber FP interferometer that is coupled to the fiber fundamental mode, or Er=rfEi+ai·tfEi, where ai is the coupling coefficient of the free-space field to the fiber mode. So the only difference with a standard FP interferometer is the factor ai, which eliminates the fraction of the total reflected field that does not have the proper amplitude profile and phase front to be coupled back into the fiber (this fraction is coupled to cladding modes, and/or reflected). The coupling coefficient can be calculated as:
where z0 is the Rayleigh range of the Gaussian beam inside the fiber FP interferometer. Using this result, the field reflection coefficient for the fiber Fabry-Perot interferometer can be expressed as:
Comparing this result to the formula for a standard FP interferometer shows that the spatial filtering performed by the single-mode fiber adds a factor of) 1/(1+jpL/z0) in each term of the series. In contrast to a standard FP interferometer, the summation expression for the fiber FP interferometer does not have a simple analytical closed form. However, a Lerch transcendent can be used to convert it into the integral form:
where
Note that the reflectivity of a standard FP interferometer can also be expressed in a similar form:
where θ=2kL. The integral form for rSFP above is more efficient in numerical calculations than the series form which requires computing a very large number of terms when the reflectivities are high.
The reflection spectrum of both a fiber FP interferometer and a standard FP interferometer are plotted in
The external-mirror fiber FP reflection amplitude can be expressed as rEFFP=∫0∞rFP(θn)e−udu, where rFP=rFP(θ) is the standard FP reflection amplitude with θ=2kL and θn=θ+(L/z0)u. The fiber FP reflection amplitude is then a weighted average of the standard FP reflection amplitude and reflection amplitudes with larger effective lengths. Thus the resonance lineshape is asymmetric and elongated towards increasing wavelength. The contrast is reduced, but can be maximized when Rf≈Rm2−L/z
The transmission spectrum from an example fiber FP interferometer was measured to confirm the predicted asymmetry. One of the FP mirrors was formed by the cleaved end of an SMF-28 fiber, which was coated with 12 nanometers of gold on a 4-nanometer chrome adhesion layer. The calculated reflectivity of the mirror formed on the fiber end was 80%, and the measured reflectivity of this mirror was 65%. The other mirror was a PCS fabricated by standard techniques on a 450-nanometer-thick silicon membrane with a square lattice of 644-nanometer diameter circular holes spaced by 805 nanometers. The calculated reflectivity of the PCS mirror at 1.55 microns was greater than 99%, although due to less-than-ideal fabrication conditions, its actual reflectivity was only 65%, a value actually better matched to the fiber mirror reflectivity. The measured transmission spectrum of this fiber FP interferometer is shown in
By reducing the bandwidth to 1 Hz (−115 dBm, the lower noise-floor curve in
For this fiber FP acoustic sensor, the reflectivity of both the fiber and the PCS was around 65%. In certain other embodiments, this number can be easily increased to above 90% to improve the sensitivity by more than 10 dB. Also, the diameter or width of the PCS silicon membrane was around 150 microns, and in certain other embodiments, this diameter or width can be larger. In certain various embodiments, the PCS has a diameter greater than about 50 microns, greater than about 100 microns, greater than about 125 microns, greater than about 150 microns, or greater than about 200 microns. For a given applied pressure, the membrane displacement increases as the fourth power of this diameter or width. Therefore, even a small increase in membrane area will greatly improve the sensitivity (e.g., 25 dB for two-fold increase in radius). With at least one of these two improvements, both of which can be implemented with straightforward fabrication steps, the sensitivity of certain other embodiments is better than about 1 μPa/Hz1/2, and in certain other embodiments, is better than the 10 nPa/Hz1/2 level in a shot-noise-limited detection.
