Two-dimensional architectures for donor-based quantum computing

Through the introduction of a new electron spin transport mechanism, a 2D donor electron spin quantum computer architecture is proposed. This design addresses major technical issues in the original Kane design, including spatial oscillations in the exchange coupling strength and cross-talk in gate control. It is also expected that the introduction of nonlocality in qubit interaction will significantly improve the scaling fault-tolerant threshold over the nearest-neighbor linear array.

Figure 1

(Color online) Top view of a fragment of a quasi-two-dimensional donor electron spin quantum computer architecture for the case of Si:P, incorporating coherent transport by adiabatic passage (CTAP).

Figure 2

(Color online) Top: Schematic of the one-electron triple donor system $3D^{2+}$ based on P donors in silicon. Two of the donors are assumed ionized, the other neutral. Bottom: multidonor CTAPn straddling schemes.

Figure 3

(Color online) Numerical simulation of the CTAP pulse scheme applied to a spin superposition at donor 1 at $t=0$, demonstrating coherent transfer to the third donor at $t=t_{\max }$ (phases relative to an untransported state).

Figure 4

(Color online) Response of the $\mathrm{P}\text{−}{\mathrm{P}}^{+}$ interdonor potential profile to the barrier gate bias $V_b=(0,\pm 500)\mathrm{mV}$.

Figure 5

(Color online) Computed energy gap for the $\mathrm{P}\text{−}{\mathrm{P}}^{+}$ system as a function of $B$-gate bias $V_b$ for $R=30\mathrm{nm}$ and basis sizes $N=55$ and 140 (device parameters are: $30\mathrm{nm}$ donor depth below the oxide interface, $5\mathrm{nm}$ oxide thickness, and $10\mathrm{nm}$ gate width).

Figure 6

Transfer error (contours) as a function of charge dephasing rate, $\Gamma$ and total transfer time for CTAP5 (solid line), CTAP7 (long dashes), and CTAP9 (short dashes) for the case of $30\mathrm{nm}$ end-donor spacings, and $20\mathrm{nm}$ between the central donors (${\Omega }_{\text{end}}^{\max }=87\mathrm{GHz}$ and ${\Omega }_S=1\mathrm{THz}$).

Figure 7

(Color online) Distribution of nonuniform interdonor tunneling rates arising from fabrication variations in donor placement, where the ${\Omega }_{S_i}$ are chosen randomly from a normal distribution with mean ${\Omega }_S$ and standard deviation $\eta {\Omega }_S$.

Figure 8

Error rates due to misalignment of the central donors as a function of the distribution parameter $\eta$ which governs the standard deviation around the target straddling donor tunneling rate ${\Omega }_S=1\mathrm{THz}$ $(t_{\max }=2\mathrm{ns})$. The line is the median error rate, over the ensemble for each $\eta$.

Figure 9

(Color online) The figure shows (counterclockwise from the bottom left) the CTAP rails connecting the interaction regions with the rest of the computer; the incorporation of interaction zones, storage, and SET readout; a unit cell including space reserved for classical driving circuitry on a chip; and a quasi-two-dimensional tiled arrangement which can be extended in the plane.