Method and apparatus for wavelength conversion and switching

[0003] This invention pertains generally to the field of optical electronic devices and particularly to optical devices of the type used in optical communication networks.

[0004] Optical fiber communication networks are being increasingly utilized for both voice and data communication. Wavelength division multiplexing (WDM) has become the method of choice for increasing the data-carrying capacity of fiber optic systems. With WDM, many (typically from 4 to 40) data streams are carried along a single fiber, using a separate wavelength for each. Redirecting the flow of information in conventional fiber optic systems requires that the optical data streams be separated, converted to electronic signals, and then routed electronically to the next span of fiber in the network. The conversions from light to electronics to light greatly limits the speed and flexibility of present fiber optic systems. To better utilize the full potential of such optical networks, efforts have been directed to the development of optical switching devices which do not require conversion of the optical signal to an electrical signal. The challenges presented for such optical switching devices are particularly significant in communication networks which utilize WDM. In certain WDM applications, it is necessary to route carriers at different wavelengths to different locations based on the wavelength of the carrier. In other applications, it may be desirable to transfer the signal from a carrier at one wavelength to a carrier at a different wavelength.

[0005] In accordance with the invention, switching of optical signals and wavelength interchange of optical signals in optical communication systems is carried out without requiring conversion of the optical signals to electrical signals, thereby allowing the capabilities of the optical switching networks to be more fully utilized and minimizing the cost of electronic components. The invention may be utilized for wavelength sensitive routers, wavelength interchange cross connects, wavelength add/drop multiplexers, spectral inverters for dispersion compensation, facilitation of wavelength hopping for data encryption, inter-conversions between wavelength division and time division multiplexing, and may be extended to three dimensions to permit logic, arithmetic, and wavelength-shifting operations to be carried out on two-dimensional optical arrays for image analysis or to implement free-space interconnects.

[0006] The invention utilizes a lattice body comprising sensitized regions arranged in a two-dimensional array in a matrix material, wherein the sensitized regions differ from the matrix material in the sign of the second order susceptibility. Preferably, the sensitized regions and matrix material have the same first order susceptibility. The lattice body may be formed of various semiconductor materials and ferroelectrics such as LiNbO₃. Two optical signals at different carrier frequencies ω₁and ω₂are coupled into the lattice body to interact with the array. A pump signal having a frequency ω_p=ω₁+ω₂is also coupled into the lattice body with the two optical signals. These signals may be collinear with one another or may intersect in the lattice body. The spacing and pattern of the sensitized regions in the array in the lattice body are selected to interact with the three optical signals so that an optical signal exits from the lattice body that is angularly displaced from the direction of propagation of the input optical signals and that comprises an optical signal at a carrier frequency ω₂that carries the information on the input signal at the carrier frequency ω₁. Another optical signal exits from the lattice body angularly displaced from the direction of propagation of the input optical signals that comprises an optical signal at a carrier frequency of ω₁that carries the information on the input signal at the carrier frequency ω₂. These two angularly displaced signals may then be received and separated from each other, as well as from the input signals at the carrier frequencies ω₁, ω₂and ω_pand cap which pass through the lattice body and exit without deflection from the initial input direction of propagation. When the pump signal is not applied to the lattice body, no wavelength interchange occurs, and the input signals at the frequencies ω₁and ω₂exit from the lattice body without change in the direction of propagation. Thus, by switching the pump signal on and off, the wavelength interchanged angularly diverging signals can be switched on and off.

[0007] Further, by utilizing a pump light signal intensity that has a larger magnitude than that of the input optical signals, the wavelength interchanged outputs can have a greater optical intensity than that of the input signals, thus effectively providing signal amplification.

[0008] The lattice body can comprise sensitized regions formed, for example, in ferroelectrics by periodic poling of the ferroelectric material, or in semiconductors by selective disordering, to provide an opposite sign of the second order susceptibility in the sensitized regions from that of the matrix material. Lateral waveguides may be formed on the lateral sides of the lattice body to receive and guide light that has been deflected from the longitudinal direction of the incoming light by interaction with the lattice.

[0009] Further objects, features and advantages of the invention will be apparent from the following detailed description when taken in conjunction with the accompanying drawings.

[0010] In the drawings:

[0011] FIG. 1 is a schematic view of an optical communications network exemplifying the invention.

[0012] FIG. 2 is a simplified perspective view of a lattice body formed in accordance with the invention.

[0013] FIG. 3 a diagram showing a reciprocal space representation of a non-linear lattice designed to perform wavelength interchange.

[0014] FIG. 4 is a graph of required real lattice spacings r_aand r_bversus of the half exit angle θ_afor wavelength interchange of two signals at λ_a=1550 nm and λ_b=1550.8 nm.

[0015] FIG. 5 is a diagram illustrating a lattice structure utilizing lateral waveguides for collecting and propagating the light deflected by the lattice.

