1] University of Colorado, Boulder, Colorado 80309, USA [2] National Renewable Energy Laboratory, Golden 80401, Colorado, USA.
Nat Commun. 2013;4:2396. doi: 10.1038/ncomms3396.
The long spin coherence time and microelectronics compatibility of Si makes it an attractive material for realizing solid-state qubits. Unfortunately, the orbital (valley) degeneracy of the conduction band of bulk Si makes it difficult to isolate individual two-level spin-1/2 states, limiting their development. This degeneracy is lifted within Si quantum wells clad between Ge-Si alloy barrier layers, but the magnitude of the valley splittings achieved so far is small--of the order of 1 meV or less--degrading the fidelity of information stored within such a qubit. Here we combine an atomistic pseudopotential theory with a genetic search algorithm to optimize the structure of layered-Ge/Si-clad Si quantum wells to improve this splitting. We identify an optimal sequence of multiple Ge/Si barrier layers that more effectively isolates the electron ground state of a Si quantum well and increases the valley splitting by an order of magnitude, to ~9 meV.
硅的长自旋相干时间和微电子兼容性使其成为实现固态量子位的理想材料。不幸的是,体硅导带的轨道(谷)简并使得难以隔离单个的两能级自旋 1/2 态,限制了它们的发展。这种简并在由锗-硅合金势垒层夹在中间的硅量子阱中被消除,但迄今为止实现的谷分裂幅度很小--大约 1 毫电子伏特或更小--降低了这种量子位中存储的信息的保真度。在这里,我们将原子赝势理论与遗传搜索算法相结合,优化了层状-Ge/Si 壳层 Si 量子阱的结构,以提高这种分裂。我们确定了一个由多个 Ge/Si 势垒层组成的最佳序列,这些层更有效地隔离了硅量子阱的电子基态,并将谷分裂提高了一个数量级,达到约 9 毫电子伏特。