Numerical study on the optical and carrier recombination processes in GeSn alloy for E-SWIR and MWIR optoelectronic applications

Stefano Dominici; Hanqing Wen; Francesco Bertazzi; Michele Goano; Enrico Bellotti

doi:10.1364/OE.24.026363

1. Introduction

Germanium-based semiconductor compounds have been considered as potential candidates for optoelectronics applications on silicon platforms. In particular, alloys of germanium and tin - a IV group semi-metal - have always been regarded with great interest due the possibility of obtaining a tunable direct band gap. Experimental investigations have shown that the fundamental energy gap changes from indirect to direct as the Sn molar fraction increases. [1] In fact, the incorporation of Sn in Ge reduces the energy gap between the conduction band minima of Γ and L valleys up to the point where the crossover take place. A strong interest toward Ge_1−xSn_x is clearly due to its potential compatibility with complementary metal-oxide semiconductor (CMOS) technology and the possibility of monolithic integration on Si-based electronics. [2–4]

The early work of Goodman [5] provided the motivations to further studies on the electronic properties of GeSn. From an experimental perspective, the growth of GeSn layers characterized by high lattice crystallinity and low structural disorder has always been challenging. Due to the limited 1% equilibrium solubility of Sn in Ge, [6] and to the instability of α-Sn at high temperatures, which quickly degrades to β-Sn, experimentalists had to rely upon non-equilibrium growth techniques, such as molecular beam epitaxy. [7] Furthermore, problems of compositional homogeneity and phase purity prevented the formation of a stable alloy [8] above 20% of Sn composition. He and coworkers [8] performed one of the first growth and characterization of Ge_1−xSn_x films at different molar fractions. Of the characterized samples, they determined the optical absorption coefficient, then correlated the low energy tail in the absorption spectrum to the structural disorder of the samples through an Urbach tail model. The persistent presence of misfits dislocation defects in fabricated Ge_1−xSn_x samples have been confirmed in by Bauer and coworkers [2, 3] by TEM analysis, detecting also possible phenomena of phase separation between Ge and Sn. Novel deposition techniques allowed to grow pseudomorphic Ge_1−xSn_x on Ge(001) [9] and relaxed Ge_1−xSn_x on Si(001) [4]. Furthermore, ad-hoc techniques for the incorporation of n-type and p-type dopant species in Ge_1−xSn_x lattice have been developed. [7,10–12] The performances of Ge_1−xSn_x based p-i-n structures, grown on Si(100) substrate, for photodetection applications have been studied and presented in a number of articles. [13–15] Finally, S. Wirths, D. Buca, and S. Mantl [16] have compiled and extensive review on the history, fabrication techniques, and applications of Si-Ge-Sn based alloys.

The electronic properties of Ge_1−xSn_x have been investigated using a variety of theoretical models, including ab-initio techniques. [17–20] A great deal of effort has been directed toward the study of the energy dependence of the two conduction band minima, at L and Γ, as a function of the Sn molar fraction and the effect of an applied strain. However, it appears that the cluster expansion method adopted by Freitas and coworkers [21] found an agreement with the majority of experimental results. In their framework, the fundamental gaps at Γ and L are considered statistical quantities characterized by an average and a standard deviation as reported as shaded regions in Fig. 1. This statistical approach makes it possible to deal efficiently with the variety of experimental data, encompassing most of them within their range of uncertainty.

Fig. 1 Experimental and calculated energy gaps at Γ and L valleys as a function of the Sn molar fraction. Shaded area and dashed lines represent standard deviation and average value of results from Freitas and coworkers. [21] The solid lines represent the values computed in this work. A number of values presented in literature for the energy gap have been reported. [4,8,12,22–26]

Download Full Size | PDF

The goal of this works is to provide a comprehensive study of the optical properties and carrier recombination mechanisms in Ge_1−xSn_x alloy for applications in the extended-short (E-SWIR) and medium (MWIR) infrared wavelength. Specifically, we consider alloys characterized by an energy gap with a nominal cutoff wavelength of 2.7 μm and 5 μm at temperatures of 240 K and 140 K, respectively. This choice has been driven by the potential use of these materials for imaging applications where layers with these alloy compositions can be integrated on silicon read-out circuits. Using a numerical model we determine the absorption coefficient, the radiative recombination rate, and the Auger recombination rate of the material under different injection levels. We investigate the alloy properties in different doping conditions: intrinsic, p-doped and n-doped with concentrations of 10¹⁵ cm⁻³ and 10¹⁷ cm⁻³, for both cases. While the numerical model can deal with strained material as well, for detector applications, where thick layer are needed, we speculate that the material will likely be relaxed. Therefore, we restrict the present effort to relaxed material and present the study of the strained material in a following work.

The manuscript is organized as follows: Section 2 will describe the numerical model used to determine the electronic properties, the absorption coefficient, and the recombination rates of Auger and radiative processes. Section 3 discusses the results and their implications, assessing the role played by radiative and non radiative processes at different injection levels. Finally, Section 4 will conclude.

2. Numerical model

We have determined the electronic structure of Ge_1−xSn_x alloys throughout the full composition range using a modified empirical pseudopotential method (EPM) [27] as implemented by Wen and Bellotti. [28, 29] The numerical model explicitly accounts for alloy disorder by means of two parameters: a disorder and spin parameters, respectively. These parameters have been used in the quadratic interpolation of atomic and spin-orbit screened atomic potentials as follows.

P_{i}^{(1, 2)} = P_{i}^{(1, 2)} - d_{(D, S O)} [x (1 - x)] (P_{i}^{(1)} - P_{i}^{(2)}) .

