CN101063731A - Distributed feedback inhibition semiconductor optical amplifier - Google Patents

Abstract

This invention relates to one non-symmetric curve source wave guide chip structure semi-conductor amplifier for 1310 nm and 1550 nm wave sections in light communication technique field, which is characterized by the following: leading non-symmetric curve consumption to reduce feedback in chip inner distribution; through leading wave guide side Goos-Hanchen displacement to form reflection displacement to interfere with reflection field phase to make light field bias from oscillation condition to realize semi-conductor amplifier for high injection down wave work to achieve high gains and low noise aim.

Description

Distributed feedback inhibition semiconductor optical amplifier

(1) technical field

The present invention relates to the semiconductor optical amplifier of the employing distributed Feedback inhibition technology of 1310nm wave band or two optical fiber communication windows of 1550nm wave band, belong to the optical communication technique field.

(2) technical background

(1) Optical Amplification Technology is an important component part of optical communication technique.Semiconductor optical amplifier has great importance to the transmission that solves the 1310nm window, wavelength Conversion and active photoswitch, light exchange, photon integrated (PIC) and the integrated technical matterss such as (OEIC) of photoelectricity in the following all-optical network, be optical communication that study with key areas product development ^{[1], [2]}

Middle nineteen nineties, the research level of semiconductor optical amplifier (SOA) has reached the application technology level of Erbium-Doped Fiber Amplifier (EDFA) ^{[3], [4]}But, very complicated and difficult by the exocoel feedback inhibition system coupling that optical elements such as non-ball lenticule and isolator constitute with packaging technology, can not satisfy the requirement that industrialization is made, high-performance semiconductor image intensifer valuable product can not satisfy the requirement in market.Hindered the optical communication technique development.

(2) the exocoel feedback is that SOA realizes high-gain, low noise major obstacle.

The feedback of SOA by chip end face and air dielectric and coupled fiber and air dielectric the reflection of totally four optical interfaces cause (Fig. 1).

The gain G of SOA satisfies

$G=\frac{(1-R_1)(1-R_2)G_s}{{(1-\sqrt{R_1{R}_2}{G}_s)}^2+4\sqrt{R_1{R}_2}{G}_s{\sin }^2[2\mathrm{\π}(v-v_0)L/c]}$

R ₁And R ₂Be respectively the residual reflectance at amplifier input and output two ends, G _sBe single-pass gain, v is a frequency, v ₀Be the centre frequency of gain spectral, L and c are respectively the long and lighies velocity in chamber.

R ₁And R ₂Feedback exists, and constitutes resonator cavity.Its gain for threshold value g _ThBe cavity loss α _InWith end face loss sum

$g_{\mathrm{th}}={\mathrm{\α}}_{\mathrm{in}}+\frac{1}{2L}\ln \left(\frac{1}{R_1{R}_2}\right)---\left(1\right)$

Because SOA is a row ripple device work, and sufficiently high gain for threshold value must be arranged.Cavity loss is determined that by material chamber length can not be too little, because the gain of SOA is a single-pass gain.Thereby enough low feedback is to guarantee that device has sufficiently high biasing, guarantees that again gain is lower than the necessary condition of threshold value.High-performance SOA requires residual reflection to reach 10 ^-6The order of magnitude ^{[3], [4]}According to existing technical conditions, adopt anti-reflection film the end face reflection of chip can be suppressed to 10 ^-4Magnitude, further inhibition will be adopted the angled end-face structure ^{[5], [6]}Though the resultant effect of the two can reach 10 ^-6The magnitude residual reflection, still, the angled end-face structure is brought the mould field mismatch of chip end face and coupled fiber, causes coupling loss.Reflect lowly more, require the oblique angle big more, mismatch is serious more, and loss is big more.The coupling loss at two ends descends except causing signal power, thereby outside the gain decline, the loss of input end is the noisiness of deterioration of device especially.In addition, the natural reflectivity of fiber end face is 10 ^-2The order of magnitude, because technologic reason, the plated film difficulty, reflection is difficult to suppress.Theoretical analysis and experimental result show, if the exocoel that the coupled fiber end face is formed does not add inhibition, device gain is difficult to surpass 14dB.The product of having released in the market all can only be operated in the low gain state.