[0016] FIG. 6 is a diagram illustrating formation of the sensitized regions in a lattice body utilizing poling of a ferroelectric material, LiNbO₃.

[0017] FIG. 7 is a schematic view of another embodiment of apparatus for wavelength interchange in accordance with the invention.

[0018] FIG. 8 is a schematic view of a further embodiment of apparatus for wavelength interchange in accordance with the invention.

[0019] With reference to the drawings, a simplified diagram of an optical communications network exemplifying the application of the invention is shown generally at 10 in FIG. 1. The network 10 includes a source 11 which provides wavelength division multiplexed signal information onto an optical waveguide such as a fiber optic cable 12. The signal on the cable 12 includes at least two carriers at wavelengths λ₁and λ₂(equivalently, at frequencies (θ₁and θ₂) which are separately modulated. The fiber optical cable 12 is connected to a coupler 14 which is also connected to a waveguide such as a fiber optic cable 15 that is selectively supplied with an optical signal at a wavelength λ_p(frequency ω_p) from a pump source 17. The coupler 14 provides an output on an input optical waveguide 18, e.g., a fiber optic cable, which is composed of the optical signals at the wavelengths λ₁, λ₂and λ_p. The cable 18 is connected to an optical wavelength interchange apparatus 20 of the invention. The outputs of the device 20 are provided on a first output optical waveguide 22 such as an optical cable, which is connected to a utilization system 23 at which the signal is detected and utilized (and which may comprise a further network and further switching devices), on a second output optical waveguide 25 such as a fiber optic cable which is connected to a utilization system 26 (which again may be a further network), and on a third output optical waveguide 28 such as a fiber optic cable which is connected to a utilization system 29 (which may be a further optical network). In accordance with the invention, the output from the device 20 on the first output waveguide optical cable 22 comprises the unmodified input optical signal that was on the input waveguide cable 18, which includes the carriers at the wavelengths λ₁, λ₂and the pump λ_p. As explained further below, when the pump source 17 is activated to apply the optical signal at the wavelength λ_pwhich corresponds to a frequency ω_p=ω₁+ω₂, the output optical signals on the two output optical waveguides 25 and 28 comprise optical signals at the wavelengths λ₁, and λ₂, respectively, which are thereby separated from one another. Further, the signals on the carriers are interchanged so that the optical signal at the carrier wavelength λ₂now carries the information originally carried by the carrier at the wavelength λ₁, and the optical signal at the carrier wavelength λ₁now carries the information that was carried by the carrier at the wavelength λ₂. Thus, the present invention selectively separates the carriers at the wavelengths λ₁and λ₂in the input waveguide signal from each other and interchanges the signal information on these carriers in the output signals. In the figures, the superscripts “A” and “B” indicate the signal information (modulation) of the carrier signals at the wavelengths λ₁and λ₂.

[0020] In accordance with the present invention, a two dimensional (2-D) second-order nonlinear lattice is utilized in the device 20 for performing a simultaneous one-step optical carrier (i.e., wavelength) interchange between a selected pair (or selected pairs) of input optical signals from the cable 18, such as those carried in a fiber optic wavelength-division multiplexed communication system. In accordance with the invention, the interchange can be made to occur by means of two simultaneous difference-frequency generation (DFG) processes. The physical process takes advantage of the two-dimensional nature of the nonlinear lattice to selectively both quasi-phase match as well as spatially segregate the interchanged optical signal pair from the unconverted inputs ——the latter being essential since not all of the two input signals wavelength-exchange. The choice of the pump frequency is uniquely defined by the sum of the input signal frequencies. In the absence of the pump beam, no wavelength interchange takes place. Applications of the wavelength interchange process of the invention include wavelength sensitive routers, wavelength interchange cross connects, wavelength add/drop multiplexers, and spectral inverters for dispersion compensation. A particular advantage of the one-step optical carrier interchange process of the invention is that it simultaneously segregates the inter-converted signals from the unconverted ones, thus eliminating coherent in-band crosstalk.

[0021] A schematic diagram of a 2-D second-order nonlinear lattice defined in a solid state body 35, and the resulting interchange process, is shown in FIG. 2. The lattice body 35 is incorporated in the wavelength interchange device 20 of FIG. 1, as discussed further below. In contrast to a conventional photonic crystal which possesses a periodic variation in the linear electric susceptibility, χ⁽¹⁾, for the purpose of introducing a photonic band gap, the lattice body 35 shown in FIG. 2 possesses a 2-D periodic variation in the second-order nonlinear electric susceptibility χ⁽²⁾. The linear electric susceptibility in the lattice body structure 35 is preferably spatially invariant, and the lattice body 35 is preferably formed of a two-dimensional array of sensitized regions 37 (shown dark in FIG. 2) embedded in and surrounded by a matrix material 38 (shown light in FIG. 1). The 2-Dχ⁽²⁾pattern of sensitized regions 37 in a matrix 38 that comprises the lattice body 35 can be produced in, e.g., ferroelectrics, such as lithium niobate (LiNbO₃), using photolithographic and electric field poling procedures such as are used in one-dimensional variation structures. In such structures, the matrix 38 and the sensitized regions 37 are formed integrally of the same material, differing only in the sign of the second order non-linear susceptibility χ⁽²⁾. See, e.g., M. L. Bortz, et al., Optics Lett. Vol. 20, pp. 49, et seq. 1995. Appropriate lattices can also be formed in a semiconductor matrix, e.g., GaAs, as by selective disordering.