In Eq. (1)d_D is the disorder parameter, d_SO the spin parameter, x is the Sn molar fraction, and $P_{i}^{(1)}$ and $P_{i}^{(2)}$ are the screened atomic potentials for each atomic specie - either Sn or Ge. We set the disorder and spin disorder parameters in order to reproduce the trend of the fundamental energy gaps, measured at L and Γ valleys, to the average values given by Freitas and coworkers [21] as reported in Fig. 1. Therefore, the tin molar fractions that correspond to the cutoff wavelengths of 2.7 and 5 μm are 9% and 18% at the corresponding temperatures of 140 K and 240 K, respectively. We have included the effect of temperature on the energy gap through a rigid shift of all conduction states. We consider first the energy shift required to match the experimentally measured energy gap of pure Ge at 300 K [30] and linearly interpolated it to the temperatures of 140 K and 240 K. [12] Fig. 2 provides the value of the tin molar fraction, and corresponding intrinsic carrier concentration, as a function of the temperature needed to achieve a cut off wavelengths of 2.7 and 5 μm in the E-SWIR and MWIR spectral regions. Specifically, on the left vertical axis we have the molar fraction required to achieve the chosen band gap at a given temperature, and on the right vertical axis the corresponding intrinsic carrier concentration at a given temperature.

Fig. 2 Sn molar fraction required to obtain the proper cutoff wavelength, function of temperature, and the corresponding intrinsic carrier concentration.

Download Full Size | PDF

2.1. Auger recombination rate and lifetime

Theoretical models of Auger processes were first developed by Beattie and Landsberg. [31] Within the framework of second order perturbation theory (SOPT) they proposed an analytical expression for Auger lifetimes in semiconductors, normally referred to as the BLB model. However, their model assumes parabolic bands and applies to direct gap semiconductors. Successively, a Green’s function based approach was introduces by Takeshima [32,33] and Bardyszewsky and Yevick [34]. This approach overcomes the numerical singularity of SOPT and allow to embed the contribution of carrier-phonon interactions into the spectral density function. In this work, we have employed a Green’s function based model to determine the Auger recombination rate and lifetime for the bulk material. We considered two categories of Auger processes: direct and indirect. In the first category, the transition is mediated just by Coulomb interactions while in the second an additional interaction, such as phonon or impurity scattering, takes part in the process. Furthermore, we have calculated separate contributions for electron and hole based processes, respectively labeled as eeh and hhe. The numerical model we have employed in based on the formalism proposed by Takeshima, [32,33] successively extended to the case of full band structure by Wen and coworkers. [28,29,35–37] The total Auger recombination rate per unit volume and time in bulk material is given by:

\begin{array}{l} R_{A R} = \frac{2 π}{ℏ} \frac{V^{3}}{{(2 π)}^{9}} [1 - e^{(μ_{v} - μ_{c}) / k_{B} T}] \int d k_{1} d k_{2} d k_{1^{'}} d k_{2^{'}} | M_{e e} |^{2} δ (k_{1} + k_{2} - k_{1^{'}} - k_{2^{'}}) \\ \times \int d E_{1} d E_{2} d E_{1^{'}} d E_{2^{'}} Θ (E_{1}) Θ (E_{2}) [1 - Θ (E_{1^{'}})] [1 - Θ (E_{2^{'}})] \\ \times Im G_{l_{1}}^{R} (k_{1}, E_{1}) Im G_{l_{2}}^{R} (k_{2}, E_{2}) Im G_{l_{1^{'}}}^{R} (k_{1^{'}}, E_{1^{'}}) Im G_{l_{2^{'}}}^{R} (k_{2^{'}}, E_{2^{'}}) . \end{array}

where

Im G_{l_{1^{'}}}^{R} (k_{1^{'}}, E_{1^{'}})

is the spectral density function representing the energy broadening of state l_1′ at k_1′, Θ(E) represent the Fermi-Dirac occupation statistical factor, |M_ee|² is the matrix element for the Auger process representing the square overlap integral between the four state wavefunctions involved in the transition. The Coulomb interaction between carriers with different wave-vectors is screened by a wave-vector dependent, static dielectric function, implemented using a analytical expression given by:

ε (q) = 1 + \frac{c_{1}}{| q |^{c_{2}} + c_{3}},

in which c₁, c₂ and c₃ are parameters obtained by fitting results from random phase approximation [38]. The presence of the twelve-fold k-space integrals in Eq. (2) makes it very difficult to use a direct integration approach. The numerical evaluation of Eq. (2) is performed using an hybrid approach that employs a Monte Carlo procedure and a direct integration based on the tetrahedron method [39]. The indirect processes are evaluated including the contribution of carrier-phonon interaction through four self-energy functions, introducing an energy broadening for the states involved in the transition (Sec. 2.3). From the knowledge of the total recombination rate it is possible to compute the Auger coefficients C_n and C_p starting from the expression:

R_{A R} = (n C_{n} + p C_{p}) (n p - n_{i}^{2}),

In the same way, the carrier lifetimes can be obtained for an arbitrary doping and injection condition from the expression:

τ_{eeh} = \frac{n p - n_{i}^{2}}{n R_{A R}^{(eeh)}}, τ_{hhe} = \frac{n p - n_{i}^{2}}{p R_{A R}^{(hhe)}} .

2.2. Absorption coefficient and radiative recombination rate and lifetime

Unlike Auger, radiative recombination processes are mediated by dipole interactions, where a photon is absorbed or emitted to conserve the total energy of the transition. A Green’s function based model has already been presented in [28] and applied by Wen and coworkers to determine the optical properties of InAs₁₋_xSb_x alloy, silicon and germanium [28,40]. The radiative recombination rate is given by:

\begin{array}{l} R_{R R} = \frac{2 e^{2} n_{r} ω_{p h}}{π ℏ m_{0}^{2} c_{0}^{3} V ε_{0}} \sum_{k_{v}, k_{c}} | 〈 k_{v} | \hat{e} \cdot P | k_{c} 〉 |^{2} \int d E_{1} \int d E_{2} Θ (E_{2}) [1 - Θ (E_{1})] \\ \times δ (k_{c} - k_{v} - k_{ph}) δ (μ_{c} - μ_{v} + E_{2} - E_{1} - ℏ ω_{p h}) Im G_{l_{v}}^{R} (k_{v}, E_{1}) Im G_{l_{c}}^{R} (k_{c}, E_{2}) . \end{array}

The the absorption coefficient can be directly evaluated using Eq. (6) with a different pre-factor and accounting for the proper initial and final valence and conduction states. The carrier-phonon interaction is included through two self-energy functions, one for conduction and one for valence states. The classical result for the direct process, obtained from Fermi golden rule can be retrieved by substituting the self-energies with delta functions. Our results include broadening for both initial and final states. We calculate the radiative coefficient and lifetimes, assuming the same concentration of electrons and holes, as follows:

τ_{n} = \frac{n p - n_{i}^{2}}{n R_{R R}}, τ_{p} = \frac{n p - n_{i}^{2}}{p R_{R R}} .

