This invention relates to Raman microlasers using photonic crystals made from silicon to achieve low-loss, low-threshold Raman lasing.
Stimulated Raman scattering (SRS) has a rich and evolving history since the development of the laser. In 1962, Woodbury and Ng discovered the SRS effect at infrared frequencies. [E. J. Woodbury and W. K. Ng, Proc. IRE 50, 2347 (1962)] Hellwarth quickly described this observation as a two-photon process with a full quantum mechanical calculation. [R. W. Hellwarth, Theory of Stimulated Raman Scattering, Phys. Rev. 130, 1850 (1963)] To account for anti-Stokes generation and higher-order Raman effects, however, Garnmire et al. and Bloembergen and Shen then adopted the coupled-wave formalism to describe the stimulated Raman effect. [E. Garmire, E. Pandarese, and C. H. Townes, Coherently Driven Molecular Vibrations and Light Modulation, Phys. Rev. Lett. 11, 160 (1963); N. Bloembergen and Y. R. Shen, Coupling Between Vibrations and Light Waves in Raman Laser Media, Phys. Rev. Lett. 12, 504 (1964); Y. R. Shen and N. Bloembergen, Theory of Stimulated Brillouin and Raman Scattering, Phys. Rev. 137, A1787 (1965)] These understandings were later improved by the inclusion of self-focusing to account for the much larger gain observed in SRS.
Recent developments include coupling a high Q (“Q” is a quality factor) silica microsphere to an optical fiber to achieve a minimum threshold of 62 μW, an example of which is illustrated in
In a different line of researches that does not use SRS, various researches have demonstrated that a laser-reflowed silicon oxide microresonator with additional Er3+ doping can achieve low-threshold lasing. [A. Polman, B. Min, J. Kalkman, T. J. Kippenberg, and K. J. Vahala, Ultralow-threshold erbium-implanted toroidal microlaser on silicon, App. Phys. Lett. 84 (7), 1037, 2004)] Concurrently, Claps et al. have demonstrated a small but first-ever Raman amplification in silicon on-chip waveguides for photonic integrated circuit applications. [R. Claps, D. Dimitropoulos, V. Raghunathan, Y. Han, and B. Jalali, Observation of stimulated Raman amplification in silicon waveguides, Optics Express 11 (15), 1731 (2003)]
However, presently, the development of sizable gain in silicon photonic integrated circuits has yet to be demonstrated. This is suspected due to un-optimized phase matching design of the optical structures. [R. H. Stolen and E. P. Ippen, Raman gain in glass optical waveguides, App. Phys. Lett. 22 (6), 276 (1973)]
Various embodiments of the present invention provide tunable laser devices and methods of manufacturing such devices. In particular, the laser device may include a layer of photonic crystal having a lattice of air-holes with defects that form an optical waveguide. The waveguide has a cross-sectional area whose dimensions are in sub-wavelength ranges, wherein the cross-sectional area is perpendicular to the propagation direction of light in the waveguide. The waveguide receives pump light and outputs Stokes light through Raman scattering. The frequencies of the pump light and the Stokes light can be selected from slow group velocity modes of the pump light and Stokes light in the waveguide. The slow group velocity can be about 1/100 of the speed of light.
The waveguide can be integrated with CMOS microelectronic devices. For instance, a p-i-n (p-type, intrinsic, n-type) diode can be integrated with the waveguide to achieve a continuous wave lasing in the optical waveguide. The waveguide can also receive pulsed pump light.
In some embodiments, the waveguide may include a pair of optically coupled cavities, whose geometries are substantially identical to each other. The cavities can be defined to cause a frequency-splitting difference between a frequency of the pump light and a frequency of the Stokes light to correspond to an optical phonon frequency in silicon for Raman scattering lasing or amplification. The optical phonon frequency is about 15.6 THz in single-crystal silicon at the room temperature.
A one-dimensional photonic crystal shaped like a bar can also form the laser device. In such embodiments, air-holes with defects can form cavities. Fabrication processes to manufacture various embodiments of the present invention are also described.