[0022] The governing equation for a material with a dominant second-order nonlinearity is $\begin{array}{cc}{\nabla }^2\overrightarrow{E}\left(\overrightarrow{r},t\right)-\frac{n^2}{c^2}\frac{{\partial }^2\overrightarrow{E}\left(\overrightarrow{r},t\right)}{\partial t^2}={\mu }_0{𝒳}^{\left(2\right)}\left(\overrightarrow{r}\right)\frac{{\partial }^2\overrightarrow{E}\left(\overrightarrow{r},t\right)}{\partial t^2}& \left(1\right)\end{array}$

[0023] where χ⁽²⁾({right arrow over (r)}) is the spatially varying instantaneous second-order nonlinearity. We assume an electric field that is composed of three time-harmonic plane waves given by $\begin{array}{cc}\overrightarrow{E}\left(\overrightarrow{r},t\right)=\frac{1}{2}\left\{A_1\left(\overrightarrow{r},t\right)\exp \left[i\left({\omega }_{1}t-{\overrightarrow{k}}_1·\overrightarrow{r}\right)\right]+A_2\left(\overrightarrow{r},t\right)\exp \left[i\left({\omega }_{2}t-{\overrightarrow{k}}_2·\overrightarrow{r}\right)\right]+A_3\left(\overrightarrow{r},t\right)\exp \left[i\left({\omega }_{3}t-{\overrightarrow{k}}_3·\overrightarrow{r}\right)\right]+c.c.\right\}& \left(2\right)\end{array}$

[0024] where ωi is the angular frequency and {right arrow over (k)}_i(ω_i) is the propagation constant vector of the i^thoptical wave, and each frequency-wavenumber pair satisfies the linear medium dispersion relation ${\left|\overrightarrow{k_i}\right|}^2-\frac{n_i^2}{c^2}{\omega }_i^2=0.$

[0025] Substituting equation (2) into (1) and assuming the slowly varying envelope approximation, the DFG resonance condition ω3=ω1 -ω2 yields the following set of coupled time-invariant equations (and their complex conjugates) that govern the spatial evolution of the wave amplitudes: $\begin{array}{cc}\overrightarrow{k_1}·\nabla A_1^{*}=\frac{1}{2}i{\mu }_0{𝒳}^{\left(2\right)}\left(\overrightarrow{r}\right){\omega }_1^2{A}_2^{*}{A}_3^{*}\exp \left[-i\left(\overrightarrow{k_1}-\overrightarrow{k_2}-\overrightarrow{k_3}\right)·\overrightarrow{r}\right]\overrightarrow{k_2}·\nabla A_2=-\frac{1}{2}i{\mu }_0{𝒳}^{\left(2\right)}\left(\overrightarrow{r}\right){\omega }_2^2{A}_1{A}_3^{*}\exp \left[-i\left(\overrightarrow{k_1}-\overrightarrow{k_2}-\overrightarrow{k_3}\right)·\overrightarrow{r}\right]\overrightarrow{k_3}·\nabla A_3=-\frac{1}{2}i{\mu }_0{𝒳}^{\left(2\right)}\left(\overrightarrow{r}\right){\omega }_3^2{A}_1{A}_2^{*}\exp \left[-i\left(\overrightarrow{k_1}-\overrightarrow{k_2}-\overrightarrow{k_3}\right)·\overrightarrow{r}\right]& \left(3\right)\end{array}$

[0026] The nonlinear susceptibility χ⁽²⁾({right arrow over (r)}) can be expanded as a Fourier series ${𝒳}_{\mathrm{bulk}}^{\left(2\right)}\sum _G{a}_G\exp \left(i\overrightarrow{G}·\overrightarrow{r}\right).$

[0027] The required frequency and phase matching conditions for normal dispersive materials can now be expressed as $\begin{array}{cc}{\omega }_3={\omega }_1^{*}-{\omega }_2\overrightarrow{k_3}\left({\omega }_3\right)=\overrightarrow{k_1}\left({\omega }_1\right)-\overrightarrow{k_2}\left({\omega }_2\right)-\overrightarrow{G}& \left(4\right)\end{array}$

