Towards the insulator-to-metal transition at the surface of ion-gated nanocrystalline diamond films

Abstract

Hole doping can control the conductivity of diamond either through boron substitution, or carrier accumulation in a field-effect transistor. In this work, we combine the two methods to investigate the insulator-to-metal transition at the surface of nanocrystalline diamond films. The finite boron doping strongly increases the maximum hole density which can be induced electrostatically with respect to intrinsic diamond. The ionic gate pushes the conductivity of the film surface away from the variable-range hopping regime and into the quantum critical regime. However, the combination of the strong intrinsic surface disorder due to a non-negligible surface roughness, and the introduction of extra scattering centers by the ionic gate, prevents the surface accumulation layer to reach the metallic regime.

Two-dimensional hole transport in ion-gated diamond surfaces: A brief review (Review article)

“…Indeed, when the applied negative V G is sufficiently large, σ 2D becomes nearly T -independent for T → 0 in both the (111) and (100) hydrogenated surfaces [45][46][47], suggesting a gate-induced insulator-tometal transition (IMT). Conversely, this low-T saturation is never observed in the NCD B-doped surfaces [50] (see Fig.3c). No evidence for SC behavior has been observed so far in any ion-gated diamond surfaces down to the lowest measured T , most likely the achieved values of n 2D are not large enough [45][46][47][48][49][50].…”

“…Namely, µ no longer monotonously decreases with increasing n 2D , showing instead an increasing trend at low n 2D , reaching a maximum and then decreasing (see Fig.3e). This behavior can be observed in all iongated diamond surfaces [45][46][47][48]50]. Since at low T the electron-phonon scattering is suppressed, this behavior can be attributed to a crossover in the scattering with defects, specifically Coulomb scattering with the ions in the EDL.…”

“…The behavior of the induced carrier density n 2D as a function of the applied V G is also dependent on the diamond surface (see Fig.2c). In general, n 2D becomes larger for larger negative values of V G , with at least one voltage range -usually at low V G -where the increase is linear and the diamond/electrolyte interface capacitance C G is nearly constant [45,47,50]. The exact values and trends are dependent on which electrolyte is used and at which temperature T one measures n 2D , as is the maximum negative V G which can be applied before the onset of irreversible electrochemical reactions.…”

“…As we show in Fig.2d, the mobilities of the (111) and (100) hydrogenated sur-faces are comparable, the values of µ for the (100) surface being slightly larger for most of the values of n 2D (though it could be partially due to sample-to-sample fluctuations) [45,48]. On the other hand, the values of µ for the B-doped NCD surface are two orders of magnitude lower at comparable n 2D , obviously due to the polycrystalline nature of the samples and their much larger surface roughness (∼ 30 nm for NCD vs. ∼ 0.5 nm for single-crystals) [50]. Despite these differences, µ is always decreasing with increasing n 2D for all diamond surfaces.…”

“…When β > 1/3, σ 2D still vanishes for T → 0, identifying the insulating side of the IMT. Notably, the NCD B-doped surface can never be brought beyond this regime even at the largest values of n 2D , most likely due to its very low mobility caused by the large surface roughness [50]. On the other hand, sufficiently large values of n 2D allow the hydrogenated single-crystal surfaces to cross the β = 1/3 line (the QC point of the IMT [72][73][74]) and reach the metallic regime where a finite σ 2D exists for T → 0.…”