Embodiments of the present invention will be best understood when read in reference to the accompanying figures wherein:
SRS is a linear inelastic two-photon process, where an incident photon interacts with an excited state of the material. In various embodiments of the present invention, which include the use of photonic crystals made of silicon, the excited state of the material refers to the longitudinal optical (LO) and transversal optical (TO) phonons of crystal silicon. In such embodiments, the strongest Stokes peak arises from single first-order Raman-phonon (threefold degenerate) at the Brillouin zone center of silicon. A microscopic description that depicts the change in the average number of photons ns at the Stokes wavelength ωs with respect to the longitudinal distance z is:
where GR is the Raman gain, αs an attenuation coefficient, μ the permeability,
the transition rate, and ρi and ρf the initial and final state populations, respectively. For ns and np (the average number of photons at ωp) significantly greater than 1,
and thus the Raman gain GR is ∝ np. For large ns and np, a mesoscopic classical description with Maxwell equations using nonlinear polarizations P(3) can also be used. The wave equations describing the interactions are:
Specifically, Ps(3)=χjkmn(3)EpE*pEs, where χjkmn(3) is the third-order fourth-rank Raman susceptibility with {j,k,m,n}={x,y,z}. The resonant terms in Ps(3) give rise to SRS, while the non-resonant terms add to self-focusing and field-induced birefringence. The Ep and Es are the electric fields at the pump and Stokes wavelengths, respectively. With χjkmn(3) obtained from bulk material properties, Equations (2) and (3) can be turned into discrete forms in the time-domain for direct ab initio numerical calculations of the nonlinear response.
As an approximation to the direct solution of this wave interpretation, the coupled-mode theory can be used to estimate the stimulated Raman gain. In particular, under the assumption of weak coupling between the pump and Stokes waves, the mode amplitudes can be given as:
where the self-coupling terms are neglected, Ep, Es and Ea denote the pump, and Stokes and anti-Stokes field amplitudes are denoted, respectively, as Ip=|Ep|2, Is=|Ep|2. βab denotes the non-resonant terms and resonant terms with no frequency dependence. κab denotes the resonant overall coupling coefficients (integrated spatially) between the modes. By determining κps(ωs) Equations (4) and (5) can be employed to determine the SRS gain. Intrinsic loss due to two-photon absorption (TPA) is assumed to be small based on the measured TPA coefficients in silicon and at pump powers on the order of 1 W. The role of TPA-induced free carrier absorption is also reduced in sub-wavelength silicon-on-insulator (SOI) waveguides of various embodiments of the present invention due to significantly shorter lifetime (compared to the recombination lifetime). This results in lower overall carrier densities.
Dimitropoulos et al. have derived a specialized form of Equations (5) to determine the Raman gain GR in waveguides. [D. Dimitropoulos, B. Houshmand, R. Claps, and B. Jalali, Coupled-mode theory of the Raman effect in silicon-on-insulator waveguides, Optics Lett. 28 (20), 1954 (2003)] In particular, GR has an approximate 1/(modal area)3/4 dependence; that is, the SRS gain increases with decreasing modal areas, such as from high-index contrast waveguide structures. In various embodiments of the present invention, as will be described in detail later, enhancements through smaller modal areas Am and length scales x, Purcell enhancements and/or slow group velocities afforded by photonic crystal structures permit increased amplification with significantly smaller device length scales.
An example photonic crystal 201 manufactured in accordance with various embodiments of the present invention is illustrated in
The lattice of air-holes also forms basic patterns 205. The example in
The Raman scattering is further described using the example waveguide 207 shown in
The small cross-sectional area of the waveguide 207 causes optical field densities to increase and causes the gain of the Raman scattering and lasing to increase as well. In addition to this enhancement, various embodiments of the present invention take advantage of slow light phenomena. That is, at the photonic band edge, photons experience multiple reflections and move very slowly through the material structure. In photonic crystal structures, line-defects in the periodic lattice permit guided-mode bands within the band gap, as shown in
With slow group velocities, it is possible to reduce the interaction length by (vg/c)2. In particular, for group velocities on order of 10−2 c, interaction lengths—between the Stokes and pump modes, for example—on order of 104 times smaller than conventional lasers can be obtained. For the same operation power, the same gain can be obtained by the time-averaged Poynting power density P (˜vg ε|E|2) incident on the photonic crystal structure. A decrease in vg leads to a corresponding increase in ε|E|2 and in the Raman gain coefficient. These line-defect waveguides can be designed for two modes (i.e., the pump and Stokes modes) to be supported within the band gap of various embodiments of the present invention.
Instead of using a lattice of air-holes to form photonic crystals, as shown in
Moreover, microcavities further enhance the same waveguide geometry with the Purcell factor η and also reduce the interaction lengths (effective length≈λQ/2πn ) for an increased amplification of 104. (More detailed descriptions on the Purcell factor are provided later.) This is obtained by forming two optically coupled microcavities that are formed substantially identical to each other in order to split the degenerate modes according to the strength of the coupling. Such a configuration supports both the pump and Stokes wavelengths in the microcavities, especially when they are formed in series. (Note that having two differently sized defects side-by-side in parallel requires special symmetries to force input/outputs into 2 of the 4 ports in the system.)