The noise mechanism of SOA is that noise and signal and spontaneous emission bat noise are clapped in spontaneous emission and spontaneous emission.The former can pass through the narrow band filter elimination, thereby noise figure F is

$F=2\mathrm{\χ}{n}_{\mathrm{sp}}+\frac{{\mathrm{\χn}}_{\mathrm{sp}}^2{m}_t\mathrm{\Δf}}{\⟨n_{\mathrm{in}}\⟩}---\left(2\right)$

χ is the constant relevant with reflection with single-pass gain, m _tBe 1 or 2, decide that Δ f is the bandwidth of wave filter,＜n on polarization _InBe the photon density of input end, n _SpIt is the population inversion parameter

$n_{\mathrm{sp}}=\frac{\mathrm{\Γ}{T}_{\mathrm{spon}}}{\mathrm{\Γ}{T}_{\mathrm{stim}}-{\mathrm{\αv}}_g}---\left(3\right)$

Γ is the light field restriction factor, and α is a loss factor, v _gBe group velocity, T _Spon, T _StimBe respectively spontaneous emission rate and stimulated emission speed.Because exocoel feedback, device are operated in low biasing down, it is low to inject carrier concentration, and a little less than the stimulated emission, the noise power of spontaneous emission is strong relatively, the noise figure height.According to the work before the applicant, under the biasing 90mA, the measured value of noise figure is generally more than 10dB.Relevant report is less than the SOA of 7dB noise figure, corresponding 400mA offset operation state.

Another spinoff that feedback is brought is to cause gain fluctuation, is caused by the modulation of feedback to gain:

$m={\left(\frac{1+\sqrt{R_1{R}_2}{G}_s}{1-\sqrt{R_1{R}_2}{G}_s}\right)}^2---\left(4\right)$

In dense wave division multiplexing transmission system, require gain fluctuation less than 0.5dB, reality is lower.If require less than 0.2dB, the single-pass gain of supposing chip is 30dB, and then residual reflection should reach 10 ^-5

(3) up to the present, the technology that suppresses exocoel feedback in the world is to adopt the complex devices structure ^{[3], [4]}, for example, adopt non-globe lens, optoisolator and angled end-face optical fiber (Fig. 2).

Non-globe lens is realized the mould field coupling between chip and optoisolator, optoisolator and the optical fiber, and isolator suppresses feedback.The coupling encapsulation of such complication system is difficulty very, can manufacture hardly.

(4) at above problem, the present patent application people proposes a kind of new technological thought, adopts asymmetrical active waveguide structure, realizes feedback inhibition voluntarily in the inside of chip active area.The theoretical foundation of present technique is: in the process that light field is propagated along curved waveguide gain, produce bending loss, the tangential angle of the size of local loss and the direction of propagation and waveguide is relevant.In asymmetrical curved waveguide structure, the forward-wave of arbitrary position, waveguide side tangentially has different angle because of the wave vector direction with waveguide with reflection wave, thereby can be under the little condition of forward-wave loss, increase the loss of reflection wave as far as possible, the energy of mirror field is discharged along waveguide side direction covering.In addition, same because forward-wave because different wave vector directions cause different phase shifts, forms phase interference with reflection wave.More than two reasons, make required amplitude and the phase-matching condition of vibration be difficult to satisfy, realize the capable ripple work under high the injection.The difference of the present invention and existing semiconductor optical amplifier feedback inhibition technology is: the latter is the inhibition at exocoel, and the present invention then is the inhibition at feedback light field amplitude and phase place.

(5) under the support of state natural sciences fund, the present patent application people has carried out at this Study on Technology work and has concentrated on the optimal design of active curved waveguide.Research is adopted Finite-Difference Time-Domain Method (FDTD) design and emulation at SOA physical model shown in Figure 3.Based on the Maxwell equation, under the condition of given active waveguide gain coefficient, import a Gauss light pulse, the SOA model of differently curved active waveguide structure is carried out the design and simulation of FDTD method.Optimize waveguiding structure by the ratio of optimizing output and feedback.Obtain desirable waveguiding structure thus.