2.3. Spectral density function

The core of the Green’s function based approach consists in the evaluation of the spectral density function, related to the imaginary part of the retarded Green’s functions $(Im G_{l_{i}}^{R})$ , that depends on the system self-energy (Σi).

Im G_{l_{i}}^{R} (k, E) = - \frac{1}{π} \frac{Im Σ_{i} (k, E)}{{[E - E_{i} - Re Σ_{i} (k, E)]}^{2} + {[Im Σ_{i} (k, E)]}^{2}} .

The imaginary part of the retarded Green’s function, presented in Eq. (8), has a lorentzian distribution shape, centered at E_i. The real and imaginary part of the self-energy term describe respectively the shift and the broadening of the distribution over the energy domain, modeling the effects of all perturbations (e.g. carrier-phonon scattering) acting on the system. The conventional approach to evaluate the self-energy is by means of the Dyson’s equation, truncating the expansion up to the desired order [28,32]. In the case of our interest, namely the calculation of Auger and ratiative recombination rates and absorption coefficient, we have truncated the expansion to the lowest order [29]. The interactions taken into account to evaluate the self-energy are non-polar optical and acoustic phonon scattering mechanisms. Their numerical contribution to the self-energy is obtained by using the appropriate matrix element, given by [28]:

| g_{AC} (q) |^{2} = \frac{Ξ_{d}^{2} ℏ ω_{a c} (q)}{2 c_{l}} \frac{q^{4}}{{(q^{2} + λ^{2})}^{2}},

| g_{NPO} (q) |^{2} = \frac{ℏ D^{2} v_{s}^{2}}{2 \bar{c} ω_{o p} (q)} \frac{q^{4}}{{(q^{2} + λ^{2})}^{2}},

In Eqs. (9)–(10), c_l and c_t are the longitudinal and transverse elastic constants, $\bar{c} = c_{l} / 3 + 2 c_{t} / 3$ is the average lattice constant, v_s is the sound velocity. Table 1 reports the parameters used to evaluate the matrix elements for each mechanism presented in Eqs. (9)–(10).

Table 1. Alloy parameters for E-SWIR (9%) and MWIR (18%) Ge_1−xSn_x. [30, 41] The values for the alloy has been obtained using a linear interpolation in the molar fraction.

View Table | View all tables in this article

Earlier calculations [42–45] made use of a simplified analytical model for the band structure (i.e. effective mass approximation). Their motivation was to avoid costly numerical calculations and investigate the limit cases. The use of analytical bands within the effective mass model is a drastic approximation, both for Ge and GeSn. In fact, in these materials both the Γ and L valleys significantly contribute the conduction band density of states, and consequently the number of available final states for the Auger transition, making both the direct and indirect processes important. Furthermore, the valence band structure and the relevant matrix elements play a significant role in determining the strength of direct and indirect transitions.

3. Numerical results

Figure 1 presents a comparison between the calculated and experimental [4,8,12,22–26] values of the energy gap values for Ge_1−xSn_x and the results of our EPM model. Based on our results, we can see that Ge_1−xSn_x is an indirect energy gap material at the E-SWIR cut off and a direct energy gap material in the MWIR case.

We have calculated the electronic structure of Ge_1−xSn_x for a Sn composition between 0% and 20% and evaluated the effective mass variations as a function of the molar fractions at Γ and L valley minimum. Subsequently, we have also averaged the effective masse values along different principal directions, that for the zincblende reciprocal lattice are denoted as: Σ, Δ, and Λ. For the i-th band the mass is computed determining the second derivative, as in Eq. (11), of the band profile using a second order interpolation close to the minima.

\frac{1}{m_{i}} = \frac{1}{m_{i}^{(Σ)}} \frac{1}{m_{i}^{(Δ)}} \frac{1}{m_{i}^{(Λ)}}

m_{i} = m_{i}^{0} + x m_{i}^{1}

Once the effective masses are known for a discrete number of molar fractions a linear interpolation, as in Eq. (12), can be used to obtain values for arbitrary compositions. First and zeroth order fitting parameters for heavy hole (HH) band, light hole (LH) band, spin-orbit (SO) band, and first conduction band are reported in Tab. 3–2. The direction averaged effective mass function of Sn molar fraction for HH band, LH band, SO band, and first conduction band are presented in Fig. 3. The effective mass for HH band, LH band, and first conduction band at Γ decrease for increasing values of molar fraction, while the effective mass for SO band and first conduction band at L increase for increasing values of molar fraction.

Table 2. Zeroth order interpolation coefficients for direction dependent effective masses. Curves were fitted to Sn molar fraction comprised between 0% and 20%.

View Table | View all tables in this article

Table 3. First order interpolation coefficients for direction dependent effective masses. Curves were fitted to Sn molar fraction comprised between 0% and 20%.

View Table | View all tables in this article

Fig. 3 Effective masses at critical points for Sn molar fractions comprised between 0% and 20%.