In
More specifically, optically coupled microcavities formed in photonic crystals can obtain a good quality factor (Q) with ultra-small sub-wavelength modal volumes (Vm). These two factors can be perceived physically as long photon lifetimes and high field intensities per photon, respectively, contributing to microcavity-enhanced processes such as cavity quantum electrodynamics (QED) and laser physics. Placed in a cavity, the Raman phenomenon on resonance is greatly enhanced by the increased final density of states per unit volume and unit frequency. This is expressed as the Purcell factor:
η=ρc/ρo≈(3λ3/4π2) (Q/Vm) (7)
where ρc and ρo are the densities of states for the cavity and free space, respectively. Using microdroplets and, more recently, silica microspheres (e.g., the structure shown in
The amplification gain improved by the coupled cavities is further enhanced by integrating a p-i-n (p-type, intrinsic, n-type) junction diode with the photonic crystal as illustrated in
Microcavities can also be formed in a two-dimensional (2D) photonic crystal 901 as illustrated in
The 2D photonic crystal 901 also includes an input port 903 and an output port 905. The input port 903 can receive pump light 911, as well as Stokes light 909 (at a low amplitude). In particular, another waveguide such as the one shown in
As a further theoretical derivation, for laser oscillation, the gain condition requires the gain GR to exceed the losses α for initiation of oscillation: (GR-α)>0. (The other condition, the phase condition, determines the lasing frequency in a cavity.) The Raman gain GR is enhanced by the Purcell factor, as illustrated in Equation (7) above, and has a Q/Vm dependence. The loss α has a dependence understood from the definition of
Equating GR with α for the lasing threshold, the estimated dependence on the lasing threshold Pth can be derived as:
where Qs and Qp are the quality factor at the Stokes and pump wavelengths, respectively, (and approximately on the same order), gR the bulk Raman gain coefficient, λs and λp the Stokes and pump wavelengths respectively, Γ the modal overlap between Stokes and pump, Ap the pump modal area, and ns the effective index at Stokes wavelengths.
According to Equation (8), the lasing threshold depends on the term
The characteristic of this threshold is similar to that derived for whispering gallery modes in microspheres, where a Vm/Q2 dependence is also observed. For cavity line widths significantly smaller than the homogenous line width of the scattering process (when Fermi's golden rule breaks down), the above estimate, and that of Equation (7), needs to be further enhanced by an approximation using the density of states and the transition rate per mode. Likewise, in the design of the microcavity for lasing, zero-threshold lasers can be achieved, where the spontaneous emission enhancement becomes beneficial. Additionally, laser modulations at higher frequencies are also possible with the reduced mode volumes.
Based on the above described theoretical derivation, two example considerations in designing microcavities of various embodiments of present invention are: (1) to find modes which have odd symmetry about mirror planes normal to their dominant Fourier components, and (2) to smoothen the dielectric variation away from the defect. With microcavities of various embodiments of the present invention, a Vm˜a factor of two larger and a lower Q on order of 104 can be achieved. This allows Q/Vm ratios on order of 6×104 to 1×105 μm−3, with Purcell enhancement factors on order of 2×104. The resultant lasing threshold, based on the estimate from Equation (8), is obtained on order of 30 μW. This estimate is made based on the above Q and Vm values, a mode overlap integral ˜0.8, silicon bulk Raman gain coefficients, and the modal volumes and cross-sectional areas involved. This estimate, using Equation (8), does not consider further enhancements on the gain coefficient from microcavity quantum electrodynamics. Yet, with the approximations, the threshold is already immediately comparable with the lasing threshold from Raman microspheres, where the minimum lasing threshold is reported at 62 μW. With further improvements, as provided by various embodiments of the present invention, the Q and Vm values drop the lasing threshold to the level of a few μW. These low-thresholds allow a directly compatible on-chip laser source with tunable wavelengths in silicon electronic-photonic integrated circuits.
Having a pair of optically coupled microcavities is not required to obtain a high amplification gain. For instance, a microcavity can be formed by linear defects (e.g., linearly missing air-holes), as illustrated in
The 2D photonic crystal 1301 is an air-bridged triangular-latticed photonic crystal layer with a thickness of 0.6a, and the radius of its air-holes is 0.29a (e.g., 300 nm), where a is the lattice period. The photonic band gap in the crystal 1031 for transverse-electric-like (TE-like) modes is around 0.25˜0.32 [c/a] in frequency. For small cavities such as L3 and L4 (i.e., linearly missing three-hole structure, L3, and linearly missing four-hole structure, L4), the calculated values of a and λ are large, which may not match the telecommunication applications (around 1550 nm wavelength). For example, in the L3 cavity, S1=0.15a, a=685 nm, λpump=2266 nm and λStokes=2568 nm. Finally, two even modes in a single L5 cavity are used as the pump and Stokes modes for Raman lasing, respectively. Here, an L5 cavity is missing five (5) air-holes, and, therefore, the length of the cavity can be on order of several micrometers (e.g., 2.5 microns). Since the width of the cavity is on order of sub-wavelength, the surface area of the cavity can be on order of several microns-squared.