Theory and method are as follows:

Consider to inject influence and the effect of dispersion of charge carrier to the material refractive index,

Being changed to of active medium refraction index

$\mathrm{\δn}(N,\mathrm{\ω})=\frac{c}{2\mathrm{\π\ω}}{\∫}_{-\∞}^{\∞}\frac{g(N,\mathrm{\ω})}{\mathrm{\ω}-{\mathrm{\ω}}^{\′}}d{\mathrm{\ω}}^{\′}---\left(5\right)$

Wherein, N and ω inject carrier concentration and light field angular frequency, g (N, ω) gain coefficient.

The polarizability of active medium is

$\mathrm{\χ}(N,\mathrm{\ω})=\frac{2}{\sqrt{{\mathrm{\ϵ}}_r}}\mathrm{\δn}(N,\mathrm{\ω})-j\frac{g(N,\mathrm{\ω})}{k}---\left(6\right)$

Polarizability is expressed as following Lorentz lorentz's form

$\left\{\begin{array}{c}\widehat{\mathrm{\χ}}(N,\mathrm{\ω})={\mathrm{\χ}}_0\left(N\right)+\underset{l=1}{\overset{N}{\mathrm{\Σ}}}\frac{A_l\left(N\right)}{j{\mathrm{\Γ}}_l\left(N\right)+({\mathrm{\δ}}_0+\mathrm{\ω}-{\mathrm{\δ}}_l\left(N\right))}\\ {\mathrm{\χ}}_0\left(N\right)=\frac{2}{\sqrt{{\mathrm{\ϵ}}_r}}\mathrm{\δ}{n}_0\left(N\right)-j\frac{g_0\left(N\right)}{k}\end{array}\right.---\left(7\right)$

Wherein, χ ₀(N), A ₁(N), Γ ₁(N), δ ₁(N) be undetermined parameter, they are the function of carrier concentration N, and

ω _cBe central angle frequency, E _gBe energy gap.

Utilize interpolation method can provide χ ₀(N), A ₁(N), Γ ₁(N), δ ₁(N) analytic relationship.

For non-magnetic media, two vorticity equations of Maxwell equation are

$\left\{\begin{array}{c}\▿\×\stackrel{\‾}{E}=-{\mathrm{\μ}}_0\frac{\∂\stackrel{\→}{H}}{\∂t}\\ \▿\×\stackrel{\→}{H}=\frac{\∂\stackrel{\→}{D}}{\∂t}\end{array}\right.---\left(8\right)$

Wherein $\stackrel{\→}{D}=\mathrm{\ϵ}\stackrel{\→}{E}+\stackrel{\→}{P}={\mathrm{\ϵ}}_0{\mathrm{\ϵ}}_r\stackrel{\→}{E}+\stackrel{\→}{P}.$

Polarization intensity for medium is provided by following formula. $\stackrel{\&RightArrow;}{P}=\stackrel{\&RightArrow;}{{\mathrm{<figure-callout\; id="0"\; label="P"\; filenames="A20061008006500092.png"\; state="\{\{state\}\}">P</figure-callout>}}_0}+\stackrel{\&RightArrow;}{P_1}+\stackrel{\&RightArrow;}{P_2}+\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;$

(9)

Wherein

$\left\{\begin{array}{c}{\stackrel{\&RightArrow;}{P}}_0\left(\mathrm{\&omega;}\right)={\mathrm{\&epsiv;}}_0{\mathrm{\&epsiv;}}_r{\mathrm{\&chi;}}_0\left(N\right)\stackrel{\&RightArrow;}{E}\left(\mathrm{\&omega;}\right)\\ {\stackrel{\&RightArrow;}{P}}_l\left(\mathrm{\&omega;}\right)={\mathrm{\&epsiv;}}_0{\mathrm{\&epsiv;}}_r{\widehat{\mathrm{\&chi;}}}_l(N,\mathrm{\&omega;})\stackrel{\&RightArrow;}{E}\left(\mathrm{\&omega;}\right),l=\mathrm{1,2}\&CenterDot;\&CenterDot;\&CenterDot;\end{array}\right.---\left(10\right)$