Analysis of Thermal Mechanical Noise
To be able to calculate the thermal mechanical noise associated with a fiber FP sensor, it is helpful to shortly review the sources of thermal noise. Thermal noise is associated with the random motion of the diaphragm, which has an equation of motion for the transverse displacement u without any incident pressure of:
The first term is the mass term, the second term is the damping term, the third term is the bending term, and the fourth term is the stress term. The damping term implies that any oscillation in the diaphragm will continue to diminish with no limit. Even the small thermal vibrations of the diaphragm would decay, leaving the diaphragm at a temperature below that of the surrounding, so the equation is flawed thermodynamically. To allow the diaphragm to be in thermal equilibrium, a pressure term is included that moves the diaphragm, which makes up for the loss due to the damping. Hence, the thermodynamically proper equation for no incident pressure is:
The pressure term PN has its origin in spontaneous thermal fluctuations which are a function of the damping B, which is included to make up for the damping loss. Thus, while the damping allows the system to transfer energy to the surrounding, it also allows the system to gather energy from the surrounding. This behavior is the basis of the fluctuation-dissipation theorem (see, e.g., H. B. Callen and T. A. Welton, “Irreversibility and generalized noise,” Phys. Rev. Vol. 83, 34-40 (1951)), which states that for every mechanism of dissipation, there is always a fluctuating force in the system. This equation can be complicated to solve, but simplifications can be made in certain embodiments in which the wavelength of the acoustic pressure wave is much larger Than the radius of the diaphragm and the frequency of the acoustic pressure wave is much less than the resonant frequency of the sensor. Under such conditions, it is possible to calculate displacements, resonances, and impedances analytically.
A generalized form of the Nyquist relation (see, e.g., C. Kittel, Elementary Statistical Physics, New York, Wiley (1958)) can be used to calculate the spectral density of the fluctuating pressure, which will be the thermal noise level of the acoustic sensor. Analogously to the regular Johnson noise in a resistor VN=√{square root over (4kBTR)}[V/√{square root over (Hz)}], the noise in a pressure sensor can be expressed as:
PN=PN2
=√{square root over (4kBTRtotal)}[Pa/√{square root over (Hz)}],
where Rtotal=B/S is the acoustic resistance related to the damping B in the system, with S=πa2 the area of the diaphragm (see, e.g., T. B. Gabrielson, “Mechanical thermal noise in micromachined acoustic and vibration sensors,” IEEE Trans. Electron Devices Vol. 40, 903-909 (1993)). The individual contributions from viscous resistance (Rvsc), and radiation resistance (Rrad) to the total acoustic resistance Rtotal=Rvsc+Rrad are discussed below, along with a more complete model of the sensor with a more accurate expression for the effective dissipation affecting the diaphragm.
The origin of viscous resistance Rvsc is Brownian motion, which is due to thermal agitation. When the diaphragm is moved through a fluid, it will experience a drag. This drag is due to the random collisions of the fluid molecules. Therefore, there is a damping term associated with this dissipation, referred to as the viscous resistance. Hence, when the diaphragm is at rest, it will still experience this random force that will contribute to the thermal noise.
The source of radiation resistance Rrad is the random pressure fluctuation in the fluid due to thermal agitation. An oscillating diaphragm will gradually lose energy through the generation of pressure waves. Therefore, there is a damping term associated with this dissipation, referred to as the radiation resistance. Hence, the diaphragm is capable of picking up the fluctuating pressure of the ambient environment, which will contribute to the thermal noise.
A.1 Uniform Diaphragm Approximation
For simplicity, any effects of the holes are first ignored, thereby assuming that the diaphragm in
so for this calculation, Rvsc=Redge.
The radiation resistance of the diaphragm can be calculated for ωa/υa>>1 by:
where ρa is the density of the fluid, and Va is the speed of sound in the fluid (see, e.g., L. E. Kinder, A. R. Frey, A. B. Coppens, and J. V. Sanders, Fundamentals of Acoustics, 3rd. ed., New York, Wiley (1982)). The contribution of the radiation resistance to the thermal noise for this case will be small. However, its contribution will be important for the cases when it is desired to decrease the overall damping for a lower thermal noise level.
The thermal noise levels for the case where we have no holes is calculated through the above expressions as Pair=33.5 μPa/Hz1/2 for air, and Pwater=246 μPa/Hz1/2 for water. These noise levels are almost an order of magnitude higher than generally desired. Thus, a more complete model that includes the effects of the holes is desired to analyze whether the presence of the holes can reduce the thermal noise level.
A.2 Diaphragm with Perforations
The presence of holes in the diaphragm creates an additional path for the viscous loss in parallel to the loss from the edges. By including the effect of the N number of holes, there will be an additional N number of parallel paths for the fluid to get out of the gap. Therefore, the viscous resistance in this case will be:
where N=(a2/rh2) is the total number of holes,
is the fill-factor, defined as the ratio of the open area to the whole area (
=π(rh/Λ)2 for a square lattice and
for a triangular lattice).