Table 1 summarizes fine-tuning the values of shift S1 of two air-holes at the edges of the microcavity. By increasing the value of S1, the calculated lattice period a decreases and the resonant wavelength λ also decreases due to the constant optical phonon frequency. The quality factors increase because the electric field profile is close to Gaussian function and has less leakage.
In
With the high Q/Vm ratios, other light scattering phenomena such as stimulated Brillouin scattering can also be achieved (although it has a 102 smaller steady-state gain in silicon and a significantly narrower gain spectrum than Raman). Higher-order Raman scattering, four-wave mixing and anti-Stokes generation are also contemplated by various embodiments of the present invention. For this reason, the photonic crystal waveguides of various embodiments of the present invention can be designed to support only a particular mode in the band gap, so as to prevent other wavelengths to appear in the Raman lasing signal. Alternatively, specific multiple modes can be intentionally designed in the band gap so that these output wavelengths can be selected as desired. In addition, self-focusing can be estimated in the various embodiments, and the optical and Raman-induced Kerr effects (field-induced birefringence) can be employed in designing various microcavities of the present invention due to the large field intensities within the sub-wavelength microcavities. Operating in this regime of high Q/Vm and high peak intensities, this host of nonlinearities can be actively implemented on-chip, for fundamental novel applications such as multiple-wavelength lasers or single biomolecular detection at the cavity.
With low intrinsic material losses in silicon-based photonic crystal at near-infrared wavelengths, optical losses were previously dominated by confinement losses in 2D planar photonic crystal structures. Radiation losses to the continuum modes by point-defect microcavities had earlier Q factors of several hundred. By forming microcavities in accordance with various embodiments of the present invention, radiation losses are significantly suppressed with Q factors on order of 102 or better; or equivalent, energy losses per cycle a factor of 102 smaller. These lower radiation losses, and hence longer photon confinement lifetimes in a cavity, permit low threshold powers for lasing. Low-loss photonic crystal waveguides with designs such as waveguides with guided defect modes far from the band edges or high-quality fabrication etching can achieve transmission losses as low as 1.8 dB/mm. 3D photonic crystals permit losses to be reduced down to surface roughness effects (as mirror symmetry is not broken) although 3D structures are less amendable to fabrication.
Various embodiments of the present invention differ from recently developed silica microspheres and Er3+-doped silica microtoroids. In contrast to silica microspheres, silicon 2D photonic crystal microcavities of various embodiments of the present invention: (1) are suitable for planar on-chip CMOS fabrication, (2) are significantly more compact (with a modal volume ˜105 times smaller), and (3) can interface directly with silicon optical waveguides and other on-chip photonic-electronic circuitry.
Moreover, in contrast with Er3+-doped silica microtoroids, silicon 2D photonic crystal microcavities of various embodiments of the present invention: (1) have lasing tunable across selectable waveguides, (2) do not require additional laser-treatment (which was used to achieve ultra-high Q in silica or Er3+ implantation), and (3) have orders of magnitude faster modulation speeds due to their ˜104 smaller modal volumes and not being limited by Er3+ lifetimes (on order 10-12 ms). The lower Q in silicon photonic crystal microcavities in comparison with the silica structures adversely affects the lasing threshold (varying at 1/Q2 as first derived in Equation (8)); however, this is compensated by the 104 larger bulk Raman gain coefficient in silicon and the 104 to 105 smaller modal volumes to bring the lasing threshold back on comparable grounds of order several to 10s of 1 W. Various embodiments of the present invention also relate to low-threshold Raman lasing in silicon 2D photonic crystal microcavities with high Q/Vm ratios.
For various embodiments of the present invention, in order to provide a net Raman gain, TPA induced the free-carrier absorption phenomenon can also be addressed using pulsed operations, where the carrier lifetime is much larger than the pulse width and much less than the pulse period. In particular, a photonic crystal 1701 with one or more microcavities is coupled to a multiplexer (MUX) 1703. The MUX 1703 receives its input from a polarization controller 1705 that combines inputs from a pulsed pump laser 1707 and a continuous wave (CW) Stokes laser 1709. The output from the photonic crystal 1701 is then input to an optical spectrum analyzer (e.g., a detector) 1711.
Various embodiments and advantages of the present invention are apparent from the detailed specification, and, thus, it is intended by the appended claims to cover all such features and advantages of the invention which fall within the true spirit and scope of the invention. Further, since numerous modifications and variations will readily occur to those skilled in the art, it is not desired to limit the invention to the exact construction and operation illustrated and described, and, accordingly, all suitable modifications and equivalents may be resorted to falling within the scope of the invention. For example, the slow speed light technique described in connection with
This application claims the priority of U.S. Provisional Application 60/589,903 filed on Jul. 20, 2004, which is incorporated herein by reference in its entirety.
Number | Date | Country | |
---|---|---|---|
60589903 | Jul 2004 | US |