Download Full Size | PDF

3.1. Optical emission and absorption properties

We have also computed the optical emission and absorption properties, commonly referred to as the radiative recombination rate and absorption coefficient, of intrinsic Ge_1−xSn_x with cutoff in the E-SWIR (Ge_0.91Sn_0.09) and MWIR (Ge_0.82Sn_0.18) spectral range at a temperature of 240K and 140K, respectively. Figure 4 presents the calculated absorption coefficient for Ge_0.91Sn_0.09 and Ge_0.82Sn_0.18 Ṫhe total absorption coefficient is represented by solid lines, while direct and indirect contributions are represented by dotted and dashed lines, respectively. The trend of direct and indirect components of the absorption coefficients in Fig. 4 confirms an indirect energy gap for the E-SWIR material in the E-SWIR, and a direct energy gap for the MWIR.

Fig. 4 Absorption coefficient for intrinsic Ge_1−xSn_x in E-SWIR and MWIR configurations.

Download Full Size | PDF

We have also calculated the radiative recombination rates for Ge_0.91Sn_0.09 and Ge_0.82Sn_0.18 through a direct numerical evaluation of Eq. (6) and reported the results in Fig. 5 and Fig. 6, respectively. Furthermore, we have considered the material under different injection levels: from 10¹⁶ cm⁻³ up to 5×10¹⁹. In case of Ge_0.91Sn_0.09 (Fig. 5) to an increase in the injection level we observed an increase in the overall rate together with a minor blue shift of the wavelength at which the emission peak occurs. Instead, in case of Ge_0.82Sn_0.18 (Fig. 6) to an increase in the injection level we observed a strong blue shift of the peak wavelength and correlated it to the contribution coming from the indirect L valley, whose minimum is located at an energy higher than Γ valley minimum. The non-zero contributions below the energy gap, as seen in Fig. 5, correspond to recombination processes close to the band minima assisted by phonon emission/absorption.

Fig. 5 Radiative recombination rate for intrinsic Ge_1−xSn_x in the E-SWIR configuration at 240K under different injection conditions.

Download Full Size | PDF

Fig. 6 Radiative recombination rate for intrinsic Ge_1−xSn_x in the MWIR configuration at 140K under different injection conditions.

Download Full Size | PDF

3.2. Auger and radiative recombination

The total Auger recombination rate is calculated separately for electron and holes, the eeh and hhe Auger processes, as a sum of direct and indirect contributions. For each carrier type, we have considered two different processes, classified by the energy band of the secondary particle final state: first and second conduction bands for eeh processes and first and second valence bands for hhe processes. Furthermore, when dealing with a doped semiconductor, we have calculated the Auger recombination rate for the dominant mechanism, determined by the type of minority carriers: eeh for p-doped and hhe for n-doped. The Auger lifetimes are obtained using Eq. (5) once the rate is known from the numerical integration of Eq. (2). To perform a quantitative comparison, we also determine the radiative lifetime that is obtained directly from Eq. (7) using the radiative recombination rate computed using Eq. (6).

For the Auger rate, the convergence of the Monte Carlo integration is achieved once the ratio between the standard deviation and the average value of the result is below 0.1%. Moreover, we have considered doping concentrations of 10¹⁵ cm⁻³ and 10¹⁷ cm⁻³, for both n-type and p-type, and the excess carrier concentration to values ranging from 10¹⁴ cm⁻³ to 10¹⁷ cm⁻³ for the lower doping and from 10¹⁶ cm⁻³ to 5×10¹⁹ cm⁻³ for the higher one. Figures 7 and 10, present the calculated radiative and Auger lifetimes for doped Ge_0.91Sn_0.09 and Ge_0.82Sn_0.18 compositions.

Fig. 7 Auger and radiative lifetimes function of excess carrier concentrations for doped Ge_0.91Sn_0.09 (E-SWIR configuration). The doping level is 10¹⁵ cm⁻³ for both n-type and p-type.

Download Full Size | PDF

Fig. 8 Auger (eeh and hhe) and radiative lifetimes over a wide range of excess carrier concentrations in case of doped Ge_0.91Sn_0.09 (E-SWIR configuration). The doping level is 10¹⁷ cm⁻³ for both n-type and p-type.

Download Full Size | PDF

Fig. 9 Auger (eeh and hhe) and radiative lifetimes over a wide range of excess carrier concentrations in case of doped Ge_0.82Sn_0.18 (MWIR configuration). The doping level is 10¹⁵ cm⁻³ for both n-type and p-type.

Download Full Size | PDF

Fig. 10 Auger (eeh and hhe) and radiative lifetimes over a wide range of excess carrier concentrations in case of doped Ge_0.82Sn_0.18 (MWIR configuration). The doping level is 10¹⁷ cm⁻³ for both n-type and p-type.

Download Full Size | PDF

We have checked that all the important contributions have been taken into account for both eeh and hhe Auger transitions. In case of eeh processes, the large majority of energy states in the third conduction have energies higher than the gap. Therefore, only a limited set of final states is available in this band, resulting in a negligible contribution to the total Auger rate. On the other hand, it is worthwhile to investigate the contribution of CHSH transitions, or hhe processes, in which the final state is found in the split-off band. [22] Since the spin-orbit splitting energy in Ge_1−xSn_x increases with molar fraction, [22] the transition rate of hhe processes having the split-off band as final state is expected to decrease as well. [46]

We calculated the contribution to the Auger coefficients of direct and indirect hhe transitions for processes where the final state for the scattered hole is located in the split-off band. We have done so in the case of intrinsic Ge_1−xSn_x for x = 0.04, x = 0.06, and x = 0.08, at 300 K, and under four excess carrier concentrations, namely n = p = 10¹⁶ cm⁻³, n = p = 10¹⁷ cm⁻³, n = p = 10¹⁸ cm⁻³, and n = p = 10¹⁹ cm⁻³. From the results reported in Fig. 11, we are able to conclude that the contribution of CHSH transitions diminishes with Sn molar fraction, regardless of the relative magnitude of the split-off energy and gap.

Fig. 11 Auger recombination coefficient as a function of excess carrier concentration for intrinsic Ge_1−xSn_x at 300 K and three molar fractions: x = 0.04, x = 0.06, and x = 0.08. The Auger coefficients decrease with increasing Sn molar fractions and the contribution from CHSH processes in case of Ge_0.91Sn_0.09 and Ge_0.82Sn_0.18 is negligible.