[0028] Here, {right arrow over (G)} is the phase matching wave vector provided by the periodic nonlinearity χ⁽²⁾({right arrow over (r)}) of the material. It can be shown (see, e.g., C. Kittel, Introduction to Solid State Physics, Wesley, New York, 1976) that if {right arrow over (a)} and {right arrow over (b)} are the real space basis vectors in the xy-plane of a two-dimensional periodic nonlinearity, then {right arrow over (G)}, a reciprocal lattice vector, is given by

[0029] {right arrow over (G)}=m{right arrow over (A)}+n{right arrow over (B)}  (5)

[0030] where m, n=0,1,2 . . . , {right arrow over (A)}=2π{right arrow over (b)}×{right arrow over (c)}/{right arrow over (a)}×{right arrow over (b)}×{right arrow over (c)} and {right arrow over (B)}=2π{right arrow over (c)}×{right arrow over (a)}/{right arrow over (a)}×{right arrow over (b)}×{right arrow over (c)}, and {right arrow over (c)}={right arrow over (z)}.

[0031] The angle (2θ) between the input collinear waves ω1 and ω2) and the output wave (ω3) is related to the magnitudes of the wavevectors and the reciprocal lattice vector. From this relationship, we have derived the following general nonlinear Bragg condition for DFG (ω3=ω1-ω2) as well as SFG (sum-frequency generation: ω3=ω1+ω2): $\begin{array}{cc}{\left(\frac{{\lambda }_3}{n_3}\right)}_{\mathrm{SFG}·\mathrm{DFG}}=\frac{2\pi }{\left|G\right|}\sqrt{{\left[1-\left(\frac{n_1{\lambda }_2\pm n_2{\lambda }_1}{n_3\left({\lambda }_2\pm {\lambda }_1\right)}\right)\right]}^2+4\left(\frac{n_1{\lambda }_2\pm n_2{\lambda }_1}{n_3\left({\lambda }_2\pm {\lambda }_1\right)}\right){\sin }^2\left(\theta \right)}& \left(6\right)\end{array}$

[0032] where λi is the free space wavelength and ni is the refractive index of the material for the i^thoptical wave. Here, (±) is replaced by (+) for SFG and by (−) for DFG.

[0033] Consider two collinear optical data streams, EA(ωA) and EB(ωB), that are to be wavelength-interchanged (resulting in EA(ωB) and EB(ωA)). To accomplish the interchange, difference-frequency generation with a pump field Ep(ωp=ωA+ωB) is used, which is also collinear with the input signals. The electric field and its subscript designate the data stream information packet, and the angular frequency and its subscript give its carrier identity. A nonlinear version of the Ewald construction (see V. Berger, Phys. Rev. Lett., Vol. 81, 1998, pp. 4136 et seq.) can be used to determine the two periodicities of the lattice necessary to produce the two phase matching conditions between the pump and each of the input signals A and B, given by: $\begin{array}{cc} {\omega }_A={\omega }_p-{\omega }_B\overrightarrow{k_A^{\prime }}\left({\omega }_A\right)=\overrightarrow{k_p}\left({\omega }_p\right)-\overrightarrow{k_B}\left({\omega }_B\right)-\overrightarrow{G_A}{\omega }_B={\omega }_p-{\omega }_A\overrightarrow{k_B^{\prime }}\left({\omega }_B\right)=\overrightarrow{k_p}\left({\omega }_p\right)-\overrightarrow{k_A}\left({\omega }_A\right)-\overrightarrow{G_B}& \left(7\right)\end{array}$

[0034] where {right arrow over (k_A¹+L )}(ω_A) is the wavevector of the new outgoing optical field generated by difference-frequency mixing between the pump and the input field EB(ωB) and now frequency-shifted to ωA. Similarly, {right arrow over (k_B¹+L )}(ω_B) is the outgoing wavevector corresponding to a frequency shifting of data stream EA(ωA→ωB).

[0035] To simplify the explanation, the two simultaneous processes may be analyzed independently. FIG. 3 is a representation of the reciprocal lattice of the nonlinear 2-D direct lattice. The wavevector difference of the two collinear optical fields Ep(ωp) and EB(ωB), {right arrow over (k_p+L )}(ωp)-{right arrow over (k_B+L )}(ωB), is drawn in FIG. 3 with its endpoint on a reciprocal lattice point. The magnitude of the wavevector for the frequency exchanged output, {right arrow over (k_A¹+L )}(ωA), is represented as the radius of a circle (Ewald sphere in two dimensions) centered at the origin of the difference wavevector, {right arrow over (k_p+L )}(ωp)-{right arrow over (k_B+L )}(ω_B) The wavevectors {right arrow over (k_A¹+L )}(ω_A) and {right arrow over (k_A+L )}(ω_A) are identical in magnitude but vary only in direction. If the Ewald sphere intersects a reciprocal lattice point, the reciprocal lattice vector {right arrow over (G_A+L )} completes the quasi-phase matching for the difference-frequency generated light. The two Ewald spheres intersect at one common point only. The difference wavevector {right arrow over (k_p+L )}-{right arrow over (k_A+L )}, which overlaps {right arrow over (k_p+L )}-{right arrow over (k_B+L )}, is slightly displaced in FIG. 3 for purposes of visualization. The modulated light EB(ωA) will exit the lattice along the direction defined by {right arrow over (k_A¹+L )}(ω_A), a direction different from that of the input waves and pump. An identical analysis can be done for the second input EA(ωA)→EA(ωB) which is also shown in FIG. 3. It is also possible to design the lattice body such that the wavelength-interchanged outputs exit the lattice body on the same side (i.e., to the left or the right of the unconverted output signal stream, and at different angles so that the two wavelength interchanged outputs can be separated).