The hole resistance will have two contributions, which are due to the horizontal flow of the fluid from the surroundings of the hole (squeeze-film flow), and the vertical flow of the fluid through the hole (Poiseuille flow). The horizontal flow contribution from each hole will be (see, e.g., Z. {hacek over (S)}kvor, “On acoustical resistance due to viscous losses in the air gap of electrostatic transducers,” Acustica, Vol. 19, 295-297 (1967-1968)):
where
The vertical flow contribution from each hole, on the other hand, will be (see, e.g., D. Homentcovschi and R. N. Miles, “Modeling of Viscous Damping of Perforated Planar Microstructures. Applications in Acoustics,” Journal of the Acoustical Society of America, Vol. 116, 2939-2947 (2004)):
where an effective thickness
is used to make corrections for the effect of the hole-end when the radius rh and the thickness h is comparable, as in the present case. Hence, the total hole resistance can be expressed as:
Therefore, if we include the contribution of the holes to the squeeze-film damping, the thermal noise level becomes: Pair=29.4 μPa/Hz1/2 for air, and Pwater=216 μPa/Hz1/2 for water. Therefore, the holes do not help much to reduce the noise level, mainly due to the high viscous resistance of the vertical flow through the small holes. For a closed cavity with no flow out of the edges, the damping will be even higher so that the thermal noise becomes: Pair=61.1 μPa/Hz1/2 for air, and Pwater=448 μPa/Hz1/2 for water.
These relatively simple calculations illustrate that to obtain a low self-noise for the fiber FP acoustic sensor, the design should be carefully considered. A more accurate model of various aspects of the sensor can be used to calculate the frequency dependent response of the sensor to an incident pressure that includes the effects of the ambient medium. Although the thermal noise in the ambient is essentially white noise, it will not transfer to the diaphragm as simple as assumed in the calculations above. The acoustic impedances of the diaphragm and the dissipative elements may not match perfectly, so that the noise transferred to the diaphragm can be more complex. Furthermore, since the impedances are in general frequency dependent, the noise transferred to the diaphragm will change with frequency. These effects are addressed in the discussion below.
B. Lumped-Element Equivalent-Circuit Modeling of the Sensor
To calculate the response and noise characteristics of the sensor, it is advantageous to have a model that describes the operation of the sensor in a simple way. One way of constructing such a model is by drawing an equivalent circuit of the sensor that includes the acoustic impedance of various elements of the sensor. In the circuit, instead of electrical resistance and capacitance, acoustic resistance R corresponding to dissipation and acoustic compliance C corresponding to the storing of potential energy are used. Likewise, instead of electrical inductance, an acoustic mass M corresponding to the storing of kinetic energy is used.
An equivalent circuit for an example fiber FP acoustic sensor is schematically illustrated by
Hd(ω)=1−ω2Cd(Mrad+Md)+jωCdRrad
Hcav(ω)=1−ω2CcavMhole+jωCcavRvsc
Hleak(ω)=1−ω2CdMhole+jωCdRvsc
The transfer function for the sensor can then be expressed as:
To calculate the noise, a noise equivalent circuit can be used, as schematically illustrated by
Acoustic resistances: Rvsc=12; Rrad=0.0038
Acoustic mass: ωMd=0.029; ωMrad=0.81; ωMhole=0.031
Acoustic compliance: (1/ωCd)=6.5; (1/ωCcav)=190,000.
In certain embodiments, the high impedance of the Ccav term will dominate the system.
To calculate the effect of these noise sources on the diaphragm, the transfer functions for these sources can be calculated just in the same way as Hsensor, yielding:
The power spectral density of the fluctuating pressure on the diaphragm will then be:
PN2=|Hvsc|2Pvsc2
+|Hrad|
Prad2
.
From the Nyquist theorem: Pvsc2
=4kBTRvsc, and
Prad2
=4kBTRrad and using the expressions for the transfer functions Hvsc and Hrad:
Therefore, the signal-to-noise ratio (SNR) on the diaphragm can be expressed as:
Thus, to design a sensor with a minimum detectable pressure of Pin=MDP [Pa/√{square root over (Hz)}], the design formula will be, using SNR≧1:
To achieve an acceptable low noise level, it is advantageous to choose the design parameters to have low Rvsc and Rrad. It is also advantageous to have a steep frequency response for the filter function Hcav, so that the noise is only amplified at very low frequencies, and not inside the bandwidth in which the sensor is to operate. This expression also gives the interesting result that the self-noise of the sensor cannot be improved by changing the stiffness of the diaphragm.
