Conductance peaks and exotic metallic behavior in an engineered nanoscale system: Signatures of correlated quasiparticles

We have studied conductance ($g$) of 1,4-butanedithiol-linked Au nanoparticle films (NPFs) through a percolation Mott-Hubbard insulator-to-metal transition (MH-MIT) from a Mott insulating phase with significant nanostructure-induced Coulomb charging barriers to a metallic phase. Near the transition, the NPFs traverse an exotic metallic phase exhibiting a pseudogaplike behavior and a zero-bias conductance peak. The peak decays rapidly with increasing temperature, yielding power-law fits with an average exponent of $3.8\pm 2.9$ and surviving in some samples up to remarkably high temperatures of $\sim$20 K. We propose that the peak is an Abrikosov-Suhl resonance previously observed in nonnanoengineered, correlated materials near MH-MITs. Since nanostructures can be controllably synthesized over a wide range of parameters, the study highlights an opportunity to use nanoengineered correlated materials as a powerful platform to study this exotic metallic phase and electron correlations.

Figure 1

Nanoparticle (NP) film morphology. (a) Evolution of film morphology with increasing number of 1,4-butanedithiol/Au nanoparticle immersion cycles. (b) A constant current scanning tunneling microscope image of a film with four immersion cycles on a silicon oxide/silicon substrate (current = 0.1 nA, tip bias = $-$1.2 V).

Figure 2

(a) A transmission electron microscope image of gold nanoparticles. (b) A UV-vis absorption spectrum of gold nanoparticles suspended in toluene. (c) UV-vis absorption spectra of self-assembled gold nanoparticle + alkanedithiol films on a glass substrate functionalized with aminosilane. The number of immersion cycles ranges from one (bottom curve) to eight (top curve). (d) Absorbance (420 nm) vs number of gold nanoparticle + alkanedithiol immersion cycles.

Figure 3

Four-probe conductance ($g$) data for an NPF at 12, 18, and 25 immersion cycles (S12, S18, and S25, respectively). (a) Zero-bias $g$ vs $T$. The data at each immersion cycle are normalized to their respective values at 300 K, $1.1\times {10}^{-5}$, $1.9\times {10}^{-4}$, and $3.5\times {10}^{-3}$ ${\Omega }^{-1}$ in order of increasing immersion cycles. Corresponding $g$ vs voltage ($V$) sweeps at various $T$s for (b) S12, (c) S25, and (d) S18. The $g$ vs $V$ sweeps at each $T$ are normalized to their respective values at (b) $-$90 and (d) $-$2 mV.

Figure 4

Four-probe control measurements exploring potential explanations for the zero-bias conductance peak observed in Fig. 3. (a) $g$ vs $V$ sweeps normalized at $-$45 mV and obtained at 2 K using a film at 6, 9, and 13 immersion cycles (S6, S9, and S13, respectively; sweeps offset for clarity). (a, inset) A magnified version of a $g$ vs $V$ sweep for S13. (b) Zero-bias $g$ vs $T$ data for S6, S9, and S13, normalized to their values at 300 K, $5\times {10}^{-4}$, $2\times {10}^{-3}$, and $8\times {10}^{-3}$ ${\Omega }^{-1}$, respectively. (c) $g$ vs $V$ sweeps at various $T$s for a different film with 11 immersions. (d) The zero-bias $g$ vs $T$ behavior corresponding to (c) with an inset of log-log plot of zero-bias conductance peak ($g_p$) vs $T$. The slope yields a power law, $g_p\propto T^{-\nu }$, with exponent $\nu =3.43$.

Figure 5

Control measurements exploring potential explanations for the zero-bias conductance peak observed in Fig. 3. The conductance vs voltage sweeps were obtained at various temperatures using aluminum/aluminum oxide tunnel junction contacts and two different films, each with 12 immersion cycles. The conductance vs voltage sweeps at each temperature in (a) are normalized to their respective values at $-$1.5 V. The insets show the corresponding conductance vs temperature behavior.

Figure 6

Schematic showing a percolation Hubbard lattice model for the nanoparticle film. A one-dimensional row array is used for illustration. As the number of immersion cycles increases, lattice sites are filled with nanoparticles, and linked nanoparticles form metallic clusters that grow. $U_i$ and $t_{ij}$ vary from site-to-site depending on local filling. Because of interactions between nanoparticles, $U_i$ tends to decrease and bandwidth tends to increase with increasing filling fraction. The evolution of density of states (DOS) of array with increasing filling fraction is shown on the left.