[0036] It can be shown that the real lattice spacings rA and r_bin the two dimensions of the lattice are given by the relation $\begin{array}{cc}{r}_{A,B}=\frac{2\pi }{\left|\overrightarrow{G_{A,B}}\right|\sin \left(\phi \right)}& \left(8\right)\end{array}$

[0037] where φ is the exterior angle subtended by {right arrow over (G_A+L )} and {right arrow over (G_B+L )} as shown in FIG. 3.

[0038] Furthermore, it can be shown that the interior angle subtended by rA and rb is also equal to φ. The following expression relates φ to the input wavelengths and desired exit angles: $\begin{array}{cc}\sin \left(\phi \right)=\left|\sqrt{{\left(\frac{\left|\overrightarrow{k_A}\right|}{\left|\overrightarrow{G_A}\right|}\sin \left(2{\theta }_A\right)\right)}^2-{\left(\frac{\left|\overrightarrow{k_A}\right|}{\left|\overrightarrow{G_A}\right|}\sin \left(2{\theta }_A\right)\frac{\left|\overrightarrow{k_B}\right|}{\left|\overrightarrow{G_B}\right|}\sin \left(2{\theta }_B\right)\right)}^2}\pm \sqrt{{\left(\frac{\left|\overrightarrow{k_B}\right|}{\left|\overrightarrow{G_B}\right|}\sin \left(2{\theta }_B\right)\right)}^2-{\left(\frac{\left|\overrightarrow{k_A}\right|}{\left|\overrightarrow{G_A}\right|}\sin \left(2{\theta }_A\right)\frac{\left|\overrightarrow{k_B}\right|}{\left|\overrightarrow{G_B}\right|}\sin \left(2{\theta }_B\right)\right)}^2}\right|& \left(9\right)\end{array}$

[0039] In the equations above,

[0040] n_pis the refractive index for the pump signal,

[0041] n_Ais the refractive index for the signal at the wavelength λ_A,

[0042] n_Bis the refractive index for the signal at the wavelength λ_B,

[0043] θ_Ais a selected desired half exit angle with respect to a normal to an output face of the lattice body for the output signal at the wavelength λ_A,

[0044] θ_Bis a selected desired half exit angle with respect to a normal to the output face of the lattice body for the output signal at the wavelength λ_B,

[0045] |{right arrow over (k_A+L )}|=2πn_A/λ_A, and |{right arrow over (k_B+L )}|=2πb_B/λ_B

[0046] and |{right arrow over (G_A+L )}| and |{right arrow over (G_B+L )}| are given by the general nonlinear Bragg condition (See Eqn. (6)). The operator (±) is replaced by (+) when EB(ωA) and EA(ωB) exit on opposite sides of the lattice and (−) when they exit on the same side of the lattice.

[0047] As an example of the invention, consider input fields at wavelengths λA=1550 nm and λB=1550.8 nm (100 GHz separation) that are to be simultaneously wavelength interchanged. An optical wavelength interchange device 20 for use with these wavelengths has a lattice body formed as an oblique nonlinear 25 lattice of sensitized regions 37 in dispersive LiNbO3 that has real lattice spacings of 5.7 μm and 11.8 μm——a parallelogram with an interior angle of φ=43.8° ——produces frequency-shifted outputs that exit the lattice on opposite sides at angles of about 4.5° for {right arrow over (k_A¹+L )} and about 10° for {right arrow over (k_B¹+L )}. For multiple input wavelengths, exchange occurs only for those input pairs for which the lattice is designed. In the absence of the pump light, all input waves pass straight through the lattice without any frequency conversion as long as the lattice is designed so that no efficient second harmonic generation occurs for the input waves. For a lattice body of LiNbO₃, the input light signals should be polarized perpendicular to the plane of the lattice body.