${\widehat{\mathrm{\&chi;}}}_l(N,\mathrm{\&omega;})=\frac{A_l\left(N\right)}{j{\mathrm{\&Gamma;}}_l\left(N\right)+({\mathrm{\&delta;}}_0+\mathrm{\&omega;}-{\mathrm{\&delta;}}_l\left(N\right))}$

Following formula as Fourier transform be

$\left\{\begin{array}{c}{\stackrel{\&RightArrow;}{P}}_0={\mathrm{\&epsiv;}}_0{\mathrm{\&epsiv;}}_r{\mathrm{\&chi;}}_0\left(N\right)\stackrel{\&RightArrow;}{E}\left(t\right)\\ \frac{d{\stackrel{\&RightArrow;}{P}}_l\left(t\right)}{\mathrm{dt}}=j{\mathrm{\&epsiv;}}_0{\mathrm{\&epsiv;}}_r{A}_l\left(N\right)[{\mathrm{\&Gamma;}}_l\left(N\right)-j({\mathrm{\&delta;}}_0-{\mathrm{\&delta;}}_l\left(N\right))]\exp \left\{\right[{\mathrm{\&Gamma;}}_l\left(N\right)-j({\mathrm{\&delta;}}_0-{\mathrm{\&delta;}}_l\left(N\right))\left]t\right\}\\ \&times;\left[\underset{0}{\overset{t}{\&Integral;}}\exp \right[-[{\mathrm{\&Gamma;}}_l\left(N\right)-j({\mathrm{\&delta;}}_0-{\mathrm{\&delta;}}_l\left(N\right))]s]\&CenterDot;\stackrel{\&RightArrow;}{E}\left(s\right)\mathrm{ds}]+j{\mathrm{\&epsiv;}}_0{\mathrm{\&epsiv;}}_r{A}_l\left(N\right)\stackrel{\&RightArrow;}{E}\left(t\right)\end{array}\right.---\left(11\right)$

Adopt two energy level gain media systems, its gain coefficient can be written as ^[6]

$g\left(\mathrm{\&omega;}\right)=\frac{{\mathrm{<figure-callout\; id="0"\; label="g"\; filenames="A20061008006500092.png"\; state="\{\{state\}\}">g</figure-callout>}}_0}{1+{(\mathrm{\&omega;}-{\mathrm{\&omega;}}_0)}^2{T}_2^2}---\left(12\right)$

In the formula, g ₀Be the gain peak that depends on the amplifier pumping level, ω is the incident light frequency, ω ₀Be the atomic transition frequency.T ₂Be dipole relaxation time (T ₂≈ 0.1ps).

Thereby,

$\mathrm{\&delta;n}\left(\mathrm{\&omega;}\right)=\frac{c}{2\mathrm{\&pi;\&omega;}}\underset{-\&infin;}{\overset{\&infin;}{\&Integral;}}\frac{{\mathrm{<figure-callout\; id="0"\; label="g"\; filenames="A20061008006500092.png"\; state="\{\{state\}\}">g</figure-callout>}}_0}{1+{({\mathrm{\&omega;}}^{\&prime;}-{\mathrm{\&omega;}}_0)}^2{T}_2^2}\frac{1}{\mathrm{\&omega;}-{\mathrm{\&omega;}}^{\&prime;}}d{\mathrm{\&omega;}}^{\&prime;}=\frac{{\mathrm{<figure-callout\; id="0"\; label="cg"\; filenames="A20061008006500092.png"\; state="\{\{state\}\}">cg</figure-callout>}}_0}{2\mathrm{\&omega;}}\frac{T_2(\mathrm{\&omega;}-{\mathrm{\&omega;}}_0)}{1+{(\mathrm{\&omega;}-{\mathrm{\&omega;}}_0)}^2{T}_2^2}---\left(13\right)$

