Spin dynamics in electrochemically charged CdSe quantum dots

We use time-resolved Faraday rotation to measure coherent spin dynamics in colloidal CdSe quantum dots charged in an electrochemical cell at room temperature. Filling of the $1S_{\mathrm{e}}$ electron level is demonstrated by the bleaching of the $1S_{\mathrm{e}}-1S_{3∕2}$ absorption peak. One of the two Landé $g$ factors observed in uncharged quantum dots disappears upon filling of the $1S_{\mathrm{e}}$ electron state. The transverse spin-coherence time, which is over $1\mathrm{ns}$ and is limited by inhomogeneous dephasing, also appears to increase with charging voltage. The amplitude of the spin precession signal peaks near the half-filling potential. Its evolution at charging potentials without any observable bleaching of the $1S_{\mathrm{e}}-1S_{3∕2}$ transition suggests that the spin dynamics are influenced by low-energy surface states.

Figure 1

(a) TEM image of a close-packed CdSe nanocrystal film. (b) Absorbance spectra $\alpha \left(E\right)$ at $V_{\text{cell}}=-0.8\mathrm{V}$ (black), $-1.0\mathrm{V}$ (gray), and $-1.2\mathrm{V}$ (light gray) along with a schematic of relevant energy levels in charging experiments. $V_{\text{cell}}=0.0\mathrm{V}$ is identical to $V_{\text{cell}}=-0.8\mathrm{V}$. (c) Bleaching spectra relative to $V_{\text{cell}}=0.0\mathrm{V}$ as described in the text. Bleaching of the $1P_{\mathrm{e}}$ electron level can be seen at $2.35\mathrm{eV}$.

Figure 2

(a)–(d) FR time-domain scans for representative voltages. Voltages not plotted represent a smooth evolution between the displayed results. The fast Fourier transforms in (e)–(h), which have been smoothed for clarity, show two frequencies in uncharged nanocrystals evolving to a single precession frequency in the charged ensemble.

Figure 3

(a) Absorbance of the probe beam, measuring the local bleaching at the laser focus spot. (b) Amplitude of the nonoscillating component ${\theta }_{\exp }$, which vanishes on charging. (c) Fit results for the spin-coherence times $T_{2,1}^{*}$ (circles), $T_{2,2}^{*}$ (triangles), and the Gaussian dephasing time ${\Delta }_t$ (squares). For $T_{2,1}^{*}⪢{\Delta }_t$, the fitting algorithm cannot quantitatively distinguish between a purely Gaussian lifetime and the exponential lifetime $T_{2,1}^{*}$, suggesting that the spin relaxation is completely dominated by inhomogeneous dephasing for $\mid V_{\text{cell}}\mid >1.0\mathrm{V}$.

Figure 4

(Color) (a) FR amplitudes ${\theta }_1$ and ${\theta }_2$ as functions of $V_{\text{cell}}$. (b) Larmor precession frequencies ${\nu }_1$ and ${\nu }_2$ corresponding to $g$ factors of $g_1=1.15$ and $g_2=1.63$. (c)–(f) Schematic diagrams of charged QD spin dynamics under ${\sigma }^{+}$ pump excitation. For ${\sigma }^{-}$ pump pulses, all spin directions should be reversed. (c) In a fraction of CdSe QDs, optically injected electrons relax nonradiatively (green arrow) into surface defect states within a few ps, leaving no $1S_{\mathrm{e}}$ electron to probe. (d) For a $V_{\text{cell}}$ that fills the surface states, optically injected electrons remain in the $1S_{\mathrm{e}}$ level and FR can be measured in all QDs. (e) A circularly polarized pump pulse can only create an exciton in a charged QD with one spin orientation of the electrically injected charge. The singlet in the $1S_{\mathrm{e}}$ state prevents probe pulses from interacting with this QD. (f) Probe pulses can measure a singly charged QD which has not absorbed a pump pulse; there is no exciton precession, however.