[0048] Since the Ewald spheres may intersect more than one reciprocal lattice point, there may be more than one direction in which quasi-phase matching is satisfied for each of the wavelengths. If it is desired to achieve quasi-phase matching in only one direction for each of the newly mapped wavelengths, there are two ways by which this can be achieved. The first method is an iterative one that involves the choice of lattice periodicity. For each pair of values of {right arrow over (G_A+L )} and {right arrow over (G_B+L )}, two families of equations are generated. The intersecting points of these families of equations define all the lattice points in reciprocal space. The requirement now is to find a pair of {right arrow over (G_A+L )} and {right arrow over (G_B+L )} such that there is only one lattice point on each of the Ewald spheres. Once this pair is found, the exit angles can be determined from the general nonlinear Bragg condition (see, Eqn. (6)) and the real space parameters (see, Eqns. (8) and (9)). The alternative method is to manipulate the shape of the unit cell nonlinearity, or equivalently the reciprocal space “structure factor”, χ⁽²⁾_bulk^a_G, upon which the efficiency of the difference-frequency generated light depends. By manipulating the structure factor, it is possible to nullify the output light in undesired directions.

[0049] Depending on the input wavelengths and desired exit angles, the general nonlinear Bragg condition as given in equation (6) and equations (8) and (9) determine the required lattice spacings and pattern. FIG. 4 shows the range of lattice spacings required as a function of half the exit angle θ_Afor the wavelength interchange λ_A=1550 nmλ_B=1550.8 nm, keeping θ_Bfixed at 5°. That is, for this graph, E_A(ω_B) always exits the lattice at 2η_B=10°. It can be seen from the graph that the smallest possible lattice spacing required is a ≈0.37 μm in the limit where θ_A=90°. It is possible to form sensitized regions (microdomains) with diameters smaller than 0.5 μm in LiNbO₃, and domains as small as 5nm have been obtained in related ferroelectrics such as BaTiO₃. Fabrication of periodic microdomains can be performed by standard photolithographic processing. The behavior of r_Bas θ_Aincreases may also be noted: r_Binitially increases until θ_Ais such that {right arrow over (G_A+L )} is tangential to its corresponding Ewald sphere (see FIG. 3) and then monotonically decreases, and then increases until its final value is the same as its initial value when θ_A=0.

[0050] The following is an example of the selection of the spacing and pattern of the sensitized regions in accordance with the invention.

[0051] Assume that it is desired to interchange the wavelengths λA=1.55 μm and λB=1.5508 μm in a nonlinear photonic crystal fabricated in lithium niobate, and that the wavelength-interchanged outputs are to appear at 4.5° (2θA) for kA and 10°(2θB) for kB, where kA,B are the wavevectors of the wavelength-interchanged outputs. In order to determine the lattice parameters, the following calculations are performed.

[0052] The pump wavelength required for such an interchange is given by ${\lambda }_p=\frac{{\lambda }_A{\lambda }_B}{{\lambda }_A+{\lambda }_B}=\frac{1.55\times 1.5508}{1.55+1.5508}=0.775\mu m$

[0053] The refractive indices in lithium niobate for the signal and pump wavelengths are nA=2.16, nB=2.16, and np=2.2. It is then necessary to calculate the corresponding reciprocal lattice vectors GA and GB. The difference-frequency version of Eqn. (6) is used, the general nonlinear Bragg condition that was derived to find these vectors. $\left|G_A\right|=\frac{2\pi n_A}{{\lambda }_A}\sqrt{{\left[1-\left(\frac{n_p{\lambda }_B-n_B{\lambda }_p}{n_A\left({\lambda }_B-{\lambda }_p\right)}\right)\right]}^2+4\left(\frac{n_p{\lambda }_B-n_B{\lambda }_p}{n_A\left({\lambda }_B-{\lambda }_p\right)}\right){\sin }^2\left({\theta }_A\right)}=0.772$ $\left|G_B\right|=\frac{2\pi n_B}{{\lambda }_B}\sqrt{{\left[1-\left(\frac{n_p{\lambda }_A-n_B{\lambda }_p}{n_B\left({\lambda }_A-{\lambda }_p\right)}\right)\right]}^2+4\left(\frac{n_p{\lambda }_A-n_B{\lambda }_p}{n_B\left({\lambda }_A-{\lambda }_p\right)}\right){\sin }^2\left({\theta }_B\right)}=1.587$

[0054] Eqn. (9) is now used to determine the value of sin(φ). The value of kA=2πnA/λA=8.756 and kB=2πnB/μB=8.751. This yields: $\sin \left(\phi \right)=\left|\sqrt{{\left(\frac{\left|\overrightarrow{k_A}\right|}{\left|\overrightarrow{G_A}\right|}\sin \left(2{\theta }_A\right)\right)}^2-{\left(\frac{\left|\overrightarrow{k_A}\right|}{\left|\overrightarrow{G_A}\right|}\sin \left(2{\theta }_A\right)\frac{\left|\overrightarrow{k_B}\right|}{\left|\overrightarrow{G_B}\right|}\sin \left(2{\theta }_B\right)\right)}^2}+\sqrt{{\left(\frac{\left|\overrightarrow{k_B}\right|}{\left|\overrightarrow{G_B}\right|}\sin \left(2{\theta }_B\right)\right)}^2-{\left(\frac{\left|\overrightarrow{k_A}\right|}{\left|\overrightarrow{G_A}\right|}\sin \left(2{\theta }_A\right)\frac{\left|\overrightarrow{k_B}\right|}{\left|\overrightarrow{G_B}\right|}\sin \left(2{\theta }_B\right)\right)}^2}\right|=0.69$

