Photonics Essentials: From Diffusion Current to Equation 3.5 ============================================================ Source ------ Thomas P. Pearsall, *Photonics Essentials: An Introduction with Experiments* (McGraw-Hill, 2003), Chapter 3, ``Photodiodes``, printed pages 38--40. Question -------- The chapter first gives a diffusion-current equation containing one spatial derivative, .. math:: J(x)=-qD_e\frac{\partial n}{\partial x}. Why does the carrier equation later contain the second derivative :math:`\partial^2 n/\partial x^2`? .. important:: Current :math:`J` does **not** turn directly into concentration. Equation 3.5 asks whether more carriers enter a small region than leave it. That requires the spatial derivative of :math:`J`, introducing one more derivative. The small-slab picture ---------------------- .. figure:: ../../../_static/knowledge_base/worked_exercises/photonics_essentials/ch03_carrier_balance.svg :alt: Carrier balance in a small semiconductor slab :align: center :width: 94% Carrier accumulation equals inflow minus outflow, plus optical generation, minus recombination. The symbols used below are: .. list-table:: :header-rows: 1 :widths: 22 48 30 * - Symbol - Meaning - Units * - :math:`\Delta n_p(x,t)` - Excess minority-electron concentration in p-type material - :math:`\mathrm{cm^{-3}}` * - :math:`J(x)` - Diffusion current density, using the book's sign convention - :math:`\mathrm{A\,cm^{-2}}` * - :math:`D_e` - Electron diffusion coefficient - :math:`\mathrm{cm^2\,s^{-1}}` * - :math:`\tau_e` - Minority-electron recombination lifetime - :math:`\mathrm s` * - :math:`G_L` - Optical generation rate per unit volume - :math:`\mathrm{cm^{-3}\,s^{-1}}` Step 1: A concentration gradient produces diffusion ---------------------------------------------------- Carriers diffuse from high concentration toward low concentration. The book writes the resulting current density as .. math:: J(x)=-qD_e\frac{\partial\Delta n_p}{\partial x}. The chapter initially writes :math:`n(x)` instead of :math:`\Delta n_p(x)`. In the neutral p-region, .. math:: n_p(x)=n_{p0}+\Delta n_p(x). The equilibrium concentration :math:`n_{p0}` is spatially constant, so .. math:: \frac{\partial n_p}{\partial x} =\frac{\partial\Delta n_p}{\partial x}. The two notations therefore give the same diffusion term. Step 2: Apply carrier conservation to a thin slab -------------------------------------------------- Consider a slab extending from :math:`x` to :math:`x+dx`. * Current density entering the slab is :math:`J(x)`. * Current density leaving it is :math:`J(x+dx)`. Expand the outgoing current to first order: .. math:: J(x+dx) \approx J(x)+\frac{\partial J}{\partial x}dx. Yes: this is a Taylor series ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The general Taylor expansion of a smooth function about :math:`x` is .. math:: J(x+dx) = J(x) +\frac{\partial J}{\partial x}dx +\frac{1}{2}\frac{\partial^2J}{\partial x^2}(dx)^2 +\cdots. The book keeps terms only through first order in the very small slab width :math:`dx`: .. math:: J(x+dx) = J(x) +\frac{\partial J}{\partial x}dx +O\!\left((dx)^2\right). After subtracting the entering and outgoing currents and dividing by :math:`dx`, the omitted part is :math:`O(dx)`. It approaches zero as the slab is made infinitesimally thin. This is why the first-order expansion is sufficient for the local conservation equation. The net carrier inflow per unit volume and time is consequently .. math:: \begin{aligned} \text{net inflow} &=\frac{J(x)-J(x+dx)}{q\,dx}\\ &=-\frac{1}{q}\frac{\partial J}{\partial x}. \end{aligned} This is the crucial extra derivative. Step 3: Add generation and recombination ----------------------------------------- Three processes change the excess-carrier concentration: .. math:: \text{rate of change} = \text{net inflow} -\text{recombination} +\text{optical generation}. In symbols, .. math:: \frac{\partial\Delta n_p}{\partial t} = -\frac{1}{q}\frac{\partial J}{\partial x} -\frac{\Delta n_p}{\tau_e} +G_L. Step 4: Substitute the diffusion current ----------------------------------------- Insert :math:`J=-qD_e\,\partial\Delta n_p/\partial x`: .. math:: \begin{aligned} -\frac{1}{q}\frac{\partial J}{\partial x} &=-\frac{1}{q}\frac{\partial}{\partial x} \left( -qD_e\frac{\partial\Delta n_p}{\partial x} \right)\\ &=D_e\frac{\partial^2\Delta n_p}{\partial x^2}, \end{aligned} where :math:`D_e` is assumed constant. The complete result is .. math:: \frac{\partial\Delta n_p}{\partial t} = D_e\frac{\partial^2\Delta n_p}{\partial x^2} -\frac{\Delta n_p}{\tau_e} +G_L. \tag{3.5} Why it is second order ---------------------- The derivative chain is .. math:: \boxed{ \Delta n_p \xrightarrow{\ \partial/\partial x\ } J \xrightarrow{\ -\partial/\partial x\ } \text{carrier accumulation} }. Therefore: * Equation 3.5 is **second order in position** :math:`x`. * It is only **first order in time** :math:`t`. * The second spatial derivative measures the curvature of the concentration profile, not simply its slope. A useful physical reading is: * constant concentration gives no diffusion current; * a straight concentration profile gives constant current but no local accumulation; * a curved concentration profile makes current vary with position, so some regions gain or lose carriers. Unit check ---------- Every term in Equation 3.5 must have units of concentration change per time: .. math:: \left[ D_e\frac{\partial^2\Delta n_p}{\partial x^2} \right] = \frac{\mathrm{cm^2}}{\mathrm s} \frac{\mathrm{cm^{-3}}}{\mathrm{cm^2}} = \mathrm{cm^{-3}\,s^{-1}}. The recombination and generation terms have the same units: .. math:: \left[\frac{\Delta n_p}{\tau_e}\right] = [G_L] = \mathrm{cm^{-3}\,s^{-1}}. Steady-state Equation 3.6 -------------------------- At steady state the concentration profile no longer changes with time: .. math:: \frac{\partial\Delta n_p}{\partial t}=0. Equation 3.5 then becomes .. math:: \frac{d^2\Delta n_p}{dx^2} = \frac{\Delta n_p}{D_e\tau_e} -\frac{G_L}{D_e}. \tag{3.6} In the dark, :math:`G_L=0`. Define the electron diffusion length .. math:: L_e=\sqrt{D_e\tau_e}. The equation reduces to .. math:: \frac{d^2\Delta n_p}{dx^2} -\frac{\Delta n_p}{L_e^2}=0, whose decaying solution in a long neutral region is .. math:: \Delta n_p(x)=\Delta n_p(0)e^{-x/L_e}. Because the spatial equation is second order, two boundary conditions are needed. Typically the junction supplies :math:`\Delta n_p(0)` and the excess concentration approaches zero far from the junction. Source correction ----------------- .. warning:: The paragraph below Equation 3.5 says that **Equation 3.4** is a second-order differential equation. Equation 3.4 is algebraic. The sentence should refer to **Equation 3.5**. Continue interactively ---------------------- Use the :doc:`ch03_interactive_lab` to change the diffusion coefficient, carrier lifetime, generation rate, and the other Chapter 3 variables while the corresponding curves update.