Download Full Size | PDF

Considering the case of Ge_0.91Sn_0.09 with a doping concentration of 10¹⁵ cm⁻³, as shown in Fig. 7, we observe that the radiative lifetime is smaller than Auger lifetimes at excess carrier concentrations below 10¹⁷ cm⁻³. Therefore the radiative recombination process is the dominant mechanism in low injection regime. In the case of Ge_0.91Sn_0.09 doped at the 10¹⁷ cm⁻³ level, as we observe in Fig. 8, the radiative processes dominates over the Auger processes at excess carrier concentrations lower than 10¹⁷ cm⁻³ and that the two Auger processes - eeh and hhe -are rather similar in magnitude at any given excess carrier concentration. Considering instead the case of a Ge_0.82Sn_0.18 with a doping concentration of 10¹⁵ cm⁻³, we observe in Fig. 9 that the radiative lifetime is smaller than Auger lifetimes at excess carrier concentrations below 10¹⁶ cm⁻³. However, the hhe Auger recombination processes still play a significant role in the low injection regime due to the small difference between radiative and hhe Auger lifetimes. In case of Ge_0.82Sn_0.18 with a 10¹⁷ cm⁻³ doping level, as it can be seen in Fig. 10, the radiative process becomes less important than the Auger processes for all the excess carrier concentrations considered and that the hhe Auger process is the overall dominant recombination mechanism. Furthermore, it is possible to observe a saturation behavior for the dominant Auger processes for increasing excess carrier concentrations and a change of monotonicity in the radiative lifetime for excess carrier concentrations above 10¹⁸ cm⁻³. As a result, for excess carrier concentrations above 10¹⁸ cm⁻³ the radiative recombination rate decrease and the corresponding lifetime slightly increases. We report results on the radiative recombination rate, function of excess carrier concentration, and separate contribution from indirect (phonon-assisted) and direct processes in Fig. 12 and 13.

Fig. 12 Auger and radiative recombination rate function of excess carrier concentrations for doped Ge_0.91Sn_0.09 (E-SWIR configuration). The doping level is 10¹⁷ cm⁻³ for both n-type and p-type. The first label in each legend entry refers to indirect (i) and direct (d) processes, respectively. The second label refers to radiative (r) or Auger class (eeh or hhe). The last label, reported for radiative processes only, refers to the type of doping (n or p respectively).

Download Full Size | PDF

Fig. 13 Auger and radiative recombination rate function of excess carrier concentrations for doped Ge_0.82Sn_0.18 (MWIR configuration). The doping level is 10¹⁷ cm⁻³ for both n-type and p-type. The first label in each legend entry refers to indirect (i) and direct (d) processes, respectively. The second label refers to radiative (r) or Auger class (eeh or hhe). The last label, reported for radiative processes only, refers to the type of doping (n or p respectively).

Download Full Size | PDF

We speculate that the driving factor behind the differences between the lifetime trends of Ge_0.91Sn_0.09 and Ge_0.82Sn_0.18 compositions can be traced back to the difference in the energy gaps and the position of Γ with respect to the L minima. In fact, considering the small difference in the effective masses between the two compositions (as reported in Fig. 3) the curvature of the band structure around the Γ and L minimum undergoes a small variations going from a composition to the other. On the other hand, the reduction of the energy gap and the different positions of Γ and L valley minimum strongly affects the quasi-Fermi energies and the carrier distribution in the k-space. We have also computed the radiative and Auger coefficients, to be used in device simulation models, accordingly to the following expressions:

B = \frac{R_{R R}}{n p - n_{i}^{2}},

C_{eeh} = \frac{R_{A R}^{(eeh)}}{n (n p - n_{i}^{2})},

C_{hhe} = \frac{R_{A R}^{(hhe)}}{p (n p - n_{i}^{2})} .

Equations (13)–(15) ascribe the dependence of radiative and Auger recombination rates to second and third order polynomials on free carrier concentration, respectively. These equations are used to fit data on total recombianation rate, without distinction between direct and indirec processes. Given the total recombination rate from the numerical evaluation of Eq.(2) and Eq.(6), we calculate the Auger and radiative recombination coefficients directly from Eqs.(13)–(15). Therefore, coefficients calculated through Eqs.(13)–(15) are valid as long as the recombination rate computed from Eq. (2) and Eq. (6) are physically accurate. Figures 14 and 17 present the calculated values, as a function of injection level, for doped Ge_1−xSn_x in E-SWIR and MWIR configuration. The same Auger recombination model and calculation methodology have been used to evaluate Auger coefficients in relaxed and strained Ge [47]. As it can be observed in Fig. 14 and Fig. 16, in the low injection regime and doping concentration of 10¹⁵ cm⁻³ both radiative and Auger coefficient are almost constant with injection level. However, radiative and Auger recombination coefficients in Ge_0.82Sn_0.18 exhibit significant variations for injection levels higher than 10¹⁶ cm⁻³. The complexity of Eq. (2), in which many different contributions are present, prevents us from being able to give a simple explanation of the differences between the calculated Auger coefficients for Ge_0.82Sn_0.18 and Ge_0.91Sn_0.09. Therefore, we have performed an additional number of simulations to discern the relative importance of the various contributions. From our analysis, we have concluded that the band structure changes, from Ge_0.82Sn_0.18 to Ge_0.91Sn_0.09, due to different molar fractions is the dominant factor, while differences in the activation energy (i.e. energy gap and phonon energies) play a lesser role. We explain the reduction of Auger coefficient in case of Ge_0.82Sn_0.18 at an injection level higher than 10¹⁷ cm⁻³ as a saturation in the number of final states available to the secondary particle. The number of these states correlates to the Auger lifetime and the dependence of Auger coefficients on the lifetimes is function of the inverse square of excess carrier concentration (obtained by combining Eqs. (14)–(15) with Eq. (5)). Therefore, the reduction of Auger coefficients in Fig. 17 is the direct consequence of Auger lifetimes saturation in Fig. 10.

