Toward air-stable multilayer phosphorene thin-films and transistors

Few-layer black phosphorus (BP), also known as phosphorene, is poised to be the most attractive graphene analogue owing to its high mobility approaching that of graphene, and its thickness-tunable band gap that can be as large as that of molybdenum disulfide. In essence, phosphorene represents the much sought after high-mobility, large direct band gap two-dimensional layered crystal that is ideal for optoelectronics and flexible devices. However, its instability in air is of paramount concern for practical applications. Here, we demonstrate air-stable BP devices with dielectric and hydrophobic encapsulation. Microscopy, spectroscopy, and transport techniques were employed to elucidate the aging mechanism, which can initiate from the BP surface for bare samples, or edges for samples with thin dielectric coating, highlighting the ineffectiveness of conventional scaled dielectrics. Our months-long studies indicate that a double layer capping of Al2O3 and hydrophobic fluoropolymer affords BP devices and transistors with indefinite air-stability for the first time, overcoming a critical material challenge for applied research and development.

Compositional analysis of fully degraded black phosphorus was performed by energy dispersive X-ray analysis (EDAX) in the scanning electron microscopy (SEM) chamber. The black phosphorus sample in Supplementary Fig. 2 was exfoliated onto an Au/Si substrate and kept in air for three days. Before SEM-EDAX analysis, the sample was baked on a hot plate at 100 °C for 10 minutes. From the AFM data, the two droplet-like spots on the black phosphorus surface show a height of 78 nm, much taller than nearby flat regions. The EDAX maps reveal significant oxygen signals on the degraded flake, indicative of oxidation in the aging process. Raman spectra of a typical degraded black phosphorus flake (red line) and the Al2O3/P ++ Si background (blue). The inset shows the detailed Raman spectrum of the degraded black phosphorus with several obvious peaks from 800 to 950 cm -1 (black line, exposed to air for 3 days) and pristine black phosphorus with much smaller or negligible peaks (red line, within an hour after exfoliation). These Raman signatures are consistent with the reported P-O stretching modes due to oxygen-containing black phosphorus compounds.
The AFM-based microwave impedance microscopy (MIM) setup is shown in Supplementary Figure 4. Details of the shielded cantilever probe are described in [S6]. The 1 GHz excitation signal is guided to the cantilever probe through an impedancematch (Z-match) section [S7]. During the scanning, the near-field interaction modifies the effective tip impedance, which results a change of the reflected microwave. The MIM circuit then amplifies and demodulates this 1 GHz signal. The two orthogonal channels are aligned to the real (MIM-Re) and imaginary (MIM-Im) parts of the tip-sample admittance by adjusting the reference phase to the mixer. Through finite-element analysis (FEA) modeling, the local permittivity () and/or conductivity () of the sample can be deduced from the two MIM outputs [S6]. In general, the MIM-Im signal, which is proportional to the tip-sample capacitance, increases monotonically as a function of the sample conductivity. The MIM-Re signal, which corresponds to loss at 1 GHz, peaks at intermediate conductivities and decreases for both good metals and good insulators.

Supplementary Figure 4. Schematic illustration of the MIM setup.
The MIM is based a commercial AFM with standard laser feedback for topographic imaging. The tip-sample interaction can be modeled as lumped elements (inset). The MIM electronics shown in the dashed box detect the impedance change and generate the two orthogonal (MIM-Im and MIM-Re) output signals.
The raw MIM/AFM data of the thin-cap black phosphorus sample in Fig. 2 are shown in Supplementary Fig. 5. The flake with a thickness d = 24 nm was capped by ~3 nm Al2O3, which was formed by the deposition of 2 nm Al and the subsequent oxidation at 120 °C. The sample was then measured daily by MIM/AFM over one week. To understand the MIM data in terms of local conductivity, numerical analysis using the commercial software COMSOL4.3 [S6] was performed. The modeling took the sample geometry in Supplementary Fig. 5d and computed the real and imaginary parts of the tip-sample admittance (reciprocal of impedance) as a function of the conductivity , which is then converted to local sheet resistance Rsh = 1/(d) for comparison with the transport data. Due to the tip wearing from repeated scans, a relatively blunt tip with effective radius of curvature 120 nm was used in the simulation. An electrically uniform black phosphorus was assumed for the 2D axisymmetric model and the result is only semi-quantitative. The MIM-Im data, for example, are strongly affected by topography and non-uniform electrical properties. As a result, we re-plotted the MIM-Re data to approximate the Rsh maps in Fig. 2b, with its false color scale derived from the simulated response in Supplementary Fig. 5e. Note that the MIM-Re signal is a non-monotonic function of Rsh [S6] so the insulating substrate is shown as blue in Rsh. The red, yellow, and green lines correspond to Rsh=10 4 , 10 5 , and 10 6 /sq, Raman mapping on the thin-cap black phosphorus sample was performed at Day 9, i.e., after the week-long AFM/MIM experiment. As shown in Supplementary Figure 6, all three characteristic peaks (Ag 1 , B2g, and Ag 2 ) were still clearly observed in the Raman maps, indicating the preservation of black phosphorus. The larger signals near the edges are probably due to the thicker material here (see Fig. 1c). Interestingly, from Supplementary Fig. 6d, significant Raman signals are also seen in the range of 800900 cm -1 . The oxidation process initiated from the sample edges is therefore important for black phosphorus with a thin dielectric capping.
A complete set of the raw MIM/AFM data of the thick-cap black phosphorus sample in Fig. 3 are shown in Supplementary Figure 7. The flake with a thickness d = 16 nm was capped by 25 nm ALD Al2O3. It is obvious from the raw data that neither the local conductivity nor the sample topography was affected by the exposure to ambient oxygen and moisture. Numerical simulation was performed based on the sample geometry in Supplementary Fig. 7d. Unlike the thin-cap sample, the thick-cap sample is electrically homogeneous so the local sheet resistance mapping can be obtained by either the MIM-Im or MIM-Re signals. For consistency purposes, we again re-plotted the MIM-Re data to approximate the Rsh maps in Fig. 3(b), with its false color scale derived from the simulated response Supplementary Fig. 7e. The insulating substrate is again shown as blue in Fig. 3b.  Fig. 4. The representative Id -Vg plots are shown here for typical (a) thin capped (~200 cm 2 /Vs), (b) thick capped (~210 cm 2 /Vs) and (c) double capped (~190 cm 2 /Vs) devices. Dashed lines are linear fits for mobility extraction. Crystal orientation of BP is unknown in this measurement.
Supplementary Media File. AFM video clip of exposed black phosphorus sample. A black phosphorus flake with initial thickness of 8 nm was monitored by AFM for over one week (170 hours). Local bubbles, a clear sign of degradation and moisture accumulation, reached up to 50 nm height after a week of measurement. AFM measurement was done in ambient condition.