C. Spectral Behavior of the Sensor
To do calculations based on the equivalent-circuit model, it is advantageous to be able to calculate the individual acoustic impedances. The model includes two dissipative elements, the viscous resistance Rvsc, and the radiation resistance Rrad. The effective viscous resistance of the N number of holes can be expressed as:
where
(see, e.g., D. Homentcovschi et al. cited above). The
radiation resistance can be expressed as
(see, e.g., L. E. Kinder et al., cited above; W. Kuhnel and G. Hess, “A silicon condenser microphone with structured back plate and silicon nitride membrane,” Sensors and Actuators A, Vol. 30, 251-258 (1992)).
Two acoustic compliances, the diaphragm compliance Cd and the cavity compliance Ccav, are considered in the model. The diaphragm compliance for a plate can be expressed as
(see, e.g., M. Rossi, Acoustics and Electroacoustics, Artech House, Inc., 1st ed., (1988)) and the compliance of a cavity with volume Vcav can be expressed as
(see, e.g., W. Kuhnel et al., cited above).
Three acoustic masses are considered in the model: the diaphragm mass Md, the radiation mass Mrad, and the hole mass Mhole. The diaphragm mass can be expressed as
(see, e.g., M. Rossi, cited above), the radiation mass as
(see, e.g., M. Rossi, cited above and W. Kuhnel et al., cited above), and the effective mass of the N number of holes as
(see, e.g., M. Zollner and E. Zwicker, Elektroakustik, Berlin, Springer Verlag, (1987)).
As discussed more fully below, calculations can be made using the model to predict the sensor performance in air and in water. In certain embodiments, optimum sensor performance is achievable by having (1) a close to unity response at a broad range, and (2) a noise level close to the ambient noise level.
C.1 Sensor in an Air Medium
In
In
C.2. Sensor in a Water Medium
D. Effect of the Holes on the Mechanics of the Diaphragm
In certain embodiments, the effect of the perforations on the stiffness of the diaphragm can be advantageously adjusted. The equation of motion discussed above for a uniform diaphragm was:
The holes will make the elasticity of the diaphragm highly anisotropic (see, e.g., M. Pedersen, W. Olthuis, and P. Bergveld, “On the mechanical behaviour of thin perforated plates and their application in silicon condenser microphones,” Sens. Actuators A, Vol. 54, 499-504 (1996)). However, the structure can be approximated as being uniform-homogenous with effective elastic parameters: Deff=D(1−√{square root over ()}); ρeff=ρ(1−√{square root over (
)}); and σeff=σ√{square root over (1−
)}. The resonance frequencies of a membrane (e.g., SiN) will change as:
which would yield, e.g., for =0.4, ωn′=1.14ωn, a 14% larger resonance frequency. The resonance frequencies of a plate (e.g., Si) on the other hand will change as:
which would yield, e.g., for =0.4, ωn′=0.783ωn, a 22% smaller resonance frequency. The response to pressure of a membrane (e.g., SiN) will change as:
which would yield, e.g., for =0.4, u′=1.29u, a 29% larger response. The response to pressure of a plate (e.g., Si) will change as:
which would yield, e.g., for =0.4, u′=2.72u, a 172% larger response.
Photonic Crystal Structure on a Fiber End
Certain embodiments described herein provide a photonic crystal structure formed on a fiber tip or end. While in certain embodiments, such a structure can be used as one of the mirrors of the fiber FP acoustic sensor, in certain other embodiments, such a structure can be used a disposable index sensor, for example to detect biological agents in blood. Certain embodiments can be used as a filter, mirror, or polarizer at the end of the fiber for applications that employ free-space to fiber interfaces. In certain embodiments, the structure can be also spliced (e.g., with fusion splicing) to another cleaved or polished fiber, such that the photonic crystal remains in between two fibers. Thus the photonic crystal can be used as an in-line reflector, filter, or polarizer. One advantage provided by certain embodiments over e.g. fiber Bragg grating based filters would be the temperature stability of such a filter due to its extremely small thickness (e.g., <500 nanometers).