Fig. 14 Radiative recombination coefficient function of injection level for doped Ge_1−xSn_x in E-SWIR and MWIR configurations. The doping level is 10¹⁵ cm⁻³ for both n-type and p-type.

Download Full Size | PDF

Fig. 15 Radiative recombination coefficient function of injection level for doped Ge_1−xSn_x in E-SWIR and MWIR configurations. The doping level is 10¹⁷ cm⁻³ for both n-type and p-type.

Download Full Size | PDF

Fig. 16 Auger recombination coefficient function of injection level for doped Ge_1−xSn_x in E-SWIR and MWIR configurations. The doping level is 10¹⁵ cm⁻³ for both n-type and p-type.

Download Full Size | PDF

Fig. 17 Auger recombination coefficient function of injection level for doped Ge_1−xSn_x in E-SWIR and MWIR configurations. The doping level is 10¹⁷ cm⁻³ for both n-type and p-type.

Download Full Size | PDF

4. Conclusions

We have presented a numerical study of the absorption coefficient, radiative and Auger recombination properties of intrinsic and doped Ge_1−xSn_x at two Sn molar fractions, 9% (E-SWIR) and 18% (MWIR), and temperatures of 240K and 140K, respectively. We have computed the absorption coefficient and radiative recombination rate for the intrinsic material and showed how the emission spectrum changes according to the injection level. Furthermore, we performed a comparison between the radiative and Auger lifetimes given two different doping concentrations for the E-SWIR and MWIR compositions.

As opposed to previous analytical work on Auger recombinations, [42–45] in this work we have applied a Green’s function based model to the full electronic band structure of the material, considering the two topmost valence and the two lowest conduction states as possible final states for the scattered particle. Therefore, the relevant contributions that previous analytical works might have neglected, due to applied approximations, have been taken into account in this study.

From this analysis, we concluded that in the E-SWIR case the radiative processes are the dominant recombination mechanism given an injection level lower than 10¹⁷ cm⁻³ while in the MWIR case the Auger processes are the dominant recombination mechanism above the injection level of 10¹⁷ cm⁻³.

Since the present study was aimed at evaluating the potential use of these alloys in detector applications, we have considered only unstrained material. In fact, absorber layers several μm thick may be required for high quantum efficiency photodiodes and these are likely to be relaxed. Nevertheless, for other kinds of applications, such as light emitters, it is likely that strain will play an important role and further investigation is needed.

Funding

Army Research laboratory (ARL) (W911NF-12-2-0023); Army Research Office (ARO) (W911NF-14-1- 0432).

References and links

1. D. Stange, S. Wirths, N. Von DenDriesch, G. Mussler, T. Stoica, Z. Ikonic, J. M. Hartmann, S. Mantl, D. Grützmacher, and D. Buca, “Optical transitions in direct-bandgap Ge₁₋_x Sn_x alloys,” ACS Photon. 2, 1539–1545 (2015). [CrossRef]

2. M. Bauer, J. Taraci, J. Tolle, A. Chizmeshya, S. Zollner, D. J. Smith, J. Menendez, C. Hu, and J. Kouvetakis, “Ge–Sn semiconductors for band-gap and lattice engineering,” Appl. Phys. Lett. 81, 2992–2994 (2002). [CrossRef]

3. M. R. Bauer, J. Tolle, C. Bungay, A. V. Chizmeshya, D. J. Smith, J. Menéndez, and J. Kouvetakis, “Tunable band structure in diamond–cubic tin–germanium alloys grown on silicon substrates,” Solid State Commun. 127, 355–359 (2003). [CrossRef]

4. S. Wirths, R. Geiger, N. von den Driesch, G. Mussler, T. Stoica, S. Mantl, Z. Ikonic, M. Luysberg, S. Chiussi, J. Hartmann, H. Sigg, J. Faist, D. Buca, and D. Grützmacher, “Lasing in direct-bandgap GeSn alloy grown on Si,” Nature Photon. 9, 88–92 (2015). [CrossRef]

5. C. Goodman, “Device applications of direct gap group IV semiconductors,” Japan. J. Appl. Phys. 22, 583 (1983). [CrossRef]

6. W. Du, Y. Zhou, S. A. Ghetmiri, A. Mosleh, B. R. Conley, A. Nazzal, R. A. Soref, G. Sun, J. Tolle, J. Margetis, H. A. Naseem, and S. Q. Yu, “Room-temperature electroluminescence from Ge/Ge₁₋_x Sn_x/Ge diodes on Si substrates,” Appl. Phys. Lett. 104, 241110 (2014). [CrossRef]

7. J. P. Gupta, N. Bhargava, S. Kim, T. Adam, and J. Kolodzey, “Infrared electroluminescence from GeSn heterojunction diodes grown by molecular beam epitaxy,” Appl. Phys. Lett. 102, 251117 (2013). [CrossRef]

8. G. He and H. A. Atwater, “Interband transitions in Sn_x Ge₁₋_x alloys,” Phys. Rev. Lett. 79, 1937 (1997). [CrossRef]

9. A. A. Tonkikh, C. Eisenschmidt, V. G. Talalaev, N. D. Zakharov, J. Schilling, G. Schmidt, and P. Werner, “Pseudomorphic GeSn/Ge (001) quantum wells: Examining indirect band gap bowing,” Appl. Phys. Lett. 103, 032106 (2013). [CrossRef]

10. A. Chizmeshya, C. Ritter, J. Tolle, C. Cook, J. Menendez, and J. Kouvetakis, “Fundamental studies of P (GeH₃)₃, As (GeH₃)₃, and Sb (GeH₃) 3: Practical n–dopants for new group IV semiconductors,” Chem. Mater. 18, 6266–6277 (2006). [CrossRef]

11. J. Xie, J. Tolle, V. D’Costa, C. Weng, A. Chizmeshya, J. Menendez, and J. Kouvetakis, “Molecular approaches to p-and n-nanoscale doping of Ge₁₋_y Sn_y semiconductors: Structural, electrical and transport properties,” Semiconductor Sci. Tech. 53, 816–823 (2009).

