Effects of oxygen on the electron transport properties of carbon nanotubes: Ultraviolet desorption and thermally induced processes

Highly sensitive electrical response of carbon nanotubes to UV exposure and thermal treatments are examined. Low intensity UV-induced conductivity change in individual single-walled carbon nanotube transistors is found to be mainly arising from oxygen desorption at the metal contacts rather than from nanotubes or possible amorphous side products. Choice of metal or thermal annealing (which changes the metal-nanotube contacts) eliminates or minimizes this low intensity UV response. However, a slow p-doping process is observed upon thermal annealing followed by equilibration in ${\mathrm{O}}_2$ or air as suggested by changes in the near infrared spectra of nanotube films and in individual nanotube transport characteristics.

Figure 1

The transfer characteristics of individual SWNT transistors before (solid curves) and after (dashed curves) $4\min$ exposure to $3.65\mathrm{eV}$ $\left(340\mathrm{nm}\right)$ UV light with intensity $\sim 3\mathrm{mW}{\mathrm{cm}}^{-2}$. Similar responses are observed for SWNT transistors fabricated using $\mathrm{C}{\mathrm{H}}_4$ CVD (a) and ethanol CVD (b). Drain-source bias is $100\mathrm{mV}$ in both cases.

Figure 2

The comparison of transport characteristics of SWNTs with different metal contacts. The transfer characteristics shown in (a) and (b) are for SWNTs contacted with $>10\mathrm{nm}$ Ti with $50\mathrm{nm}$ Au capping layer (Ti-SWNT contacts). Pd-contacted SWNT is shown in (c). Titanium as contacts can lead to high on-state conductance with linear $I\text{−}V$ characteristics (inset a) at large negative gate voltages but sometimes non-linear rectifying behavior is also seen (inset b). The $I\text{−}V$ characteristic of Pd-contacted SWNT at the indicated gate voltage is shown in the inset of (c).

Figure 3

The $I\text{−}V$ curves before (thin curve) and after (bold curve) $11\min$ UV (photon $\text{energy}=3.65\mathrm{eV}$) exposure under Ar for a Ti-only contacted SWNT transistor (a). The lower inset shows the corresponding transfer characteristics at $V_{\mathrm{DS}}=100\mathrm{mV}$. Note that the bold transfer curve in this inset is measured while the UV source is still on. The schematic of band alignment for $p$-channel operation before and after UV desorption of oxygen is shown in the upper inset. The change in the $p$-channel conductance, plotted as the natural log of the normalized conductance, due to exposure to UV (intensity $\sim 1.6\mathrm{mW}{\mathrm{cm}}^{-2}$ for $3.94\mathrm{eV}$ photons) is shown in (b). Only small voltages ($V_{\mathrm{DS}}=10\mathrm{mV}$ and $V_g=0\mathrm{V}$) were used to minimized any effects due to applied electrostatic potential. The solid curve is the least squares fit to Eq. (2) given in the text. The inset is the photon energy dependence of the photodesorption cross sections obtained from the least squares fits. Prior to measurements at each photon energy, the sample was heated in air at $100°\mathrm{C}$ for $30\min$ to accelerate full recovery by enhancing surface oxidation (similar process as in Ref. 30).

Figure 4

The transfer characteristics showing the response of SWNT transistors to UV exposure before and after thermal annealing under Ar at $500°\mathrm{C}$. $\mathrm{Ti}∕\mathrm{Au}$ contacted SWNT (a) exhibits large $p$-channel decay upon UV (photon $\text{energy}=4.14\mathrm{eV}$ with intensity $\sim 1.8\mathrm{mW}{\mathrm{cm}}^{-2}$) irradiation before annealing but no longer responds after (inset). On the other hand, Ti-only contacted SWNT (b) exhibits large response to UV (photon $\text{energy}=3.65\mathrm{eV}$ with intensity $\sim 3\mathrm{mW}{\mathrm{cm}}^{-2}$) before and after (inset) annealing.

Figure 5

The changes in the transfer characteristics of Pd-contacted SWNT to UV (photon $\text{energy}=3.65\mathrm{eV}$ with intensity $\sim 10\mathrm{mW}{\mathrm{cm}}^{-2}$). The inset compares the $p$-channel conductance (average conductance between $V_g=-13$ and $-15\mathrm{V}$ in the transconductance measurements) decay of Pd-contacted SWNT with Ti-contacted SWNT transistors at the same photon energy and intensity.

Figure 6

The transfer characteristics of Pd-contacted SWNT under ${\mathrm{O}}_2$ atmosphere (a) and in vacuum $(\sim {10}^{-7}\text{torr})$ after thermal annealing at $150°\mathrm{C}$ for $2\text{days}$ (b). The drain-source voltage is $100\mathrm{mV}$ for both measurements. The $p$-channel output characteristics under ${\mathrm{O}}_2$ and $n$-channel output under vacuum are shown in (c) and (d), respectively, with the corresponding gate voltages.

Figure 7

The transfer characteristics of the same device as shown in Fig. 6 but prior to complete equilibration under ${\mathrm{O}}_2$. Both the $p$ channel at negative gate voltages and the $n$ channel at positive gate voltages are observed. The inset is the same curve plotted in log scale. The dashed line corresponds to subthreshold swing of $1\mathrm{V}$ per decade.

Figure 8

The transfer characteristics of thick-$\mathrm{Ti}∕\mathrm{Au}$-contacted SWNT before (solid line) and after thermal annealing under Ar at $500°\mathrm{C}$ (dashed line). Both forward and reverse sweeps are plotted to show that the threshold shift after thermal annealing followed by equilibration in air is larger than the hysteresis.

Figure 9

The near IR absorption spectra of SWNT mats after thermal annealing under vacuum at $400°\mathrm{C}$ for $3\mathrm{h}$. The solid curve is measured under vacuum and the dashed curve is measured in air after equilibrating for $4\text{days}$.