Electroluminescence of copper-nitride nanocrystals

Nanocrystals can behave as quantum boxes with confined electronic states governing their optoelectronic properties. The formation of nanometer-size crystals of copper nitride $\left({\mathrm{Cu}}_3\mathrm{N}\right)$ grown by nitrogen sputtering of a Cu(110) surface is reported. Scanning tunneling spectroscopy shows that the nanocrystals exhibit a series of well-defined sharp electronic resonances, which correspond to confined free-electron-like states. We observe that electrons from a scanning tunneling microscope tip induce the emission of light with a larger efficiency than on the bare metal surface. The spectral analysis of the emitted photons reveals various radiative inelastic pathways enabled by the confined states, which explain the enhanced light emission. Thus, the ${\mathrm{Cu}}_3\mathrm{N}$ nanocrystals can be employed as nanometer-size light sources.

Figure 1

(a) STM image of copper-nitride nanocrystals formed on Cu(110). The corrugated striped phase corresponds to the ${\mathrm{Cu}}_2\mathrm{N}$ monolayer, and the bright rectangular protrusions are the Cu3N nanocrystals. (b), (c) Atomically resolved structure of ${\mathrm{Cu}}_3\mathrm{N}$ nanocrystals: tunneling current STM image (b) and AFM (c). The size of the unit cell obtained from the images fits with the unit cell dimensions of the (100) face of a ${\mathrm{Cu}}_3\mathrm{N}$ crystal. (d) Schematic structure of ${\mathrm{Cu}}_3\mathrm{N}$ monolayer. The N atoms are incorporated into the Cu layer. The gray rectangle marks the unit cell of the monolayer, the yellow square the unit cell of the nanocrystal. (e) Schematic crystal structure of three-dimensional ${\mathrm{Cu}}_3\mathrm{N}$. The arrow shows the direction of stripes in the ${\mathrm{Cu}}_3\mathrm{N}$ network (image processing using WSxM [36]).

Figure 2

(a) Large-scale STM image of a Cu(110) surface partially covered by the ${\mathrm{Cu}}_2\mathrm{N}$ network, showing different types of ${\mathrm{Cu}}_3\mathrm{N}$ nanocrystals. (A), (B) Nanocrystals embedded within the Cu surface, (C) nanocrystal grown on ${\mathrm{Cu}}_2\mathrm{N}$ terraces, (D) nanocrystal protruding above clean Cu areas. Height profiles taken along the plotted lines above the different nanocrystals are shown in the graph. (b) STM-current (error signal) image of a region at the border between a nanocrystal and the ${\mathrm{Cu}}_2\mathrm{N}$ film. The atomic lattice is continuous as shown in the model. (c) Height profiles of a nanocrystal (inset) show a curved shape in both STM and AFM measurements. Electronic effects in the measurement are excluded since AFM data are taken in $\Delta f$ feedback mode.

Figure 3

Comparison of constant-current $dI/dV$ spectra recorded on a ${\mathrm{Cu}}_3\mathrm{N}$ nanocrystal and on the bare ${\mathrm{Cu}}_2\mathrm{N}$ covered Cu(110) surface.

Figure 4

$dI/d{V}_s$ spectrum recorded on a ${\mathrm{Cu}}_3\mathrm{N}$ nanocrystal. The tip-sample distance was linearly decreased with bias (2.8 Å) in order to extend the dynamic range of the current voltage converter to cover the wide range of bias values. Various features are labeled as QW (quantum well state), ${\mathrm{S}}_1$, and ${\mathrm{S}}_2$ (surface states), and described in the text. The insets show constant-height (left) and constant-current (right) $dI/d{V}_s$ spectra around the energy of the resonances QW, and ${\mathrm{S}}_1$, and ${\mathrm{S}}_2$, respectively.

Figure 5

(a) QW state: $dI/dV$ images at the bias corresponding to the substructure marked with arrows in the spectrum shown in the rightmost panel. (b) Similar $dI/d{V}_s$ maps of resonances ${\mathrm{S}}_1$ and ${\mathrm{S}}_1^{\prime }$ and ${\mathrm{S}}_2$ and ${\mathrm{S}}_2^{\prime }$ for a different nanocrystal.

Figure 6

(a) Comparison of light intensity per tunneling electron $I\left(h\nu \right)$ spectra obtained on the ${\mathrm{Cu}}_3\mathrm{N}$ nanocrystal of Fig. 2, on the ${\mathrm{Cu}}_2\mathrm{N}$ layer, and on the Cu(110) surface $(I=20$ nA, $V_s=1.6$ V, time = 180 s). The dotted gray line shows the reference spectrum, measured on a bare Cu(110) region with $V_s=3$ V, used to normalize the data in (b). (b) Plots of the spectral yield $Y\left(h\nu \right)$ obtained by normalization of the data of (a). By removal of the plasmon-related features, the resulting spectra are tip-independent, as described in the Appendix. (c) Spectral yields of a ${\mathrm{Cu}}_3\mathrm{N}$ nanocrystal and of the ${\mathrm{Cu}}_2\mathrm{N}$ layer relative to that of the Cu(110) surface. The plots quantify the larger photon emission efficiency of the nanocrystal, and reveal a photon energy cutoff associated to the position of the QW resonance.

Figure 7

(a) Photon yield spectra versus applied bias. The photon yield (color coded) is obtained using a reference spectrum measured on the bare Cu(110) surface at $V_s=3$ V. Dashed lines mark cutoffs in the spectra which shift linearly with the applied bias. These correspond to the maximum photon energy for a certain transition into a resonance, and thus allow identifying the different final states. The different radiative transitions observed are labeled as I, II, III, and Lm. Each spectrum is acquired for 120 s, with $I_t=50$ nA. (b) $dI/d{V}_s$ (constant-current mode, $I=0.5$ nA) acquired over the nanocrystal with a QW state at 0.8 V. Arrows indicate the decay processes identified in (a). Dotted red curve shows the integrated photon counts.

Figure 8

(a) Evolution of the photon yield spectra $Y\left(h\nu \right)$ from Fig. 7 at sample bias between 3.6 and 4.3 V. The peak Lm which evolves in the yield spectra corresponds to the direct electronic transition between the states ${\mathrm{S}}_2$ and ${\mathrm{S}}_1$ of the nanocrystal, as shown in the top right inset image. The bottom left inset shows the fit of the luminescence spectra by a Gaussian peak. (b) Bias dependence of the total quantum efficiency and the quantum efficiency of the luminescence process. (c) Schematic image of the inelastic decay processes giving rise to direct (III) and internal (Lm) radiative transition mechanisms.

Figure 9

(a) Set of light emission spectra obtained on a bare Ag(111) surface using a W tip, presumably coated by silver after indentations into the substrate. Each spectrum is measured by recording photons for 4 minutes with the STM feedback stabilized at $I=5$ nA and at different sample bias values. The spectra are divided by the number of tunneling electrons $I_t\times t$. (b) Spectral yield obtained by dividing each of the spectra in (a) by the spectrum recorded at the highest bias value (2.5 V) and by an additional factor of 2.2 and 1.7 introduced to fit the resulting Yield with the dashed model function $\frac{\left|eV\right|-h\nu }{|e{V}_{bg}|-h\nu }$ described in the text. The origin of this constant factor is probably due to a decrease in the plasmon enhancement with tip-sample separation.