Fig. 1. Current–voltage characteristics across Ag-Al2O3-CdSe MB-Au heterojunction. (a) I-V characteristics of three samples of Al2O3 thickness ∼3–4 nm. Data from sample 2 are multiplied by a factor of 10 for clarity. The onset voltages for six samples (three extra samples shown in Appendix C) are summarized into the histogram in the inset. The error bar indicates the standard deviation from all samples. The positive and negative onset voltages are the intersection voltages with the zero current axis obtained by linear fits to the data points in the corresponding voltage regions. The wide field optical image shows the heterojunction of sample 1 with the scale bar 5 μm. (b) COMSOL simulations on the I-V characteristics across Ag-Al2O3  (3 nm) -CdSe MB (500 nm)-Au heterojunction. In the simulations, the Ag work function and CdSe energy levels are kept the same while the conduction band of Al2O3 and Au work function are allowed to vary. The results demonstrate that the I-V can change significantly with even a small change in Al2O3 conduction band and Au work function. The inset shows the logarithm of electron density distribution, log(n), at zero bias across the entire heterojunction with the conduction band of Al2O3 at 3.74 eV and Au work function of 5.13 eV. Here the unit of n is m−3.

Fig. 2. Photocurrent across Ag-Al2O3 (3–4 nm)-CdSe MB-Au heterojunction. (a) Diagram of experimental setup. (b) Wide field optical image showing the laser beam incident on the Ag-Al2O3-CdSe MB heterojunction (sample 1). (c), (d) Peak photocurrent versus bias voltage on Au electrode of sample 1 and sample 4 at different incident laser powers. (e), (f) Zero bias photocurrent as a function of incident laser power for sample 1 and sample 4, respectively. The error bars indicate the fluctuation of reading from the lock-in amplifier. (g), (h) COMSOL simulation results of the photocurrent density versus bias voltage on Au electrode with Al2O3 conduction band at 3.74 eV (g) and 3.77 eV (h) and Au work function at 5.13 eV (g) and 5.1 eV (h) for incident laser power of 1×10−4  W. The simulated zero bias photocurrents as a function of incident laser power are given as insets.

Fig. 3. Impact of carrier redistribution on optical absorption loss in Ag-Al2O3-CdSe MB. (a) Diagram showing Ag-Al2O3-CdSe cross section and thicknesses of Al2O3 and CdSe used in this simulation. (b), (c) Energy diagrams of Ag-Al2O3-CdSe heterojunction and model system at zero bias, respectively. The affinity level (conduction band edge) of Al2O3 is set to 3.74 eV with a band gap of 2.6 eV. The work function of the metal in the model system is set to 5.366 eV, in the middle of CdSe bandgap. The bandgap of the oxide is set to 3 eV with the middle of the bandgap aligned with the metal work function. (d), (e) Electron density along the edge highlighted in red in (a) as a function of input power for Ag-Al2O3-CdSe (b) and the model system (c), respectively. The side plots give the electron density as a function of position at input power of 1 W. (f), (g) fv−fc as a function of input power corresponding to (d) and (e) in the z range from 0 nm to 80 nm, respectively. The side plots show the fv−fc as a function of position at input power of 1 W.

Fig. 4. Photocurrent across Ag-Al2O3 (4–5 nm)-CdSe MB-Au heterojunction. (a) Peak photocurrent as a function of bias voltage on Au for three devices measured with two different laser systems. The black and blue curves are offset for clarity. The dashed lines indicate the zero current lines for each curve. The arrows indicate the step-like onsets of photocurrents. Samples 5 and 6 were excited with a supercontinuum laser (repetition rate 40 MHz, pulse width ∼80  ps) after passing a bandpass filter (600±20  nm). The inset shows a wide field optical image of the laser spot focused on the Ag-Al2O3-CdSe MB heterojunction of sample 6, with scale bar of 5 μm. Sample 2 was excited with the Spark Antares laser at 532 nm (80 MHz, pulse width 5–6 ps). The peak input intensities associated with the three curves are 18.3  kW/cm2, 2585  kW/cm2, and 1495  kW/cm2 for samples 6, 5, and 2, respectively. (b) Sample 6 photocurrent as a function of bias voltage on Au for different incident laser powers. After the initial sharp increase, the current plateaus or follows a gradual linear increase. The inset shows the linear dependence of the plateaued current on the input power. The error bars indicate the range of current variation in the gradual linear increase regions.

Fig. 5. Geometry of Ag-Al2O3 (3 nm)-CdSe (200 nm)-Au heterojunction defined in COSMOL.

Fig. 6. Variation of dark current density across the Ag-Al2O3 (3 nm)-CdSe-Au heterojunction with the Ag and Au electrode separation (a) and thickness of the CdSe microbelt (b). The black curve in both panels is the same as the black curve in Fig. 1(b), with Al2O3 conduction band edge at 3.74 eV and Au work function at 5.1 eV. In current figure, all parameters are kept the same except for the one indicated in figure annotations. The thickness of the CdSe microbelt is 500 nm in (a), and the electrode separation is 400 nm in (b).

Fig. 7. Dark I-V characteristics of three more samples with Al2O3 thickness ∼3–4 nm.

Fig. 8. Hole density distribution (logp) at zero bias across the Ag-Al2O3 (3 nm)-CdSe (500 nm)-Au heterostructure with conduction band of Al2O3 at 3.74 eV and Au work function of 5.13 eV.

Fig. 9. Hole density, fv, and fc distributions at zero bias as a function of input power.

Fig. 10. Loss compensation measurements on the fundamental hybrid plasmonic mode supported on CdSe nanobelt-Al2O3-Ag and photonic mode supported on CdSe nanobelt on glass. Plots are adapted from Ref. [5].

Fig. 11. (a) Electric energy density distribution within Ag-Al2O3 (3 nm)-CdSe (200 nm thickness × 800  nm width) cross section when the fundamental hybrid plasmonic mode at 716 nm is excited. The side plot gives the electric energy density profile along the dashed line. The shadowed region indicates CdSe. (b) Electric energy density averaged fv−fc¯ for systems of Ag-Al2O3-CdSe with CdSe thickness of 200 nm, 100 nm, 50 nm, fv−fc¯ directly averaged over 10 nm in CdSe of 200 nm thickness closest to Ag, and fv−fc¯ for the model system. The dashed lines guide the eyes to the threshold input power, at which the dashed lines intercept with the top x axis.

Fig. 12. Photocurrent measurements on sample 6 at three different wavelengths. (a) Typical photocurrent versus bias voltage at different incident laser wavelengths. The curve obtained from 800 nm excitation is multiplied by a factor of 20 for clarity. (b) Dependence of the plateaued current on the input power. The error bars indicate the range of current variation in the gradual linear increase regions.

Fig. 13. Photocurrent measurement on four samples of Al2O3 thickness around 5 nm. The peak input intensity is 50  kW/cm2 for samples 7 and 8, 40  kW/cm2 for sample 9, and 75  kW/cm2 for sample 10.