$\mathrm{\&chi;}\left(\mathrm{\&omega;}\right)=\frac{{\mathrm{cg}}_0}{T_2\sqrt{{\mathrm{\&epsiv;}}_r}[-{\mathrm{\&omega;}}_0+\mathrm{j\&alpha;}]}[\frac{1}{\mathrm{\&omega;}}-\frac{\mathrm{\&omega;}-{\mathrm{\&omega;}}_0}{{(\mathrm{\&omega;}-{\mathrm{\&omega;}}_0)}^2+{\mathrm{\&alpha;}}^2}+j\frac{\mathrm{\&alpha;}}{{(\mathrm{\&omega;}-{\mathrm{\&omega;}}_0)}^2+{\mathrm{\&alpha;}}^2}]---\left(14\right)$

Wherein, α=1/T ₂

(2-6) formula is made Fourier transform can be got

$\mathrm{\&chi;}\left(t\right)=\left\{\begin{array}{c}\frac{{\mathrm{cg}}_0}{T_2\sqrt{{\mathrm{\&epsiv;}}_r}[-{\mathrm{\&omega;}}_0+\mathrm{j\&alpha;}]}[1-e^{({\mathrm{j\&omega;}}_0-\mathrm{\&alpha;})t}+j{e}^{({\mathrm{j\&omega;}}_0-\mathrm{\&alpha;})t}],t>0\\ \frac{{\mathrm{cg}}_{0}j}{T_2\sqrt{{\mathrm{\&epsiv;}}_r}[-{\mathrm{\&omega;}}_0+\mathrm{j\&alpha;}]},t=0\\ \frac{{\mathrm{cg}}_0}{T_2\sqrt{{\mathrm{\&epsiv;}}_r}[-{\mathrm{\&omega;}}_0+\mathrm{j\&alpha;}]}[-1+e^{({\mathrm{j\&omega;}}_0+\mathrm{\&alpha;})t}+j{e}^{({\mathrm{j\&omega;}}_0+\mathrm{\&alpha;})t}],t<0\end{array}\right.---\left(15\right)$

Equally, by (2-6) formula and (2-2b) formula can obtain being expressed as of polarization intensity

P _r(ω)＝ε ₀χ(ω)E(ω)＝ε ₀βE(ω)/{ω[ω-ω ₀+jα]} (16)

Wherein, $\mathrm{\&beta;}={\mathrm{<figure-callout\; id="0"\; label="cg"\; filenames="A20061008006500092.png"\; state="\{\{state\}\}">cg</figure-callout>}}_0/\left(T_2\sqrt{{\mathrm{\&epsiv;}}_r}\right).$ The Fourier transform of following formula is

$\frac{\&PartialD;{\stackrel{\&RightArrow;}{P}}_r\left(t\right)}{\&PartialD;}=e^{(j{\mathrm{\&omega;}}_0+\mathrm{\&alpha;})t}[c_0-\underset{0}{\overset{t}{\&Integral;}}{\mathrm{\&epsiv;}}_0\mathrm{\&beta;}{e}^{-(j{\mathrm{\&omega;}}_0+\mathrm{\&alpha;})s}\mathrm{ds}]---\left(17\right)$

Work as P _rWhen (t) being a tempolabile function, be carved with in the time of can thinking t=0 at the starting point place $\&PartialD;{\stackrel{\&RightArrow;}{P}}_r\left(t\right)/\&PartialD;t{|}_{t=0}=0,$ C is arranged thus ₀=0 (18)

With two vorticity equations of above-mentioned polarization vector substitution Maxwell equation be

Obtain:

