The applied method is similar to the sinusoidal steady state analysis (S3A) used for semiconductor device simulation (Kurata
1971, Temple and Shewchun
1973, Laux
1985). S3A method involves the use of complex sinusoidal perturbation of the generation rate with infinitesimal amplitudes to excite the device in the frequency domain. In Langevin-like method these sinusoidal excitations of device parameters (electrical potential Ψ, quasi Fermi energy for electrons and holes Φ
n and Φ
p and temperature T) are caused by perturbation amplitudes connected with spectral density of random source terms (noise sources). Knowing the spectral density of noise sources, we can determine the modules of these complex amplitudes. The equations obtained in this way are called the transport equations for fluctuations (TEFF) (Jóźwikowski
2001). These equations were modified and developed in subsequent works (Jóźwikowski et al.
2004), eventually taking the form of Eq. (
4) in (Jóźwikowski et al.
2016). Let the set of transport Eqs. (
1)–(
4) in (Jóźwikowski et al.
2016) at the node k in the computation mesh be symbolically given [similarly as in (Laux
1985)] by:
$$\begin{aligned} F_{\Psi }^{k} \left( {{\text{N}}_{\text{D}}^{ + } ,{\text{N}}_{\text{A}}^{ - } ,{\text{n}},{\text{p}}} \right) = 0 \hfill \\ \frac{\partial n}{\partial t} - F_{n}^{k} \left( {{\text{n}},{\text{p}},\upmu_{\text{e}} ,({\text{G}} - {\text{R}})} \right) = 0 \hfill \\ \frac{\partial p}{\partial t} - F_{p}^{k} \left( {{\text{n}},{\text{p}},\upmu_{\text{h}} ,({\text{G}} - {\text{R}})} \right) = 0 \hfill \\ \frac{\partial T}{\partial t} - F_{T}^{k} \left( {{\text{n}},{\text{p}},\upmu_{\text{e}} ,\upmu_{\text{h}} } \right) = 0 \hfill \\ \end{aligned}$$
(1)
In the first step the set (
1) is solved to obtain steady-state values by using the Newton iterative method. The system is obtained by expressing the intensive parameters in the form
\(\varPsi = \varPsi^{0} +\updelta\varPsi\),
\(\varPhi_{\text{n}} = \varPhi_{\text{n}}^{0} +\updelta\varPhi_{\text{n}} ,\varPhi_{\text{p}} = \varPhi_{\text{p}}^{0} +\updelta\varPhi_{\text{p}}\) and
\(T = T^{0} + \delta T\). 0 superscript denotes the steady-state solution for the device, and δ refers to small time dependent terms. All physical quantities which depend on intensive parameters may be properly expanded: for example the electron mobility may be expressed as
$$\mu_{e} = \mu_{e}^{0} + \delta \mu_{e} = \mu_{e}^{0} + \left( {\frac{{\partial \mu_{e} }}{{\partial {\text{n}}}}\frac{\partial n}{\partial \Psi } + \frac{{\partial \mu_{e} }}{{\partial {\text{p}}}}\frac{\partial p}{\partial \Psi }} \right)\delta \Psi + \frac{{\partial \mu_{e} }}{{\partial {\text{n}}}}\frac{\partial n}{{\partial \Phi_{\text{n}} }}\delta \Phi_{\text{n}} + \frac{{\partial \mu_{e} }}{{\partial {\text{p}}}}\frac{\partial p}{{\partial \Phi_{\text{p}} }}\delta \Phi_{\text{p}} + \left( {\frac{{\partial \mu_{e} }}{{\partial {\text{n}}}}\frac{\partial n}{{\partial {\text{T}}}} + \frac{{\partial \mu_{e} }}{{\partial {\text{p}}}}\frac{\partial p}{{\partial {\text{T}}}}} \right)\delta \text{T}$$
(2)
To analyze the noise problem we treat them as fluctuations and introduce additional noise sources.
In this way we can obtain Eq. (
4), which is the matrix–vector notation of Eq. (
4) in (Jóźwikowski et al.