[0055] The real space parameters are now given by derived Eqn. (8): $r_A=\frac{2\pi }{\left|\overrightarrow{G_A}\right|\sin \left(\phi \right)}=11.761\mathrm{\mu m}$ $r_B=\frac{2\pi }{\left|\overrightarrow{G_B}\right|\sin \left(\phi \right)}=5.718\mathrm{\mu m}$

[0056] The interior angle subtended by rA and rB is:

[0057] φ=sin⁻¹(0.692)=43.82°

[0058] FIG. 5 illustrates an exemplary structure for providing lateral confinement of the wavelength interchanged spatially separated light. The solid state body 35 in which the lattice of sensitized regions 37 is formed is bordered laterally by waveguides 50 and 51. If desired, the lattice body may also be formed between cladding layers (not shown) to provide transverse confinement of light to the lattice body. The waveguides 50 and 51 are preferably formed to have the same propagation constant as the z-component of the wave vector in the bulk crystal of the solid state body 35. When the three collinear input signals (at frequencies (ω₁, ω₂, and ω_p) are passed into the body 35, the wavelength interchanged signal at the frequency ω₁is deflected into and confined within the waveguide body 50, while the wavelength interchanged signal at the frequency ω₂is deflected into and confined to the waveguide 51. The signals in the waveguides 50 and 51 then can exit the device at positions spatially separated from but propagating in the same direction as the original input signals at the wavelengths ω₁, ω₂, and ω_p. This process is analogous to prism coupling. See, e.g., Topics in Applied Physics, Vol. 7, Integrated Optics, Springer Verlag, Berlin, 1979, pp. 83-134.

[0059] As noted above, in the 1.0 to 1.6 μm wavelength region, the required periodicities of the sensitized regions in the lattice are about 5 to 10 μm, which can be readily produced utilizing standard photolithographic processes. The pattern of sensitized regions 37 can be readily produced in ferroelectrics, such as LiNbO₃, utilizing photolithographically-defined E-field poling processes, as illustrated in FIG. 6. The solid state body 35 (e.g., bulk LiNbO₃) is mounted in a reaction chamber 55 which holds a suitable reactant (e.g., LiCl for patterning a LiNbO₃wafer). A power source 56 applies a high voltage to the reaction chamber 55 that is applied across the solid state body 35 to produce the sensitized regions utilizing conventional processing techniques. See, e.g., M. L. Bortz, et al., supra. The solid state body 35 may also be formed in ferroelectric thin films, which have a number of advantages over the bulk material. The inherently strong optical field confinement provided by thin films can result in much more efficient nonlinear interactions, and poling voltages are much lower, potentially permitting electrically programmable crystal structures. 2-D nonlinear photonic crystals may, for example, be formed in films made by wafer bonding and thinning LiNbO₃. An example of this structure is an ˜50 μm thick LiNbO₃film formed on a silicon substrate. Further, it is noted that it is possible to grow films of LiNbO₃with excellent crystallinity using a modulated chemical beam epitaxy method. Such thin film epitaxy may be utilized to extend the nonlinear photonic crystals into three dimensions.

[0060] The wavelength interchange device of the present invention may be utilized for various applications and the invention may be embodied in related devices without departing from the spirit and scope of the present invention. One example is the use of the present invention to facilitate wavelength hopping for data encryption. The same process described above for wavelength routing can be used to shift portions of the data stream to new wavelengths. In this case, a data stream (analogous to a train) is composed of a set of packets (analogous to rail cars) of different wavelengths. This has the effect of intermixing or scrambling the data streams. The signals may then be decoded by reversing the process at the receiving end by utilizing a set of pumps that are activated in the reverse order.

[0061] The nonlinear photonic crystal can be extended to three dimensions to permit logic, arithmetic, and wavelength-shifting operations to be carried out on two-dimensional optical arrays for image analysis or to implement free-space interconnects. It is noted that the nonlinear domains in many ferroelectric films are in the nanoscale range (that is, in the range of about 10 nm), thus making it possible to provide sufficiently large phase matching to directly convert mid-infrared light to the visible for applications in imaging spectroscopy or to long wavelengths (>2 μm) for applications such as environmental sensing. It is also possible to use these optically nonlinear nanocrystals as digital 2-D holographic patterns, subsequent illumination of which by, e.g., an infrared beam, would generate a holographic image in the visible range by way of second harmonic generation.

[0062] The invention may be utilized to provide inter-conversions between wavelength division and time division. By properly choosing the pump wavelength and the direction of propagation of the pump, input and output optical data streams, two or more data streams, each at its own wavelength, may be converted to a common wavelength which exits a common port. This process amounts to conversion of WDM data to time division multiplexed (TDM) data. The reverse can also be achieved by reversing the directions of output and input data streams.