$\begin{array}{cc}\left\{\begin{array}{c}\frac{\&PartialD;E_z}{\&PartialD;y}-\frac{\&PartialD;E_y}{\&PartialD;z}=-{\mathrm{\&mu;}}_0\frac{\&PartialD;H_x}{\&PartialD;t}\\ \frac{{\&PartialD;E}_x}{\&PartialD;z}-\frac{\&PartialD;E_z}{\&PartialD;x}=-{\mathrm{\&mu;}}_0\frac{\&PartialD;H_y}{\&PartialD;t}\\ \frac{\&PartialD;E_y}{\&PartialD;x}-\frac{\&PartialD;E_x}{\&PartialD;y}=-{\mathrm{\&mu;}}_0\frac{\&PartialD;H_z}{\&PartialD;t}\end{array}\right.& \left\{\begin{array}{c}\frac{\&PartialD;H_z}{\&PartialD;y}-\frac{\&PartialD;H_y}{\&PartialD;z}=\mathrm{\&epsiv;}\frac{\&PartialD;E_x}{\&PartialD;t}+\frac{\&PartialD;P_{\mathrm{rx}}}{\&PartialD;t}\\ \frac{\&PartialD;H_x}{\&PartialD;z}-\frac{\&PartialD;H_z}{\&PartialD;x}=\mathrm{\&epsiv;}\frac{\&PartialD;E_y}{\&PartialD;t}+\frac{\&PartialD;P_{\mathrm{ry}}}{\&PartialD;t}\\ \frac{\&PartialD;H_y}{\&PartialD;x}-\frac{\&PartialD;H_x}{\&PartialD;y}=\mathrm{\&epsiv;}\frac{\&PartialD;E_z}{\&PartialD;t}+\frac{\&PartialD;P_{\mathrm{rz}}}{\&PartialD;t}\end{array}---\left(19\right)\right.\end{array}$

Utilize the Yee trellis algorithm under the rectangular coordinate system, obtain its discrete form.

$\left\{\begin{array}{c}{E}_x(\stackrel{\&RightArrow;}{r},n+1)=E_x(\stackrel{\&RightArrow;}{r},n)+\frac{\mathrm{\&Delta;t}}{\mathrm{\&epsiv;}}[\frac{{\&PartialD;H}_z}{\&PartialD;y}-\frac{{\&PartialD;H}_y}{\&PartialD;z}]-\frac{\mathrm{\&Delta;t}}{\mathrm{\&epsiv;}}\frac{{\&PartialD;P}_{\mathrm{rx}}(\stackrel{\&RightArrow;}{r},t)}{\&PartialD;t}{|}_{t=n+\frac{1}{2}}\\ E_y(\stackrel{\&RightArrow;}{r},n+1)=E_y(\stackrel{\&RightArrow;}{r},n)+\frac{\mathrm{\&Delta;t}}{\mathrm{\&epsiv;}}[\frac{{\&PartialD;H}_x}{\&PartialD;z}-\frac{{\&PartialD;H}_z}{\&PartialD;x}]-\frac{\mathrm{\&Delta;t}}{\mathrm{\&epsiv;}}\frac{{\&PartialD;P}_{\mathrm{ry}}(\stackrel{\&RightArrow;}{r},t)}{\&PartialD;t}{|}_{t=n+\frac{1}{2}}\\ E_z(\stackrel{\&RightArrow;}{r},n+1)=E_z(\stackrel{\&RightArrow;}{r},n)+\frac{\mathrm{\&Delta;t}}{\mathrm{\&epsiv;}}[\frac{{\&PartialD;H}_y}{\&PartialD;x}-\frac{{\&PartialD;H}_x}{\&PartialD;y}]-\frac{\mathrm{\&Delta;t}}{\mathrm{\&epsiv;}}\frac{{\&PartialD;P}_{\mathrm{rz}}(\stackrel{\&RightArrow;}{r},t)}{\&PartialD;t}{|}_{t=n+\frac{1}{2}}\end{array}\right.---\left(20\right)$

