Analysis and Design of Vertical Cavity Surface Emitting Lasers

Chapter 4.2.2 - Modeling of Self- and Cross-Gain Saturation Coefficients

4.2.2 Modeling of Self- and Cross-Gain Saturation Coefficients

General expressions for the linear and nonlinear optical susceptibility tensors can be deduced using the density matrix approach [22]. The quantum-mechanical Liouville equation for a density matrix ρ is given by

where k can also be interpreted as the wavevector of the crystallographic direction, say, [001]. The Hamiltonian operator H(k) consists of three parts:

Where H0 is the unperturbed Hamiltonian for electron and hole and H' takes into account the interaction of matters such as electron–photon interaction, and can be expressed as

where M is the dipole operator and Hrelax includes the relaxation effects due to incoherent scattering process. Therefore, the general expression for the Liouville equation is given by

and

where ρ(0) is the density matrix at thermal equilibrium. It is noted that the total electron wavefunctions can be expanded in terms of Ψp, where Ψp, refers to the wavefunction of the pth band. If the density matrix in this basis function is written as Ψpρ (k, t)Ψq ≡ ρp,q (k, t), the corresponding quantum mechanical Liouville equation can be expressed as

where γp,q is the population relaxation rate ( p = q) or phase relaxation rate ( p ≠ q)· Please note that this perturbation approximation is applied only to the case of weak optical fields.

In order to solve the quantum-mechanical Liouville equation, the perturbation method is again used to expand the density matrices as below:

in which diagonal components are the Fermi distributions in each band with vanishing off-diagonal components. Substituting (4.22) into (4.21) gives

for n = 0,1,2 and Ψpρ(n)Ψq = is defined in the calculation. If it is assumed that

where ωp,q ( = Ep – Eq/ħ ) is the oscillation frequency of the generated photon due to the recombination of carriers between energy levels Ep and Eq, (4.23) can be expressed in terms of basis functions as shown below:

Now, if the optical field E (r, t) and the density matrix are written in Fourier expressions

(where c.c. = complex conjugate) then, by substituting (4.26), (4.27) into (4.25) and matching the time-dependent exponential terms, it can be shown that

for ωη = ωv or ωh. Hence, the frequency dependence of can also be solved in a similar manner, provided the solution of is known.

The polarization per unit volume P (ω) is calculated from the trace of the matrix product of the dipole moment matrix M and the density matrix ρ:

where Lw is the width of QW

and Mp,q is a 3 X 3 matrix. The subscripts c and v represent the conduction and valence bands, respectively. Hence, the ith component of the polarization density, to the first order in the optical field, is given by

and, with the definition of the component of electric susceptibility tensor of first order, , it is expressed as

Compare (4.31) and (4.32) and using (4.28), the linear susceptibility tensor of the QW is given by

where i and i' are the crystallographic axis directions (i.e., x, y, or z direction). The summation indices c and v run over conduction subbands and valence sub-bands, respectively. They are the lowest subbands (spin-degenerate two bands, c±) in the conduction band and the six subbands (three spin-degenerate pairs: v1±, v2± and v3±) in the valence band. are the dipole transition matrix elements in i direction. are the diagonal components of the lowest order density matrix in the subbands c±(v1±, v2±, v3±), which are Fermi distribution functions with specified Fermi levels and temperatures. ħµc,v (k) is the energy difference between the conduction and valence bands:

where Ec (Ev) is the energy of the electron (hole) with wave vector k. Higher-order susceptibility tensor in the frame of (,,) can also be deduced in a similar manner, and their expressions are given in Section 4A.1.

In the derivation above, the linear and nonlinear susceptibility tensors are given in the frame of (, , ) but the natural coordinate system of VCSELs is spanned by (, ,), which are used to express the optical gain and saturation coefficients in the rate equation model. The coordinate transformation from (,,) to (, ,) is a simple rotation by 45° on z, which is expressed by a 3 X 3 matrix R [23]. For example, it can be shown that

where η = v or h.

Evaluation of the expression of χ(1) and χ(3) requires the knowledge of the dipole transition matrix elements Mc,ν (k) and the energy dispersions Ec,v (k), which can be obtained from the band calculations. The energy dispersion of the conduction band can be approximated by the parabolic band with effective masses for the well and barrier materials [22]. The valence band states are calculated by the envelope function approximation with the 4 x 4 Luttinger–Kohn Hamiltonian [22]. The spin-orbit splitoff band can be ignored in the envelope function approximation, as its contribution to the gain saturation is small in the energy range of interest. However, the axial approximation has not been used because the anisotropy in the (kx, ky) plane is considered to be important for the polarization dependence of saturation coefficients. Detailed modeling of energy band structure of QWs can be found in section 4A.2.