[0063] The same periodically oriented domains in a ferroelectric such as LiNbO₃which give rise to wavelength interchange also will produce a periodic array of refractive index differences with application of a uniform electric field across the substrate. The result is a simultaneous 2-D (or 3-D) linear-plus-nonlinear photonic crystal. Such crystals may be utilized for their ability to steer the direction of the exiting light.

[0064] The nonlinear photonic crystals can also be used to perform χ⁽²⁾:χ⁽²⁾cascading, thus mimicking a third order process. For three input signals at frequencies ω₁, ω₂, and ω₃, one of the 2-D reciprocal lattice vectors can be designed to match ω₁+ω₂while another reciprocal lattice vector can be designed to match ω₁+ω₂−ω₃. The advantage of the 2-D nonlinear photonic crystal in comparison with its 1-D counterpart is that it has the ability to obtain pure optical products that are spatially segregated and free of other near-degenerate products. The interchange efficiency will depend upon the polarization of the input and pump beams (by way of the nonlinear susceptibility tensor of the lattice material). The present invention also can be extended to three dimensional lattices having multiple layers of sensitized regions.

[0065] It should be understood that it is not essential in accordance with the invention for the input signals and the pump signal to be coupled into the lattice body collinear with one another as was illustrated in FIG. 1. As will be apparent from examination of the design equations given above, the lattice body can be selected to provide an appropriate angle of coupling of the input signals into the lattice body to provide wavelength interchanged output signals which exit either collinear with one another or exit at desired angles with respect to each other.

[0066] To illustrate these principles, FIG. 7 shows the apparatus of the invention in which the input signals and the pump signal are not collinear with one another, but rather are coupled to the lattice body 35 at angles to each other. The optical waveguide 60 carrying the pump signal at the frequency ω_pis coupled to the input face 61 of the lattice body 35 at an angle with respect to a normal to the face 61 of the lattice body (illustrated by the dashed line labeled 62 in FIG. 7). An input waveguide 64 carrying one of the input signals at a frequency ω₁and a second input waveguide 65 carrying an input signal at a frequency ω₁are coupled to the input face 61 of the lattice body 35 at further angles off the normal 62 to the face of the lattice body. Utilizing a design criteria similar to that described above, the spacing and pattern of the sensitized regions in the lattice body 35 can be selected such that the signals carried by the waveguide 60, 64 and 65 will interact with the lattice where these signals intersect within the lattice body 35 to provide collinear output signals which are coupled to an output waveguide 68 at an angle which may be normal to the output face 70 of the lattice body 35 as shown, or at an angle, with the waveguide 68 carrying output signals at wavelengths ω₁and ω₂that are collinear with one another. This arrangement effectively both interchanges the information carried by the input signals at the frequencies ω₁and ω₂and also multiplexes the signals so that the two separate signals are now carried on a single output waveguide 68. As indicated by the nomenclature in FIG. 7, the information “A” carried by the input signal at the frequency ω₁is now carried by the signal at the frequency ω₂in the output, while the information “B” in the input signal at the frequency ω₂is carried in the output by the signal at the frequency ω₁. The spacing and pattern of sensitized regions in the lattice body can also be selected to provide separated output signals, as illustrated in FIG. 8, wherein the output signal at the frequency ω₁, exits from the lattice body 35 at an angle to the output face 70 of the lattice body and is received by a waveguide 72, while the output signal at the frequency ω₂exits at a different angle to the output face 70 and is received by a second output waveguide 73. In both cases, the output signals at the frequencies ω₁and ω₂are angularly separated from the pump signal provided on the input waveguide 60, which passes without deflection through the lattice body and exits at an angle different from the angles to the output face 70 of the lattice body at which the output signals at the frequencies ω₁and ω₂exit.

[0067] In all of the foregoing embodiments of the invention, it is understood that the apparatus in accordance with the invention may provide not only wavelength interchange of the input signals at the frequencies ω₁and ω₂, but also parametric amplification of such signal where the pump signal at the frequency cop has greater intensity than the input signals. In fact, it is apparent that the apparatus of the invention may be utilized to carry out amplification of two input signals with or without wavelength interchange. For example, utilizing the apparatus of FIG. 1, the input signals can be separated into two angularly diverging output signals at frequencies ω₁and ω₂, which are wavelength interchanged, which may then be supplied to the input waveguides 64 and 65 of the apparatus of FIG. 7 or FIG. 8 to provide amplified output signals in which the original carrier frequencies each carry their original information either collinear with each other, as for the apparatus of FIG. 7, or separated from one another, as for the apparatus of FIG. 8.

[0068] It is understood that the invention is not confined to the particular embodiments set forth herein as illustrative, but embraces all such modified forms thereof as come within the scope of the following claims.