12. H. Tran, W. Du, S. A. Ghetmiri, A. Mosleh, G. Sun, R. A. Soref, J. Margetis, J. Tolle, B. Li, H. A. Naseem, and S. Q. Yu, “Systematic study of Ge₁₋_x Sn_x absorption coefficient and refractive index for the device applications of Si-based optoelectronics,” J. Appl. Phys. 119, 103106 (2016). [CrossRef]

13. C. Senaratne, P. Wallace, J. Gallagher, P. Sims, J. Kouvetakis, and J. Menendez, “Direct gap Ge₁₋_y Sn_y alloys: Fabrication and design of mid–IR photodiodes,” J. Appl. Phys. 120, 025701 (2016). [CrossRef]

14. Y. Zhou, W. Dou, W. Du, T. Pham, S. A. Ghetmiri, S. Al-Kabi, A. Mosleh, M. Alher, J. Margetis, J. Tolle, G. Sun, R. Soref, B. Li, M. Mortazavi, H. Naseem, and S. Q. Yu, “Systematic study of GeSn heterostructure-based light-emitting diodes towards mid–infrared applications,” J. Appl. Phys. 120, 023102 (2016). [CrossRef]

15. W. Du, Y. Zhou, S. A. Ghetmiri, A. Mosleh, B. R. Conley, A. Nazzal, R. A. Soref, G. Sun, J. Tolle, J. Margetis, H. A. Naseem, and S. Q. Yu, “Room–temperature electroluminescence from Ge/Ge₁₋_x Sn_x/Ge diodes on Si substrates,” Appl. Phys. Lett. 104, 241110 (2014). [CrossRef]

16. S. Wirths, D. Buca, and S. Mantl, “Si–Ge–Sn alloys: From growth to applications,” Progress in Crystal Growth and Characterization of Materials 62, 1–39 (2016). [CrossRef]

17. P. Moontragoon, Z. Ikonić, and P. Harrison, “Band structure calculations of Si–Ge–Sn alloys: achieving direct band gap materials,” Semiconductor Sci. Tech. 22, 742 (2007). [CrossRef]

18. N. Bouarissa and F. Annane, “Electronic properties and elastic constants of the ordered Ge₁₋_x Sn_x alloys,” Mater. Sci. Eng. B 95, 100–106 (2002). [CrossRef]

19. J. D. Sau and M. L. Cohen, “Possibility of increased mobility in Ge–Sn alloy system,” Phys. Rev. B 75, 045208 (2007). [CrossRef]

20. S. Gupta, B. Magyari-Köpe, Y. Nishi, and K. C. Saraswat, “Band structure and ballistic electron transport simulations in GeSn alloys,” Proc. Simul. Semicond. Processes Devices pp. 3–6 (2012).

21. F. Freitas, J. Furthmüller, F. Bechstedt, M. Marques, and L. Teles, “Influence of the composition fluctuations and decomposition on the tunable direct gap and oscillator strength of Ge₁₋_x Sn_x alloys,” Appl. Phys. Lett. 108, 092101 (2016). [CrossRef]

22. V. R. D’Costa, C. S. Cook, A. Birdwell, C. L. Littler, M. Canonico, S. Zollner, J. Kouvetakis, and J. Menéndez, “Optical critical points of thin-film Ge₁₋_y Sn_y alloys: A comparative Ge₁₋_y Sn_y/Ge₁₋_x Sn_x study,” Phys. Rev. B 73, 125207 (2006). [CrossRef]

23. T. Tsukamoto, N. Hirose, A. Kasamatsu, T. Mimura, T. Matsui, and Y. Suda, “Formation of GeSn layers on Si (001) substrates at high growth temperature and high deposition rate by sputter epitaxy method,” J. Mater. Sci. 50, 4366–4370 (2015). [CrossRef]

24. J. Mathews, R. Beeler, J. Tolle, C. Xu, R. Roucka, J. Kouvetakis, and J. Menéndez, “Direct-gap photoluminescence with tunable emission wavelength in Ge₁₋_y Sn_y alloys on silicon,” Appl. Phys. Lett. 97, 221912 (2010). [CrossRef]

25. S. A. Ghetmiri, W. Du, J. Margetis, A. Mosleh, L. Cousar, B. R. Conley, L. Domulevicz, A. Nazzal, G. Sun, R. A. Soref, J. Tolle, B. Li, H. A. Naseem, and S. Q. Yu, “Direct-bandgap GeSn grown on silicon with 2230 nm photoluminescence,” Appl. Phys. Lett. 105, 151109 (2014). [CrossRef]

26. H. Lin, R. Chen, W. Lu, Y. Huo, T. I. Kamins, and J. S. Harris, “Investigation of the direct band gaps in Ge₁₋_x Sn_x alloys with strain control by photoreflectance spectroscopy,” Appl. Phys. Lett. 100, 102109 (2012). [CrossRef]

27. M. L. Cohen, J. R. Chelikowsky, and J. R. Meyer-Arendt, “Electronic structure and optical properties of semiconductors,” Applied Optics 28, 2388 (1989).