2016) at each analyzed node k.
$$\begin{aligned} \left[ {\begin{array}{cccc} {\frac{{\partial F_{\Psi }^{k} }}{{\partial \Psi }}} & {\frac{{\partial F_{\Psi }^{k} }}{{\partial \Phi _{{\text{n}}} }}} & {\frac{{\partial F_{\Psi }^{k} }}{{\partial \Phi _{{\text{p}}} }}} & {\frac{{\partial F_{\Psi }^{k} }}{{\partial {\text{T}}}}} \\ {\frac{{\partial n}}{{\partial \Psi }}\frac{\partial }{{\partial t}} - \frac{{\partial F_{{\text{n}}}^{k} }}{{\partial \Psi }}} & {\frac{{\partial n}}{{\partial \Phi _{{\text{n}}} }}\frac{\partial }{{\partial t}} - \frac{{\partial F_{{\text{n}}}^{k} }}{{\partial \Phi _{{\text{n}}} }}} & { - \frac{{\partial F_{{\text{n}}}^{k} }}{{\partial \Phi _{{\text{p}}} }}} & {\frac{{\partial n}}{{\partial {\text{T}}}}\frac{\partial }{{\partial t}} - \frac{{\partial F_{{\text{n}}}^{k} }}{{\partial {\text{T}}}}} \\ {\frac{{\partial p}}{{\partial \Psi }}\frac{\partial }{{\partial t}} - \frac{{\partial F_{{\text{p}}}^{k} }}{{\partial \Psi }}} & { - \frac{{\partial F_{{\text{p}}}^{k} }}{{\partial \Phi _{{\text{n}}} }}} & {\frac{{\partial p}}{{\partial \Phi _{{\text{p}}} }}\frac{\partial }{{\partial t}} - \frac{{\partial F_{{\text{p}}}^{k} }}{{\partial \Phi _{{\text{p}}} }}} & {\frac{{\partial n}}{{\partial {\text{T}}}}\frac{\partial }{{\partial t}} - \frac{{\partial F_{{\text{p}}}^{k} }}{{\partial {\text{T}}}}} \\ { - \frac{{\partial F_{{\text{T}}}^{k} }}{{\partial \Psi }}} & { - \frac{{\partial F_{{\text{T}}}^{k} }}{{\partial \Phi _{{\text{n}}} }}} & { - \frac{{\partial F_{{\text{T}}}^{k} }}{{\partial \Phi _{{\text{p}}} }}} & {\frac{\partial }{{\partial t}} - \frac{{\partial F_{{\text{T}}}^{k} }}{{\partial {\text{T}}}}} \\ \end{array} } \right]\left[ {\begin{array}{c} {\delta \,\Psi } \\ {\delta \,\Phi _{{\text{n}}} } \\ {\delta \,\Phi _{{\text{p}}} } \\ {\delta \,T} \\ \end{array} } \right] \hfill \\ = \left[ {\begin{array}{c} {\frac{{\partial F_{{p}}^{k} }}{{\partial {\mu }_{{\text{e}}} }}\frac{{\partial {\mu }_{{\text{e}}} }}{{\partial {\tau }_{{{\text{rel}}}}^{{\text{e}}} }}\delta {\tau }_{{{\text{rel}}}}^{{\text{e}}} + \frac{{\partial F_{{\text{n}}}^{k} }}{{\partial ({\text{G}} - {\text{R}})}}\left( {\delta ({\text{G}} - {\text{R}})_{{SHOT}} ({\text{G}} - {\text{R}})_{{1/f}} + F_{n} (t)} \right)} \\ {\frac{{\partial F_{{\text{o}}}^{k} }}{{\partial {\mu }_{{\text{h}}} }}\frac{{\partial {\mu }_{{\text{h}}} }}{{\partial {\tau }_{{{\text{rel}}}}^{{\text{h}}} }}\delta {\tau }_{{{\text{rel}}}}^{{\text{h}}} + \frac{{\partial F_{{\text{p}}}^{k} }}{{\partial ({\text{G}} - {\text{R}})}}\left( {\delta ({\text{G}} - {\text{R}})_{{SHOT}} ({\text{G}} - {\text{R}})_{{1/f}} + F_{p} (t)} \right)} \\ {{\mathbf{F}}_{{\mathbf{C}}} ({\mathbf{t}}) + {\mathbf{G}}_{{\mathbf{C}}} ({\mathbf{t}})} \\ \end{array} } \right] \hfill \\ \end{aligned}$$
(4)
The expressions (
1)–(
4) use standard conventions; the indexes n or e refer to electron, h or p refer to the holes, respectively,
\(\updelta\) denotes fluctuation,
\(\varvec{G}\) is the carrier generation rate,
R is the carrier recombination rate. G–R processes are influenced by Auger 1 and Auger 7 inter-band mechanisms and enhanced additionally by SHR processes caused by metal vacancies and dislocations (Kopytko and Jóźwikowski
2015). Other symbols are explained in (Jóźwikowski et al.