Fabricating such devices can be challenging, e.g., to deposit materials such as silicon at the end of a fiber, and use lithography techniques to pattern small structures on it. The high temperatures involved (e.g., several 100° C.) in depositing materials such as silicon, and especially etching patterns into it with a hot plasma, can damage the polymeric coating of the optical fibers. Also, the extremely large aspect ratio of a piece of fiber makes it difficult to pattern, due to space limitations for putting such a structure into most micro-fabrication equipments.
Certain embodiments described herein utilize an optical structure (e.g., a photonic crystal structure) formed on an end of the fiber. In certain such embodiments, a method of fabrication is used that overcomes temperature issues and other physical constraints, and certain such embodiments can be used to fabricate photonic crystal structures on silicon on the end of a single-mode fiber.
In certain embodiments, the method 800 further comprises depositing a dielectric layer 720 onto the prepared fiber end in an operational block 820. In certain embodiments, the dielectric layer 720 is evaporated onto the fiber end by e-beam evaporation. For example, amorphous silicon (or other high-refractive-index dielectrics such as germanium) can be evaporated onto the fiber end using an e-beam evaporator. In certain other embodiments, the dielectric layer 720 can be deposited by other methods including but not limited to sputtering or chemical vapor deposition. In certain other embodiments, silicon dioxide, silicon nitride, magnesium fluoride, or silicon monoxide can be used as the dielectric layer 720.
In certain embodiments, the method further comprises patterning the dielectric layer 720 in an operational block 830. For example, in certain embodiments, the dielectric layer 720 is patterned by ion-milling using a focused ion beam which patterns a photonic crystal lattice 710 directly onto the dielectric layer 720 at the end of the fiber 730. The photonic crystal of certain embodiments comprises a periodic array of holes that extend either partially or completely through the dielectric layer 720. In certain embodiments, the method 800 further comprises covering the dielectric layer 720 with a thin layer of gold (e.g., thickness of about 10 nanometers) prior to the patterning process to reduce charging effects during the patterning process. In certain other embodiments, the dielectric layer 720 can be patterned using lithography techniques (e.g., covering the fiber end with a photo-resist material, exposing it with an optical or electron-beam lithography tool, developing the photo-resist, etching through the dielectric layer 720, and stripping the photo-resist). In certain embodiments, the periodic array of holes of the photonic structure 710 covers at least part of the area of the core region of the fiber 730, and in certain such embodiments, the periodic array of holes covers the whole core region of the fiber 730. The nature of the e-beam evaporation and ion-milling does not pose any temperature issues or problems regarding the length of the fiber structures.
The holes of the photonic crystal pattern in certain embodiments can have a variety of shapes, including but not limited to, circle, square, rectangle, and keyhole. These holes can be laid out on different types of two dimensional lattices including but not limited to a square lattice, a triangular lattice, and a hexagonal lattice. The patterns can also have non-uniformities, such as hole sizes being different in the center of the pattern as compared to the outer parts of the pattern. Similarly, the pattern can also have defects, such as missing holes, or individual holes with different sizes compared with the rest of the pattern.
In certain embodiments, the device 700 can be spliced (e.g., with fusion splicing) to another cleaved or polished fiber 740, such that the photonic crystal 710 remains in between two fibers 730, 740, as depicted in
In certain embodiments, the modes in the photonic crystal 710 can be coupled directly to optical near-fields, and the device 700 can be used as a compact near-field sensor. The optical properties of the photonic crystal 710 can change drastically when the index of refraction of the surrounding medium 750 changes. In certain such embodiments, the device 700 can be used as an index sensor. For example as schematically illustrated by
The method 900 comprises forming a first dielectric layer 1020 on an end of an optical fiber 1010 in an operational block 910. In
The method 900 further comprises forming a sacrificial layer 1030 over at least a portion of the first dielectric layer 1020 in an operational block 920. In certain embodiments, forming the sacrificial layer 1030 comprises forming a 100-nanometer chrome layer 1032 over the first dielectric layer 1020 (e.g., by e-beam evaporation), and forming a 500-nanometer gold layer 1034 over the chrome layer 1032 (e.g., by e-beam evaporation). In certain embodiments, the first dielectric layer 1020, the chrome layer 1032, and the gold layer 1034 are formed subsequently to one another in the same e-beam evaporation system. As shown in
The method 900 further comprises forming a second dielectric layer 1040 over the sacrificial layer 1030 in an operational block 930. For example, as shown in
The method 900 further comprises patterning the first dielectric layer 1020 to form a first photonic crystal slab 1050 in an operational block 940, and patterning the second dielectric layer 1040 to form a second photonic crystal slab 1060 in an operational block 950. For example, as shown in
The method 900 further comprises removing the sacrificial layer 1030 in an operational block 960. For example, as shown in
In certain embodiments, the fabrication methods described above can be used to fabricate other optical devices besides photonic crystals at the end of an optical fiber. For example, for several applications including telecommunication and fiber-based physical devices such as gyroscopes, it is desirable to have one or more sub-wavelength gratings which can manipulate the phase distribution, electric-field amplitude distribution, and polarization distribution in an optical mode.