$\left\{\begin{array}{c}\frac{\&PartialD;P_{\mathrm{rx}}(\stackrel{\&RightArrow;}{r},t)}{\&PartialD;t}{|}_{t=n+\frac{1}{2}}=-{\mathrm{\&epsiv;}}_0{e}^{(\mathrm{\&alpha;}+j{\mathrm{\&omega;}}_0)(n+\frac{1}{2})\mathrm{\&Delta;t}}\{\underset{l=1}{\overset{n}{\mathrm{\&Sigma;}}}\mathrm{\&beta;}{E}_x(\stackrel{\&RightArrow;}{r},l)\mathrm{\&Delta;t}{e}^{-(\mathrm{\&alpha;}+j{\mathrm{\&omega;}}_0)\mathrm{l\&Delta;t}}\\ +\mathrm{\&beta;}\frac{3E_x(\stackrel{\&RightArrow;}{r},n)-E_x(\stackrel{\&RightArrow;}{r},n-1)}{2}\mathrm{\&Delta;t}{e}^{-(\mathrm{\&alpha;}+j{\mathrm{\&omega;}}_0)(N+\frac{1}{2})\mathrm{\&Delta;t}}\}\\ \frac{\&PartialD;P_{\mathrm{ry}}(\stackrel{\&RightArrow;}{r},t)}{\&PartialD;t}{|}_{t=n+\frac{1}{2}}=-{\mathrm{\&epsiv;}}_0{e}^{(\mathrm{\&alpha;}+j{\mathrm{\&omega;}}_0)(n+\frac{1}{2})\mathrm{\&Delta;t}}\{\underset{l=1}{\overset{n}{\mathrm{\&Sigma;}}}\mathrm{\&beta;}{E}_y(\stackrel{\&RightArrow;}{r},l)\mathrm{\&Delta;t}{e}^{-(\mathrm{\&alpha;}+j{\mathrm{\&omega;}}_0)\mathrm{l\&Delta;t}}\\ +\mathrm{\&beta;}\frac{3E_y(\stackrel{\&RightArrow;}{r},n)-E_y(\stackrel{\&RightArrow;}{r},n-1)}{2}\mathrm{\&Delta;t}{e}^{-(\mathrm{\&alpha;}+j{\mathrm{\&omega;}}_0)(n+\frac{1}{2})\mathrm{\&Delta;t}}\}\\ \frac{\&PartialD;P_{\mathrm{rz}}(\stackrel{\&RightArrow;}{r},t)}{\&PartialD;t}{|}_{t=n+\frac{1}{2}}=-{\mathrm{\&epsiv;}}_0{e}^{(\mathrm{\&alpha;}+j{\mathrm{\&omega;}}_0)(n+\frac{1}{2})\mathrm{\&Delta;t}}\{\underset{l=1}{\overset{n}{\mathrm{\&Sigma;}}}\mathrm{\&beta;}{E}_z(\stackrel{\&RightArrow;}{r},l)\mathrm{\&Delta;t}{e}^{-(\mathrm{\&alpha;}+j{\mathrm{\&omega;}}_0)\mathrm{l\&Delta;t}}\\ +\mathrm{\&beta;}\frac{3E_z(\stackrel{\&RightArrow;}{r},n)-E_z(\stackrel{\&RightArrow;}{r},n-1)}{2}\mathrm{\&Delta;t}{e}^{-(\mathrm{\&alpha;}+j{\mathrm{\&omega;}}_0)(n+\frac{1}{2})\mathrm{\&Delta;t}}\}\end{array}\right.---\left(21\right)$

Select the modulation Gauss pulse as importing light:

$E\left(t\right)=\cos \left({\mathrm{\&omega;}}_{0}t\right)e^{-{\left(\frac{t-\mathrm{nT}}{\mathrm{mT}}\right)}^2}---\left(22\right)$

Its frequency domain form is

$E\left(\mathrm{\&omega;}\right)=\frac{\sqrt{\mathrm{\&pi;}}\mathrm{mT}}{2}{e}^{-\frac{m^2{T}^2{(\mathrm{\&omega;}+{\mathrm{\&omega;}}_0)}^2}{4}}{e}^{-\mathrm{jnT}(\mathrm{\&omega;}+{\mathrm{\&omega;}}_0)}[1+e^{j2\mathrm{nT}{\mathrm{\&omega;}}_0}{e}^{m^2{T}^2\mathrm{\&omega;}{\mathrm{\&omega;}}_0}]---\left(23\right)$

Through reasonably analyzing, get Gauss pulse parameter m=19, n=1, T=0.7 * 10 ^-15S.