2016). As the set of Eq. (
4) is linear one can express all random variables by means of the Fourier series and separately consider all Fourier coefficients at any frequency. The random source terms, i.e.
\(\varvec{\delta}{\varvec{\uptau}}_{{{\mathbf{rel}}}}^{{\mathbf{e}}}\),
\(\varvec{\delta}{\varvec{\uptau}}_{{{\mathbf{rel}}}}^{{\mathbf{h}}}\),
\(\varvec{\delta}\left( {\varvec{G} - \varvec{R}} \right)_{{\varvec{SHOT}}} ,\varvec{ }\)
\(\varvec{\delta}\left( {\varvec{G} - \varvec{R}} \right)_{{1/\varvec{f}}}\),
\({\mathbf{F}}_{{\mathbf{C}}} \left( {\mathbf{t}} \right)\),
\({\mathbf{G}}_{{\mathbf{C}}} \left( {\mathbf{t}} \right)\),
\(\varvec{F}_{\varvec{n}} \left( \varvec{t} \right)\) and
\(\varvec{F}_{\varvec{p}} \left( \varvec{t} \right)\) are modeled and inserted into TEFF (set (4)) to determine the fluctuations of
\(\varPsi\),
\(\varPhi_{\text{n}}\),
\(\varPhi_{\text{n}}\) and T. Here
\(\varvec{\delta}{\varvec{\uptau}}_{{{\mathbf{rel}}}}^{{\mathbf{e}}}\) and
\(\varvec{\delta}{\varvec{\uptau}}_{{{\mathbf{rel}}}}^{{\mathbf{h}}}\) are the fluctuations of electron relaxation time and hole relaxation time, respectively. (Kousik et al.
1985), based on Handel’s theory of 1/f noise (Handel
1975,
1980) obtained theoretically spectral intensity of
\(\varvec{\delta}{\varvec{\uptau}}_{{{\mathbf{rel}}}}^{{\mathbf{e}}}\) for silicon. We have adopted their results for HgCdTe in some previous works (Jóźwikowski
2001,
2004,
2016). Handel’s theory of 1/f noise is based on the fact, that in accordance with the quantum electromagnetic field theory, electric charge carriers are accompanied by photons. Interactions leading to the change in carrier velocity are the sources of creation or annihilation of photons (Bremsstrahlung). They are called “soft photons” and are not energetic enough to be detected, however, the possibility of their absorption or emission must be taken into account in the calculation of the scattering amplitude. The number of these photons is inversely proportional to their energy, i.e. their frequency. This way both scattering processes determined by relaxation time and G–R processes determined by G–R terms are the potential sources of 1/f noise. In (Jóźwikowski et al.
2016) we have determined the Hooge coefficients (Hooge
1969) for 1/f noise caused by Auger1, Auger7, radiative and SHR G–R mechanisms in HgCdTe. The influence of dislocation on 1/f noise was determined as well by using our original model. In the presented paper, all those 1/f noise sources are included in TEFF in
\(\varvec{\delta}\left( {\varvec{G} - \varvec{R}} \right)_{{1/\varvec{f}}}\) terms.
\({\mathbf{F}}_{{\mathbf{C}}} \left( {\mathbf{t}} \right)\) denotes the fluctuation of heat stream and
\({\mathbf{G}}_{{\mathbf{C}}} \left( {\mathbf{t}} \right)\) denotes the fluctuation of heat generation rate (Jóźwikowski et al.
2004).