By patterning structures such as holographic waveplates (or computer-generated holograms) on the amorphous silicon layer at the fiber tip, certain embodiments described herein can be used to convert the fundamental Gaussian mode of a single-mode fiber into higher order Laguerre-Gaussian modes. Such modes include, but are not limited to, doughnut-shaped modes and even more complex modes that can be employed in applications such as optical trapping and manipulation of microscopic particles. For some applications such as confocal microscopy, a doughnut-shaped focal point is desirable at the front focal plane of high numerical-aperture objectives. Holographic waveplates also allow the creation of modes with unconventional space-variant polarizations (conventional here refers to either linearly, circularly, or elliptically polarized), which are usually radial and azimuthal polarizations. By using these space-variant polarizations states, in certain embodiments, it is even possible to focus light to a spot smaller than the diffraction limit (see, e.g., R. Dorn et al, “Sharper focus for a radially polarized light beam,” Phys. Rev. Lett., Vol. 91, 233901 (2003)).
In certain embodiments described herein, a fiber can be formed with a holographic waveplate at its end, that has an output mode with either a useful intensity distribution such as a doughnut mode or a polarization distribution such as radial polarization. Such a mode can be used to trap and apply torque to microscopic particles in various embodiments. A doughnut-shaped mode can also increase the trapping force to trap particles with lower refractive index than the surrounding fluid. One advantage of certain embodiments described herein over standard optical trapping and manipulation methods is that the optics are all integrated into the fiber, so that it is easy to dip it into a solution, target a specific area or particle group, and move around both the fiber and the trapped particles.
In certain embodiments, one or more polarizers can be created by fabricating sub-wavelength gratings at the fiber end. One method in accordance with certain embodiments described herein can be used to obtain a conventional, linear polarizer by creating a regular one-dimensional grating pattern on the amorphous silicon layer at the fiber end. For certain applications and wavelength ranges, a metal layer can be added to the silicon layer, or the silicon layer can be completely substituted by a metal layer. These types of polarizers are usually referred to as wire-grid polarizers, schematically illustrated by
These fibers, which are patterned with a sub-wavelength structure at the end, can also be combined in certain embodiments with a second fiber, such as through splicing, or using a lower temperature gluing method. Then, the sub-wavelength structures will remain in-line in between the fibers. Certain embodiments can, for example, interface two different types of fibers with a holographic waveplate, and facilitate the coupling between the fibers. Also, certain embodiments can create a fiber polarizer, by embedding a sub-wavelength polarizer in between two fibers.
Various embodiments have been described above. Although the invention has been described with reference to these specific embodiments, the descriptions are intended to be illustrative of the invention and are not intended to be limiting. Various modifications and applications may occur to those skilled in the art without departing from the true spirit and scope of the invention as defined in the appended claims.
This application is a divisional from U.S. patent application Ser. No. 11/971,714, filed Jan. 9, 2008, incorporated in its entirety by reference herein, which claims the benefit of U.S. Provisional Patent Appl. No. 60/879,465, filed Jan. 9, 2007, which is incorporated in its entirety by reference herein, and U.S. Provisional Patent Appl. No. 61/000,003, filed Oct. 23, 2007, which is incorporated in its entirety by reference herein.
