Currently, I am reading the original paper about spin field effect transistor proposed by Supriyo Datta and Biswajit Das. In the last part of this paper, to obtain a larger overall current modulation the authors introduced the following Hamiltonian:

\begin{equation} H= \begin{bmatrix} -\dfrac{\hbar^2}{2m^*} \left( \dfrac{\partial^2}{\partial x^2} + \dfrac{\partial^2}{\partial z^2} \right) + V(z) -i\eta \dfrac{\partial}{\partial x} & i\eta \dfrac{\partial}{\partial z} \\ -i\eta \dfrac{\partial}{\partial z} & -\dfrac{\hbar^2}{2m^*} \left( \dfrac{\partial^2}{\partial x^2} + \dfrac{\partial^2}{\partial z^2} \right) + V(z) + i\eta \dfrac{\partial}{\partial x} \end{bmatrix} \end{equation} in which $\eta$ is the spin-orbit coupling constant and $V(z)$ is a confining potential that confines the electrons in a waveguide.

To find the eigenstates of $H$ one can use the unperturbed ($\eta=0$) eigenstates as a basis set. These eigenstates can be labeled with three indices: the subband index $m$, the wave vector $k$ in the $x$ direction, and the spin. For unperturbed Hamiltonian $H_0$:

\begin{align} & H_0|m,k\rangle = E_{m,k}|m,k\rangle \\ & E_{m,k}=\epsilon_m+\dfrac{\hbar^2 k^2}{2m^*} \end{align} the two spins $[1,0]^T$ and $[0,1]^T$ are degenerate. The subband energy $ \epsilon_m $ is obtained by solving the eigenequation

\begin{equation} \left( -\dfrac{\hbar^2}{2m^*}\dfrac{d^2}{dz^2} + V(z) \right)\phi_m(z)=\epsilon_m \phi_m(z) \end{equation}

The Rashba term $H_R$ ($\eta$ related terms in $H$ matrix) leads to matrix elements coupling the eigenstates of $H_0$ as follows: ($ + \leftrightarrow [1,0]^T ;- \leftrightarrow [0,1]^T $):

\begin{align} \langle m',k',+|H_R|m,k,+ \rangle & = + \eta k \delta_{m',m} \delta_{k',k} \\ \langle m',k',-|H_R|m,k,- \rangle & = - \eta k \delta_{m',m} \delta_{k',k} \\ \langle m',k',+|H_R|m,k,- \rangle & = \dfrac{\eta}{\hbar} \langle m'|p_z|m\rangle \delta_{k',k} \\ \end{align}

My question is how to derive these matrix elements? Or what's the mathematical form of the state $|m,k,+\rangle$ ?


You can't know what's exactly the mathematical form of $|m,k,+\rangle$ without knowing what $V(z)$ is. Nevertheless, you can still obtain the matrix elements without knowing the details of it. You should have realized that $|m\rangle = \phi_m(z) $ and need to know that $ |m,k,+\rangle = |m\rangle \otimes |k\rangle \otimes |+\rangle$

Also, guess what? $|k\rangle$ are just plane-waves because no potential in the x-direction: $|k\rangle = A e^{ikx}$.

That said now you can determine the first element: \begin{align} \langle m',k',+|H_R|m,k,+ \rangle & = \left[ \langle m',k'|, 0 \right] \begin{bmatrix} -i\eta \dfrac{\partial}{\partial x} & i\eta \dfrac{\partial}{\partial z} \\ -i\eta \dfrac{\partial}{\partial z} & i\eta \dfrac{\partial}{\partial x} \end{bmatrix} \left[ |m,k\rangle, 0 \right]^T \\ & = \langle m',k'| \left(-i\eta \dfrac{\partial}{\partial x} \right) |m,k\rangle \\ & = -i\eta\ \langle m' | m\rangle \ \langle k'|\dfrac{\partial}{\partial x}|k\rangle\\ & = -i\eta\ \delta_{m',m}\ (ik) \ \langle k'|k\rangle\\ & = \eta\ k\ \delta_{m',m} \ \delta_{k,k'}\\ \end{align}

Please note that $|m\rangle$ are not eigenstates of $\dfrac{\partial}{\partial x}$.

I trust you can now derive the results for other elements! Good luck!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.