Substitutional Tin Acceptor States in Black Phosphorus

Abstract

Nominally pure black phosphorus (BP) is commonly found to be a p-type semiconductor, suggesting the ubiquitious presence of impurity species or intrinsic, charged defects. Moreover, scanning tunneling microscopy (STM) images of black phosphorus reveal the presence of long-range doublelobed defect features superimposed onto the surface atomic lattice. We show that both the p-type doping of BP and the defect features observed in STM images can be attributed to substitutional tin impurities. We show that black phosphorus samples produced through two common synthesis pathways contain tin impurities, and we demonstrate that the ground state of substitutional tin impurities is negatively charged for a wide range of Fermi level positions within the BP band gap. The localized negative charge of the tin impurities induces hydrogenic states in the band gap, and it is the 2p level that sits at the valence band edge that gives rise to the double-lobed features observed in STM images.

Identification and Manipulation of Defects in Black Phosphorus

“…At negative bias, these defects have an asymmetric dumbbell shape oriented along the armchair direction and extending over ∼8 nm. Such features have been remarked on by previous authors. ,,, Importantly, we observe that, on going from negative to positive bias, there is a contrast reversal (see Figure e and f) and shape change, so that these same defects (labeled as type- A defects) now appear as dark circles; this has not been noted previously in the literature. We identify five salient features of defect A : (1) bright contrast (higher LDOS with respect to pristine BP) at negative bias, (2) dumbbell shape at negative bias, (3) comparatively dark center (see magenta marker) between the lobes of the dumbbell at negative bias, (4) dark contrast (lower LDOS) with respect to pristine BP at positive bias, and (5) circular shape at positive bias.…”

“…Previous studies have reported spectroscopic signatures of dumbbell defects but focused only on a low energy range (below 0.4 eV). Different low energy characteristics have been reported such as two resonant states around the Fermi level separated by 27 meV, two acceptor states at −0.13 and −0.21 eV, or a peak at 0.02 eV and a shoulder at 0.12 eV . We do not observe such low energy features; however, we find clear electronic states at higher energies, as described below.…”

“…We note that, to date, no studies have shown a successful match between theoretically simulated and experimental STM images; indeed, previous authors reported that their simulated STM images of a MV or an Sn impurity in the topmost phosphorene sublayer (Sn-1) resulted in localized features quite different from the extended dumbbell imaged experimentally . We note too that the identification of the dumbbells as arising from MVs is called into question by MVs having a very low diffusion barrier, and also being unstable with respect to divacancies . We therefore believe that there are still several open questions regarding the nature of experimentally observed defects in BP.…”

“…While defects in BP can be a limiting factor for materials performance, they can also confer desired properties . Defects on the BP surface have been imaged using scanning tunneling microscopy (STM). − A commonly observed defect feature is an asymmetric dumbbell shape, extending several nanometers along the armchair direction; this has also been associated as the source of p-doping found in BP crystals. However, there is hitherto no consensus in the literature on the type of defect corresponding to such a dumbbell feature: some authors have identified the dumbbells as arising from monovacancies (MVs), ,, whereas others have concluded they arise from neutral or charged Sn substitutional defects. , Further confusion has arisen from scanning tunneling spectroscopy (STS) data on these defects from different groups showing peaks at different energies. ,, …”

“…Until now, it has proved difficult to perform full-fledged density functional theory (DFT) calculations on candidate defect systems because of the extended length scales over which the effect of the defects are manifested. Therefore, previous authors have made use of tight binding calculations , and/or modeled the defect as a hydrogenic acceptor , whose ground and first excited states look like an ellipse and dumbbell, respectively. We note that, to date, no studies have shown a successful match between theoretically simulated and experimental STM images; indeed, previous authors reported that their simulated STM images of a MV or an Sn impurity in the topmost phosphorene sublayer (Sn-1) resulted in localized features quite different from the extended dumbbell imaged experimentally .…”