The physical model that is used for three-dimensional artificial is shown in (Fig. 3).Dash area is an absorption layer, and absorption layer inside is the zoning.In the zoning, be the incident light district in the axial left bottom of z; The right side is the fiber section of output, and wherein, the centre is a fiber cores, and both sides are fibre cladding.The centre of zoning is the semiconductor optical amplifier chip.For convenience of calculation, we are vacuum with zone between optical fiber and the anti-reflection film and incident light area definition.

Utilize the Finite Difference-Time Domain separating method to find the solution the Maxswell equation, provide electromagnetic field distribution in the zoning.According to electromagnetic field distribution, can obtain:

1) electromagnetic energy of output optical fibre place transmission

2) reflected energy that receives of incident light district

Utilize the reflected energy in incident light district and the energy of output optical fibre place transmission recently curved waveguide to be optimized design

(3) description of drawings

Fig. 1 is that optical fiber directly is coupled and forms the synoptic diagram of optical interface;

Fig. 2 is the SOA structural representation that adopts optoisolator;

Fig. 3 is the SOA chip structure physical model that is used for FDTD emulation;

Fig. 4 is active curved waveguide structural representation.

(4) concrete enforcement

The example of a design is seen (Fig. 4).

Material and structural parameters are as follows:

Active layer thickness d=1.0 μ m

Fiber core diameter D=10.0 μ m

Active waveguide refractive index (InGaAsP): 3.5

Cladding index (InP): 3.2

Fiber cores refractive index (SiO2): 1.46

Fibre cladding refractive index (SiO2): 1.45

Wavelength: 1.55 μ m

R/R ₁/R ₂＝1/1.58/0.42

Simulation result:

Output optical power 1.0mW, the ratio of the curved waveguide of corresponding equal length (straight length) and the feedback of straight wave guide is about 5.3%

(5) main reference document

[1]Stephen?Hardy，SONATA?project?to?play?all-optical?tune，Lightwave，1998，September，23-24.

[2]Stephen?Hardy，Amplifier?research?seeks?to?open‘seeks?window’，Lightwave，1998，November，36-38.

[3]Tiemeijer?L.F.，High?performance?MQW?laser?amplifiers?for?transmission?systemsoperating?in?the?1310nm?window?at?10Gbit/s?and?beyond，Proceedings?of?21th?Europeconference?on?optical?communication，ECOC’95-Brussel，259-266.

[4]L.F.Tiemeijer?et?al.，Packaged?high?gain?unidirectional?1300nm?MQW?amplifiers，Proc.45 ^th?Elec.Comp.&?Tech.Conf.，1995，751-758

[4] cross-talk is firm etc., a kind ofly is suitable for a large amount of preparation row ripple SOA anti-reflection film new technologies, laser technology, 2004 the 5th phases.

[6]Duan?Zigang，Huang?Dexiu，Experimental?study?on?semiconductor?optical?amplifierswith?angled?facets，Chinese?Journal?of?lasers?B，Vol.8，N1，Feb.20，1999.

Claims (4)

1 distributed feedback inhibition semiconductor optical amplifier, its chip structure is: because the asymmetry of waveguide bend has the directivity of light field transmission on the structure;

2 distributed feedback inhibition semiconductor optical amplifiers, its chip structure is: the input and output side at active waveguide has the straight wave guide part respectively, to improve the coupling of waveguide input, output mould field and coupled fiber mould field;

3 distributed feedback inhibition semiconductor optical amplifiers, its chip structure is: the chip end face adopts the perpendicular end surface of natural cleavage plane, under the condition that needs the higher degree feedback inhibition, adopts inclined end face;

4 distributed feedback inhibition semiconductor optical amplifiers, its chip structure is: the sweep of active waveguide has the radius-of-curvature of variation.Progressively reduce along light field direction of propagation radius-of-curvature, adopt the smooth transition that is connected in the curved waveguide and the junction of straight wave guide.

Images