The interplay between light and matter is the basis of many fundamental processes and various applications1. Harnessing light–matter interactions in principle allows operation of solid state devices under new physical principles: for example, the a.c. optical Stark effect (OSE) has enabled coherent quantum control schemes of spins in semiconductors, with the potential for realizing quantum devices based on spin qubits2, 3, 4, 5. However, as the dimension of semiconductors is reduced, light–matter coupling is typically weakened, thus limiting applications at the nanoscale. Recent experiments have demonstrated significant enhancement of nanoscale light–matter interactions, albeit with the need for a high-finesse cavity6, 7, ultimately preventing device down-scaling and integration. Here we report that a sizable OSE can be achieved at substantial energy detuning in a cavity-free colloidal metal–semiconductor core–shell hetero-nanostructure, in which the metal surface plasmon is tuned to resonate spectrally with a semiconductor exciton transition. We further demonstrate that this resonantly enhanced OSE exhibits polarization dependence and provides a viable mechanism for coherent ultrafast spin manipulation within colloidal nanostructures. The plasmon–exciton resonant nature further enables tailoring of both OSE and spin manipulation by tuning plasmon resonance intensity and frequency. These results open a pathway for tailoring light–matter–spin interactions through plasmon–exciton resonant coupling in a judiciously engineered nanostructure, and offer a basis for future applications in quantum information processing at the nanoscale. More generally, integrated nanostructures with resonantly enhanced light–matter interactions should serve as a test bed for other emerging fields, including nano-biophotonics and nano-energy8, 9.
Figures at a glance
Figure 1: Au–CdSe as a hetero-core–shell nanostructure, integrating a surface plasmon of the Au core with an exciton of the monocrystalline CdSe shell. 
a, Typical large scale TEM image, showing uniform core–shell nanostructures. Scale bar, 50 nm. Every core–shell nanostructure shows a monocrystalline wurtzite lattice, as highlighted in b. b, Typical linear optical absorption spectra of Au–CdSe (pink) and CdS–CdSe (blue) at 2 K. For CdS–CdSe, the CdSe shell manifests well-defined quantum well optical features19. These two spectra are raw data without normalization to the density of the nanostructures in the thin film. Green curve is a theoretical calculation of the surface plasmon resonance of nanoparticles with a 4.3-nm Au core and a 2.8-nm CdSe shell (see Supplementary Information), showing energy resonance with an exciton of the CdSe quantum shell. Red arrow indicates a typical sub-resonant excitation energy with ΔE = 186 meV for observing OSE in Au–CdSe. Insets, high resolution TEM images of Au–CdSe (pink box) and CdS–CdSe (blue box), clearly revealing the monocrystalline semiconductor shell feature. Red and orange dashed curves are guides to the eye for core-boundary and shell-boundary, respectively. Scale bar, 5 nm. c, A model schematically showing transitions and resonant coupling (blue arrow) between core (continuum spectrum of plasmon) and shell (discrete interband exciton) in a hetero-core–shell configuration under laser excitation (red arrows).
Figure 2: Tunable plasmon–exciton resonantly enhanced OSE and its polarization dependence in Au–CdSe nanostructures. 
a, Time-resolved linear optical absorption spectra with sub-resonant excitation energy at 1.892 eV of 1.45 GW cm−2. The sub-resonant and probe laser beams are co-polarized. The absorption intensity is shown colour-coded. b, Polarization dependence of OSE measured by DTS. Red and blue circles are experimental DTS (at a probe energy of 2.077 eV) obtained with respectively co- and counter-polarized sub-resonant excitation at 1.892 eV of 1.12 GW cm−2. Red and blue curves are fits to the data (Supplementary Information). c, Temporal and spectral mapping of the DTS difference of co- and counter-polarized sub-resonant excitation and probe, and its dependence on the diameter of the Au core: left, 4.3 nm; middle, 3.3 nm; right, 2.4 nm. For all three samples, the thickness of the monocrystalline CdSe shell remains the same (2.8 nm). The intensity difference of the DTS spectra is shown colour-coded; the colour scale (left) applies to all three mappings. The ΔE for data shown left, middle and right are respectively 186, 172 and 106 meV. Insets, high resolution TEM images of samples from which the data were taken. Red and orange dashed curves are guides to the eye for the core-boundary and the shell-boundary, respectively. Scale bar, 5 nm.
Figure 3: Ultrafast coherent spin manipulation in Au–CdSe enacted by resonantly enhanced OSE. 
a, Schematic experimental set-up. Photograph of an optically clear Au–CdSe thin film sample (~1 cm × 1 cm) is shown. OPA, optical parametric amplifier laser. b–e, TRFR measurement (left) and Bloch sphere (right) showing evolution of spin (pink lines on Bloch sphere highlight spin trajectory). All TRFR data are taken at H0 = 3.5 T with pump energy at 2.293 eV of 0.31 GW cm−2 and probe energy at 2.087 eV of 0.04 GW cm−2. Sub-resonant energy of all tipping pulses is fixed at 1.892 eV. a.u., arbitrary units. b, Absence of tipping pulse. Red circles, experimental data; black curve, fit to data with theoretical expression of Sx(t). c, Tipping pulse applied at tTIP = 13.5 ps (arrowed). Tipping pulse intensity is 1.20 GW cm−2. Blue squares, experimental data; red dashed curve, data from b. d, Tipping pulse applied at tTIP = 9.3 ps (arrowed). Tipping pulse intensity is 0.56 GW cm−2. Blue squares, experimental data; red dashed curve, data from b. e, Three tipping pulses are applied sequentially (blue arrows). Tipping pulse intensities are 0.67, 0.80 and 0.64 GW cm−2. Blue squares, experimental data; red dashed curve, data from b.
Figure 4: Dependence of ultrafast spin manipulation on tipping pulse intensity and the size of the Au core. 
a, Left: TRFR curves at different tipping pulse intensities with H0 = 2.3 T. Tipping pulses are applied at fixed delay tTIP = 13.7 ps, and with intensities (GW cm−2) as follows: black curve, 0 (absence of tipping pulse); red circles, 0.44; green down-triangles, 0.67; blue up-triangles, 1.31. Right, Bloch sphere schematically highlights different tipping effects (represented by dashed arrows of different colour) resulting from different tipping pulse intensities; pink lines highlight spin trajectory. b, Dependence of tipping angle on tipping pulse intensity as well as on Au core size. Squares, 4.3-nm Au core; circles, 3.3-nm Au core. Dash-dot line represents π/2 angle. Error bars are defined as the standard deviation of the data set measured from different areas in the same sample and/or different samples. Larger error bars shown at very high tipping pulse intensity are related to the damage threshold of samples (Supplementary Information).
right