28. H. Wen and E. Bellotti, “Rigorous theory of the radiative and gain characteristics of silicon and germanium lasing media,” Phys. Rev. B 91, 035307 (2015). [CrossRef]

29. H. Wen, B. Pinkie, and E. Bellotti, “Direct and phonon-assisted indirect Auger and radiative recombination lifetime in HgCdTe, InAsSb, and InGaAs computed using Green’s function formalism,” J. Appl. Phys. 118, 015702 (2015). [CrossRef]

30. M. Otfried, Semiconductors: data handbook (Springer, 2004).

31. P. Landsberg and A. Beattie, “Auger effect in semiconductors,” J. Phys. Chem. Solids 8, 73–75 (1959). [CrossRef]

32. M. Takeshima, “Unified theory of the impurity and phonon scattering effects on Auger recombination in semiconductors,” Phys. Rev. B 25, 5390 (1982). [CrossRef]

33. M. Takeshima, “Green’s-function formalism of band-to-band Auger recombination in semiconductors. correlation effect,” Phys. Rev. B 26, 917 (1982). [CrossRef]

34. W. Bardyszewski and D. Yevick, “Compositional dependence of the Auger coefficient for InGaAsP lattice matched to InP,” J. Appl. Phys. 58, 2713–2723 (1985). [CrossRef]

35. F. Bertazzi, M. Goano, and E. Bellotti, “A numerical study of Auger recombination in bulk InGaN,” Appl. Phys. Lett. 97, 231118 (2010). [CrossRef]

36. F. Bertazzi, M. Goano, and E. Bellotti, “Numerical analysis of indirect Auger transitions in InGaN,” Appl. Phys. Lett. 101, 011111 (2012). [CrossRef]

37. F. Bertazzi, X. Zhou, M. Goano, G. Ghione, and E. Bellotti, “Auger recombination in InGaN/GaN quantum wells. A full-Brillouin-zone study,” Appl. Phys. Lett. 103, 081106 (2013). [CrossRef]

38. J. P. Walter and M. L. Cohen, “Frequency and wave-vector dependent dielectric function for silicon,” Phys. Rev. B 5, 3101 (1972). [CrossRef]

39. C. Jungemann and B. Meinerzhagen, Hierarchical device simulation: the Monte-Carlo perspective (Springer Science & Business Media, 2012).

40. H. Wen and E. Bellotti, “Optical absorption and intrinsic recombination in relaxed and strained InAs₁₋_x Sb_x alloys for mid-wavelength infrared application,” Applied Physics Letters 107, 222103 (2015). [CrossRef]

41. M. V. Fischetti and S. E. Laux, “Band structure, deformation potentials, and carrier mobility in strained Si, Ge, and SiGe alloys,” J. Appl. Phys. 80, 2234–2252 (1996). [CrossRef]

42. L. Huldt, “Auger recombination in germanium,” Phys. Status Solidi A 24, 221–229 (1974). [CrossRef]

43. L. Huldt, “Phonon-assisted Auger recombination in germanium,” Phys. Status Solidi A 33, 607–614 (1976). [CrossRef]

44. A. Haug, “Phonon-assisted Auger recombination in degenerate semiconductors,” Solid State Commun. 22, 537–539 (1977). [CrossRef]

45. A. Haug, “Auger recombination of electron-hole drops,” Solid State Commun. 25, 477–479 (1978). [CrossRef]

46. C. Senaratne, P. Wallace, J. Gallagher, P. Sims, J. Kouvetakis, and J. Menendez, “Direct gap Ge₁₋_y Sn_y alloys: Fabrication and design of mid-IR photodiodes,” J. Appl. Phys. 120, 025701 (2016). [CrossRef]

47. S. Dominici, H. Wen, F. Bertazzi, M. Goano, and E. Bellotti, “Numerical evaluation of Auger recombination coefficients in relaxed and strained germanium,” Appl. Phys. Lett. 108, 211103 (2016). [CrossRef]

Parameter	Unit	Ge_0.91Sn_0.09	Ge_0.82Sn_0.18
c_l	GPa	117.950	114.100
c_t	GPa	64.769	65.522
ε₀	–	16.880	17.440
ε_∞	–	16.880	17.440
ħω_op	meV	35.930	35.060
Ξ_d	eV	9.1	9.1
D(L₆_c)	10⁸ eV/cm	12.17	12.17
D(L_v)	10⁸ eV/cm	10	10

$m_{direction}^{valley}$	HH	LH	SO	CB
$m_{(111)}^{(0), Γ}$	0.7361	0.0533	0.1318	0.0487
$m_{(100)}^{(0), Γ}$	0.2845	0.0600	0.1323	0.0486
$m_{(110)}^{(0), Γ}$	0.5451	0.0545	0.1322	0.0486
$m_{(111)}^{(0), Γ}$	–	–	–	1.4469

$m_{direction}^{valley}$	HH	LH	SO	CB
$m_{(111)}^{(0), Γ}$	−0.1294	−0.1855	0.4387	−0.1532
$m_{(100)}^{(0), Γ}$	−0.4434	−0.2149	0.4364	−0.1533
$m_{(110)}^{(0), Γ}$	−0.5269	−0.1922	0.4369	−0.1546
$m_{(111)}^{(0), Γ}$	–	–	–	0.3572

Parameter	Unit	Ge_0.91Sn_0.09	Ge_0.82Sn_0.18
c_l	GPa	117.950	114.100
c_t	GPa	64.769	65.522
ε₀	–	16.880	17.440
ε_∞	–	16.880	17.440
ħω_op	meV	35.930	35.060
Ξ_d	eV	9.1	9.1
D(L₆_c)	10⁸ eV/cm	12.17	12.17
D(L_v)	10⁸ eV/cm	10	10

$m_{direction}^{valley}$	HH	LH	SO	CB
$m_{(111)}^{(0), Γ}$	0.7361	0.0533	0.1318	0.0487
$m_{(100)}^{(0), Γ}$	0.2845	0.0600	0.1323	0.0486
$m_{(110)}^{(0), Γ}$	0.5451	0.0545	0.1322	0.0486
$m_{(111)}^{(0), Γ}$	–	–	–	1.4469

Numerical study on the optical and carrier recombination processes in GeSn alloy for E-SWIR and MWIR optoelectronic applications

Abstract

1. Introduction

2. Numerical model

2.1. Auger recombination rate and lifetime

2.2. Absorption coefficient and radiative recombination rate and lifetime

2.3. Spectral density function

3. Numerical results

3.1. Optical emission and absorption properties

3.2. Auger and radiative recombination

4. Conclusions

Funding

References and links

Cited By

Figures (17)

Tables (3)

Equations (15)

Optics Express