Atomistic quantum-dot physics
How do atomic structure, symmetry and interactions determine the light a quantum dot emits? My work links that question to detailed models, experiments and computational tools.
Tight-binding electronic structure · Configuration interaction · Parallel computing · Machine learning
Changing the character of the confined hole
Increasing a nanowire dot’s height does more than lower its confinement energy. In my 2013 atomistic study of InAs/InP dots, sufficiently tall structures developed a light-hole excitonic ground state instead of the more usual heavy-hole character.
The optical signature is a different fine structure: a dark state and three bright states, including a higher-energy transition polarized along the wire. Dot geometry therefore controls not just the emission energy, but also the polarization and ordering of the exciton levels.
Coupling as a way to control exciton fine structure
With Michał Świderski, I studied how crystal orientation, structural asymmetry and coupling shape the exciton spectrum of two InAs/InP dots in a nanowire. The atomistic symmetry can differ from what a cylindrical outline suggests. Alloy randomness introduces another source of symmetry breaking.
We then showed that an electric field can reduce, cancel or reverse the bright-exciton splitting. Strong coupling can avoid the severe loss of optical activity found in weakly coupled regimes. In selected alloy configurations, the splitting falls below 1 µeV while pronounced optical activity is recovered. These are theoretical conditions relevant to entangled-photon sources, not a demonstration of entanglement by themselves.
One material, two crystal phases, unexpected states
A crystal-phase dot is defined by alternating zinc-blende and wurtzite segments rather than by changing chemical composition. In collaborative experimental and theoretical work on InP nanowires, we connected the measured spectra to hybridized hole states and demonstrated single-photon emission and a photon-pair cascade.
Later, with Martyna Patera, I showed that a crystal-phase dot can have an antibonding hole ground state. Even weak strain from the phase mismatch can substantially alter its lowest hole levels. A companion study established the importance of crystal-field splitting and spontaneous polarization, treated consistently with the electron–hole interaction.
Why a nominally dark exciton can emit light
Dark-exciton optical activity is not a single mechanism. With Y. Don and D. Gershoni, I modelled how reduced shape symmetry activates dark transitions in self-assembled dots. Related experimental work with the wider collaboration demonstrated optical initialization and coherent control of the dark-exciton spin using short pulses.
My 2021 atomistic study showed that alloy randomness can activate dark excitons even in cylindrical InGaAs dots, without additional faceting or elongation. Two contributions matter: exchange-induced mixing with bright configurations, and changes to the optical transition matrix elements themselves. Their interplay makes intensity and polarization sensitive to the particular atomic arrangement.
Changing optical properties after a dot has been grown
Work with Garnett Bryant and collaborators showed how mechanical strain can reshape confined states and modify exciton exchange, including its phase, with consequences for optical polarization. Strain is therefore more than a correction to the band gap: it is a control parameter for the quantum states.
In a subsequent experiment–theory collaboration, a straining SiO2 envelope around an InP nanowire tuned the emission of an embedded InAsP dot over more than 200 meV. Atomistic calculations connected the observed shifts to strain. This provides a route to post-growth spectral tuning without changing the dot’s original composition.
Making atomistic many-body calculations feasible
Resolving every atom is only useful if the calculation can be completed. With Piotr Różański, I developed methods combining atomistic tight-binding states with configuration interaction for nanostructures containing more than ten million atoms.
A major bottleneck is evaluating Coulomb and exchange matrix elements. Our FFT-based methods replace costly direct summation; Coulombo-Lattice uses the underlying crystal lattice without transferring wave functions to an auxiliary basis. For the Coulomb calculation, it achieves O(N log N) scaling and agreement with direct summation to numerical precision within the same model. Both code releases are linked below.
- Linear scaling approach for atomistic calculation of excitonic properties of 10-million-atom nanostructures
- Efficient computation of Coulomb and exchange integrals for multi-million atom nanostructures
- Exploiting underlying crystal lattice for efficient computation of Coulomb matrix elements in multi-million atoms nanostructures
Learning the energies without repeating every calculation
My study of double InAs/InP nanowire dots used neural networks and transfer learning to connect geometry with electron and hole ground-state energies. Training on a fraction of the structural search space yielded errors of about 1 meV, measured as root-mean-square deviations within the studied domain.
The useful role of the surrogate is to explore many structures before committing to expensive atomistic calculations. The demonstrated accuracy concerns single-particle energies: it does not establish microelectronvolt accuracy for exciton fine structure, or reliable extrapolation to arbitrary new structures.
Composition, surfaces and strain are part of the design
With Agnieszka Gajewicz-Skretna, I examined ZnCuInS/ZnS nanocrystals using atomistic and excitonic calculations. Composition changes carrier localization and electron–hole overlap; a band-gap value alone does not fully describe the confinement or optical response.
For ZnSe/ZnS, we found a strong role for surface termination and stoichiometry. In the studied large-core regime, adding a shell can produce a strain-driven blue shift, unlike the red shift found for smaller cores. These results inform cadmium-free materials design; the absence of cadmium alone is not evidence that a nanocrystal is non-toxic.