Distinguishing carbon nanotube defect chemistry using scanning gate spectroscopy

Electronic scattering at individual defect sites is presumably sensitive to defect chemistry. Here, we combine advances in carbon nanotube device fabrication and scanning probe characterization to investigate this correspondence. Specifically, we apply scanning gate spectroscopy (SGS) to the study of defects introduced into single walled carbon nanotubes by point functionalization in water, sulfuric acid, or hydrochloric acid. SGS measures the energy-dependent transmission functions of defect sites, and by working in the dilute limit of individual, isolated defects we empirically distinguish the three chemical types. A preliminary analysis proposes a scattering model in order to motivate further theoretical investigations of this one-dimensional scattering system.

Figure 1

(a) SGM and (b) corresponding SGS image of an s-SWNT having Schottky barriers at both electrodes and an additional, weakly scattering defect in the channel at $\Delta x$ $=$ 1.2 μm. Approximate position of source and drain electrodes are indicated with grey striping. Inset depicts the measurement geometry. (c) SGS for a similar s-SWNT over a different range of V${}_{\mathrm{tip}}$. (d) SGS image for a strongly scattering defect in a SWNT with only one sensitive interface. (e) SGS image for a pristine, small-bandgap SWNT that exhibits no appreciable Schottky barriers.

Figure 2

(a) Geometry defining the electrostatic coupling between a segment of the SWNT and $V_{\mathrm{tip}}$ and $V_{\mathrm{bg}}$. (b) Band diagram of a localized defect state interrupting the conduction band at the position of a defect. The potential of the state $V_{\mathrm{defect}}$ relative to $V_{\mathrm{sd}}$ is varied by the tip position and potential. The SWNT cartoon depicts an ether functionality believed to result from H${}_2$SO${}_4$ oxidation. (c) and (d) SGS profile of a defect with and without overlaid curves indicating fits to the contours of constant conductance. The voltage- and $x$-dependent contours determine the values of $C_{\mathrm{tip}-\mathrm{bg}}$ and $V_{\mathrm{defect}}$, as described in the text. (e) and (f) An example of the agreement between experimental SGS data (e) and a simple electrostatic model (f) that correctly simulates the voltage and $x$ dependencies.

Figure 3

Characteristic data for s-SWNTs point functionalized in (a) H${}_2$O, (b) H${}_2$SO${}_4$, and (c) HCl. Each depicts a portion of an SGS data set surrounding the defect site, the maximal $G$($V_{\mathrm{tip}}$) measured directly over the defect, and a transmission function $T$($E$) calculated from $G$($V_{\mathrm{tip}}$) and the contact resistance, as described in the text.

Figure 4

(a) A defect created in H${}_2$SO${}_4$ produces two similar but distinct $G$($V_{\mathrm{bg}}$) behaviors. For each state, two overlapping curves depict measurements at $V_{\mathrm{sd}}$ $=$ $+$100 and $-$100 mV. (b) In the first state (red curve), the SGS image depicts scattering at $V_{\mathrm{tip}}$ < 0. (c) In the second state (blue curve), SGS imaging only occurs for $V_{\mathrm{tip}}$ > 0. (d) During some acquisitions, the device switches between its two states as the probe passes the defect site responsible for the response. A single color scale connects all three images to the $G$($V_{\mathrm{bg}}$) data.

Figure 5

(a) A defect produced in H${}_2$O results in two dissimilar $G$($V_{bg}$) behaviors that are predominantly $p$ or $n$ type. (b) A single SGS image from the same device, capturing switching between the two states as the tip traverses the defect site. Defects produced in H${}_2$O prove to be the most difficult to image, even when acquired using modest $V_{\mathrm{tip}}$ ranges.

Figure 6

Variability of point functionalization in HCl for three SWNT devices. (a) One or two discrete drops in conductance occurring in $\Delta t$ < 1 s indicates the most likely creation of a covalent point defect. (b) Multiple drops with bistable dynamics over a few seconds suggests more complex damage, such as the creation and clustering of multiple sites. (c) A continuous decrease over many seconds indicates extensive and irreversible oxidation. In all three examples, individual data points correspond to 5 kHz sampling rate. Insets show example SGS images for the less ideal cases.