| Number | Name | Date | Kind |
|---|---|---|---|
| 3831137 | Cuomo | Aug 1974 | A |
| 4446543 | McLandrich et al. | May 1984 | A |
| 4449781 | Lightstone et al. | May 1984 | A |
| 5488504 | Worchesky et al. | Jan 1996 | A |
| 5910286 | Lipskier | Jun 1999 | A |
| 6289143 | Berthold et al. | Sep 2001 | B1 |
| 6777244 | Pepper et al. | Aug 2004 | B2 |
| 6798960 | Hamada | Sep 2004 | B2 |
| 6807342 | Fan | Oct 2004 | B2 |
| 6822784 | Fukshima et al. | Nov 2004 | B2 |
| 6885784 | Bohnert | Apr 2005 | B2 |
| 7024072 | Chen et al. | Apr 2006 | B2 |
| 7054011 | Zhu et al. | May 2006 | B2 |
| 7134343 | Suzuki et al. | Nov 2006 | B2 |
| 7155087 | Suh et al. | Dec 2006 | B2 |
| 7187816 | Huang | Mar 2007 | B2 |
| 7190869 | Jin et al. | Mar 2007 | B2 |
| 7224465 | Balachandran et al. | May 2007 | B2 |
| 7233729 | Romagnoli et al. | Jun 2007 | B2 |
| 7280265 | Miles | Oct 2007 | B2 |
| 7283716 | Park et al. | Oct 2007 | B2 |
| 7308163 | Bratkovski et al. | Dec 2007 | B2 |
| 7333703 | Hatsuda et al. | Feb 2008 | B2 |
| 7474823 | Wang et al. | Jan 2009 | B2 |
| 7483144 | Sanders | Jan 2009 | B2 |
| 7489846 | Grot et al. | Feb 2009 | B2 |
| 7526148 | Kilic et al. | Apr 2009 | B2 |
| 7545513 | Kiesel et al. | Jun 2009 | B2 |
| 7619744 | Liess | Nov 2009 | B2 |
| 7630589 | Kilic et al. | Dec 2009 | B2 |
| 7751055 | Sanders et al. | Jul 2010 | B2 |
| 7809219 | Kilic et al. | Oct 2010 | B2 |
| 7881565 | Kilic et al. | Feb 2011 | B2 |
| 7944567 | Asano | May 2011 | B2 |
| 7973936 | Dantus | Jul 2011 | B2 |
| 20020135863 | Fukshima et al. | Sep 2002 | A1 |
| 20040008934 | Takiguchi et al. | Jan 2004 | A1 |
| 20040080726 | Suh et al. | Apr 2004 | A1 |
| 20040208449 | Chen et al. | Oct 2004 | A1 |
| 20040228575 | Kim et al. | Nov 2004 | A1 |
| 20050052724 | Suzuki et al. | Mar 2005 | A1 |
| 20050169590 | Alkeskjold | Aug 2005 | A1 |
| 20050175304 | Romagnoli et al. | Aug 2005 | A1 |
| 20050186117 | Uchiyama et al. | Aug 2005 | A1 |
| 20050191025 | Kim et al. | Sep 2005 | A1 |
| 20050201660 | Grot et al. | Sep 2005 | A1 |
| 20050237602 | Yanagisawa | Oct 2005 | A1 |
| 20060227331 | Vollmer et al. | Oct 2006 | A1 |
| 20070081165 | Kilic et al. | Apr 2007 | A1 |
| 20080034866 | Kilic et al. | Feb 2008 | A1 |
| 20080089645 | Wang et al. | Apr 2008 | A1 |
| 20090028407 | Seibel et al. | Jan 2009 | A1 |
| 20090208163 | Kilic et al. | Aug 2009 | A1 |
| 20100092125 | Kilic et al. | Apr 2010 | A1 |
| 20110041616 | Kilic et al. | Feb 2011 | A1 |
| 20110268384 | Akkaya et al. | Nov 2011 | A1 |
| Number | Date | Country |
|---|---|---|
| 56-081802 | Jul 1981 | JP |
| 2002-328243 | Nov 2002 | JP |
| 2003-185864 | Jul 2003 | JP |
| 2005-045463 | Feb 2005 | JP |
| 2005-077711 | Mar 2005 | JP |
| 2008-541513 | Nov 2008 | JP |
| 2009-535977 | Oct 2009 | JP |
| 2010-516124 | May 2010 | JP |
| WO 2006119200 | Nov 2006 | WO |
| WO 2007130152 | Nov 2007 | WO |
| WO 2008086448 | Jul 2008 | WO |
| WO 2011115933 | Sep 2011 | WO |
| Number | Date | Country | |
|---|---|---|---|
| 20100092125 A1 | Apr 2010 | US |
| Number | Date | Country | |
|---|---|---|---|
| 60879465 | Jan 2007 | US | |
| 61000003 | Oct 2007 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 11971714 | Jan 2008 | US |
| Child | 12587551 | US |