Consider the numerical calculation of linear gain and saturation coefficients of a In0.732Ga0.268As0.582P0.418/InP QW material at operation wavelength around 1.3 µm. The concentrations of Ga and As are chosen so that the well material is lattice-matched to InP. In the calculation, the conduction band offset and the thickness of the well are set to 0.4 and 10 nm, respectively. Inside the well (barrier), the energy bandgap, dielectric constant ε0, and effective mass of electron are assumed to be 0.95 eV (1.35 eV), 13.3 (12.4), and 0.0573 (0.08), respectively. In addition, the Luttinger parameters of the well (barrier) are assumed to be Γ1 = 7.8 (4.95), Γ2 = 2.8 (1.65), and Γ3 = 3.6 (2.35) for the InGaAsP/InP QW. Figure 4.2 plots the gain spectra of the two orthogonal polarized modes varying with sheet density of carrier concentration at room temperature. Because of the crystal isotropy along any of the transverse crystallographic axes of the (001) plane, the linear gain of h mode [gh (ω)], and that of v mode, [gv (ω)], are identical. Two peaks at photon energies of 1.028 and 1.052 eV are observed from the gain spectra at sheet density of 3 X 1012 cm¯2. The band calculation shows that the first subband state (v1) in the valence band is dominated by v; , ±components and the second subband (v2) by v; , ±½ components in the vicinity of the zone center. The transition between the conduction (c) and (v1), which is c; ½,±½ → v; , ± dominant, leads to the first gain peak and the second peak is caused by the transition between c–v2, which is c; ½,±½ → v; , ±½. These selection rules also affect the spectra of saturation coefficients. Figure 4.3 shows the spectra of the corresponding self-saturation coefficient of the two orthogonal polarizations with different sheet density concentrations. Again because of the crystallographic isotropy, the self-saturation coefficients of the h and v modes [ζsh and ζsv], are identical for the entire spectrum. A peak is observed near photon energy of 1.02 eV, which implies that the c – v1 transition has influence on the self-saturation coefficient.

There are two types of cross-saturation coefficients: (1) ζcvh, which represents the gain saturation of lasing light Sv caused by the injected orthogonal polarization Sh, is the saturation coefficient for two orthogonal polarizations; and (2) ζchv, which has a meaning similar to those inferred from indices. These two cross-saturation coefficients are the function of two optical frequencies, ωv and ωh. It is noted that the cross-saturation coefficients are maximized only when both

Figure 4.2 Plot of linear gain spectra of h and υ modes in InGaAsP/InP quantum wells at room temperature (after Ref. 20).

Figure 4.3 Self-saturation coefficients of h and υ-modes at resonance (i.e., ωh ≈ ωυ ) in InGaAsP/InP quantum wells (after Ref. 20).

Figure 4.4 Cross-saturation coefficients of h and υ-modes at resonance (i.e., ωh ≈ ωυ) in InGaAsP/1nP quantum wells (after Ref. 20).

photon energies are in resonance (i.e., ωv ≅ ωh) but vanish at the other photon energies. Furthermore, due to the isotropy of the crystal structure, ζchv = ζcvh is expected and the corresponding spectra of ζchv = (ζcvh) at resonance is shown in Figure 4.4. It is noted that the profile of cross-saturation coefficients at resonance is dependent on the photon energies and has a peak at 1.02 eV, which indicates the domination of c – v1 transition.

From Figures 4.3 and 4.4, it can be concluded that although the optical gains of the two orthogonal polarizations are identical, the presence of self-and cross-saturation coefficients may induce optical gain anisotropy due to the high polarization intensity at a particular resonance frequency. In addition, the in-plane orientation of the two orthogonal polarizations can be defined by the external injection of the polarized field. Hence, this formation of in-plane optical gain anisotropy can be used to control the polarization characteristics of VCSELs such as polarization switching and polarization bistability [19,20]

UNLIMITED FREE
ACCESS
TO THE WORLD'S BEST IDEAS

SUBMIT
Already a GlobalSpec user? Log in.

This is embarrasing...

An error occurred while processing the form. Please try again in a few minutes.

Customize Your GlobalSpec Experience

Category: Light Emitting Diodes (LED)
Finish!
Privacy Policy

This is embarrasing...

An error occurred while processing the form. Please try again in a few minutes.