Revealing the impact of ambient molecular contamination on scanning tunneling microscopy and spectroscopy of layered materials

Abstract

Hydrocarbon contamination is an ever-present factor to consider in surface science measurements. In the case of van der Waals material surfaces, the structure of this contamination has become known in recent years as a self-assembled layer of normal-alkanes, resulting from a few days' exposure to ambient air. Knowledge of its composition and structure enables systematic investigation of its influence on surface properties. Here, we investigate the effect of this contamination on scanning tunneling microscopy (STM) and spectroscopy measurements by comparing clean and ambient alkane-contaminated surfaces of graphite. Our results reveal that the ambient alkane layer suppresses the well-known phonon-induced gap near the Fermi energy, resolving a long-standing inconsistency in STM studies, where this feature is often absent. Furthermore, we show that the presence of the contamination layer alters the current-distance ($I(z)$) characteristics, flattening its exponential decay by a factor of 1.5 to 5 compared to the clean surface. This change arises from extra conductance channels through the alkane layer alongside the tunnel junction, as the tip penetrates the contaminant overlayer. Finally, based on the $I(z)$ characteristics, we provide a practical guide to detect the presence of surface contamination in STM measurements.

Figures (4)

• Figure 1: The effect of ambient contamination in STM measurements of graphite.(a) Schematic image of the n-alkane self-assembled monolayer stemming from ambient air contamination palinkas on a graphite surface. While scanning, the STM tip can either penetrate this layer, or can scan above it. (b) Constant current STM topography image of contaminated graphite, with the STM scanning above the contaminant layer. The self-assembled structure of the n-alkanes is visible palinkas. (c) Constant current STM topography image of UHV cleaved graphite sample. (d) Constant current STM topography image when the tip penetrates the alkane layer. The honeycomb lattice of graphene is marked with red hexagons. The images represent raw data, processed only by line-by-line linear background subtraction along the fast scan direction. Scale bars: 5 nm in (b), 2 nm in (c, d). Tunneling current setpoint ($I_{\mathrm{t}}$) and sample bias ($V_{\mathrm{b}}$) values are displayed on the top of the panels. Data in (b) and (d) are measured with the same tip. (e) Representative tunneling conductance measurement ($dI/dV$) performed on UHV cleaved and contaminated graphite samples. The green dashed lines show the position of the phonon induced gap edges Natterer2015-xy. Spectra are offset for clarity. Lock-in parameters: 10 mV modulation at 1267 Hz. (f) Single examples of $I(z)$ spectroscopy on clean, contaminated graphite and gold samples, by measuring the tunneling current as the tip is retracted. The current of the curves are normalized to 1, by dividing by the setpoint current ($I_{\mathrm{t}}$), for better comparison of the decay. The tip - sample distance defined by the current setpoint is chosen as $z = 0$.
• Figure 2: Influence of surface contamination on $I(z)$ spectra.(a-d)$I(z)$ spectroscopy curves presented as 2D histograms measured on a contaminated HOPG sample with increasing setpoint. The histogram color scale presents the number of data points in a specific tip - sample distance ($z$) and tunneling current ($I$) range. The histograms are composed of data from more than 40000 individual $I(z)$ spectra, measured by first retracting the tip from the surface ("backward"), then approaching the sample ("forward"). The red and green dotted curves represent the averaged forward and backward curves. The blue curves are measured on UHV-cleaved samples. Setpoint $I_{\mathrm{t}}$ values are shown on top, $V_{\mathrm{b}} = 500$ mV. All spectra were recorded with the same tip. (e) Schematic representation of the tip-sample interaction with the n-alkane layer, for different tip-sample separations as marked by numbers in (d). In all panels, the tip - sample distance defined by the current setpoint is chosen as the $z = 0$ value.
• Figure 3: Conductance fluctuations.(a) Examples of individual $I(z)$ curves for the contaminated and clean graphite surfaces, measured by retracting the tip. (b) Histogram of conductance values for specific $I$ and $z$ values, for the alkane contaminated surface. Data from 20480 individual $I(z)$ curves. The spectra were measured by retracting the tip from the distance set by the current setpoint ($I_\mathrm{t}$ = 250 pA). The conductance fluctuation is calculated by subtracting the mean ($\left < I_{\mathrm{t}}(z) \right >$) from all individual $I(z)$ curves, dividing by the sample bias and taking the absolute value of it: $\left | \left ( I_{\mathrm{t}}(z) - \left < I_{\mathrm{t}}(z) \right > \right ) / V_{\mathrm{b}} \right |$. (c) Histogram of conductance fluctuations for all $z$ values. Increasing $I_{\mathrm{t}}$, we increase the number and magnitude of conductance fluctuations. In all cases $V_{\mathrm{b}}$ = 500 mV. The tip - sample distance defined by the current setpoint is chosen as $z = 0$.
• Figure 4: Ab-initio calculations and measured decay constants.(a) DOS decay of a bilayer graphite surface with and without an alkane monolayer on top. The alkane layer modifies the decay of the graphite wave functions into the vacuum. The zero value of the distance coordinate is the position of the topmost C atom. The DOS values are integrated from the Fermi level to 0.5 eV for both cases. Oscillations of the DOS at large distances are due to numerical error. (b) Decay of the ab-initio calculated DOS into the vacuum of the pure bilayer graphite surface and the n-alkane covered one. Both the clean and alkane covered DOS plot starts from the same DOS value to match STM $I(z)$ conditions. Same data as in (a). (c) Decay constant of the calculated DOS, obtained by differentiating the $\mathrm{log}$ of the DOS values versus distance from the surface $z$. (d)$I(z)$ curves measured on clean and contaminated graphite samples, with the tip at various distances inside the contaminant monolayer. Current setpoints shown in the legend. (e) Decay constants ($\kappa$, see eq. \ref{['eq:decay']}) for the $I(z)$ curves in (d). The error bars correspond to two standard deviations of the current noise at zero current (tip far from sample). The decay constant and DOS decay are calculated by taking the $log$ of the values in panels b and d and calculating the derivative with respect to distance $z$. In panels d and e, the tip - sample distance defined by the current setpoint is chosen as $z = 0$.