\(\varvec{F}_{\varvec{n}} \left( \varvec{t} \right)\) and
\(\varvec{F}_{\varvec{p}} \left( \varvec{t} \right)\) denote electron and hole diffusion noise, respectively (van der Ziel
1954,
1959). In the presented structures diffusion noise plays a marginal role and may be omitted. Knowing the SD of random sources (for example
\(S_{a} \left( f \right)\) of
\(\delta a\left( {\vec{r},t} \right)\)), one may determine the complex amplitude
\(a_{f}\) of their Fourier coefficients defined as follows:
$$\delta a\left( {\vec{r},t} \right) = \mathop \int \limits_{0}^{\infty } \frac{1}{\sqrt 2 }c_{f} \exp \left( {i\varphi_{f} } \right)\exp \left( {i2\pi ft} \right)df;\quad a_{f} = \frac{1}{\sqrt 2 }c_{f} \exp \left( {i\varphi_{f} } \right);S_{a} \left( f \right)\Delta f = a_{f} a_{f}^{*}$$
(5)
Here
\(S_{a} \left( f \right)\) denotes the SD of fluctuation quantity,
\(\Delta f = 1Hz\) and
\(c_{f} exp\left( {i\varphi_{f} } \right)\) is the complex Fourier coefficient for frequency
\(f\) (van der Ziel
1976). The numerical method for solving TEFF equations is described in (Jóźwikowski
2001,
2004,
2016). Knowing the SD of fluctuation quantity being the random noise source we are able to determine
\(c_{f}\) only, which is the modulus of the complex coefficient
\(c_{f} exp\left( {i\varphi_{f} } \right).\) As we cannot determine the angle
\(\varphi_{f}\) being some number between 0 and
\(2\pi\), we generate it randomly using a computer. This procedure is applied independently for all random noise terms contained in TEFF. We usually generate about 100 trials to obtain a corvergent spatial distribution of mean values of the SD of fluctuations of intense parameters for arbitrarily chosen frequency. Having them we can calculate the fluctuation of all physical quantities contained in TEFF. The solver of TEFF has been implemented in Fortran Intel Compiler and complex numbers. To solve the set of Eq. (
4) one has to know the SD of mobility fluctuations caused by fluctuations of relaxation times and the carrier concentration [this was shown in (Jóźwikowski et al.
2004)] as well as the SD of fluctuations of g–r processes generating both 1/f noise and the shot noise. Similar to (Jóźwikowski et al.
2016), the influence of dislocations on noise current was also taken into account in this paper. Current noise measured in electronic system connected with detector is caused by fluctuations of current density inside the detector. The idea to find it is based on the fact that the density of the Joule power,
\(\rho_{P}\) in heterostructure is the product of the current density and the gradients of quasi-Fermi energies, i.e.:
$$\mathop \int \limits_{V}^{{}} \left( { - \vec{j}_{n} \cdot \nabla \varPhi_{\text{n}} + \vec{j}_{p} \cdot \nabla \varPhi_{\text{p}} } \right)dV = \mathop \int \limits_{V}^{{}} \rho_{P} dV = UI$$
(6)
where
\(\vec{j}\) denotes the current density,
\(U\)-the bias voltage,
\(I\)-the total electric current,
\(V\)-the volume of the detector.
$$\vec{j}_{n} = - n\mu_{n} \nabla \varPhi_{\text{n}} ,\quad \vec{j}_{p} = p\mu_{p} \nabla \varPhi_{\text{p}}$$
(7)
The noise power density is here treated as a result of the fluctuations of the quasi-Fermi levels as well as fluctuations of the current density, i.e.
$$\delta \rho_{P} = - \vec{j}_{n} \cdot \delta \left( {\nabla \varPhi_{n} } \right) - \nabla \varPhi_{n} \cdot \delta \vec{j}_{n} + \vec{j}_{p} \cdot \delta \left( {\nabla \varPhi_{p} } \right) + \nabla \varPhi_{p} \delta \vec{j}_{p}$$
(8)
The fluctuations of the total noise power density may be expressed by
$$\delta P_{N} = \int\limits_{V} {\delta \rho_{P} dV = \delta UI + U\delta I \approx U\delta I}$$
(9)
The fluctuation of the density of Joule’s heat power
\(\delta \rho_{P}\) is now considered as a noise power density. Taking into account the energy balance of Eq. (
9), one may write an expression for the effective SD of the total noise current:
$$S_{I} = \frac{1}{{U^{2} }}S_{{P_{N} }} .$$
(10)