Direct Observation of Deep Excitonic States in the Photoluminescence Spectra of Single-Walled Carbon Nanotubes

Low-energy, dark excitonic states have recently been predicted to lie below the first bright ($E_{11}$) exciton in semiconducting single-walled carbon nanotubes [Phys. Rev. Lett. 93, 157402 (2004)]. Decay into such deep excitonic states is implicated as a mechanism which reduces photoluminescence quantum yields. In this study we report the first direct observation of deep excitons in SWNTs. Photoluminescence (PL) microscopy of suspended semiconducting single-walled carbon nanotubes (SWNTs) reveals weak emission satellites redshifted by $\sim 38–45$ and $\sim 100–130\mathrm{meV}$ relative to the main $E_{11}$ PL emission peaks. Similar satellites, redshifted by 95–145 meV depending on nanotube species, were also found in PL measurements of ensembles of SWNTs in water-surfactant dispersions. The relative intensities of these deep exciton emission features depend on the nanotube surroundings.

Figure 1

Typical microphotoluminescence maps (PL intensity vs excitation and emission wavelengths) of suspended SWNTs (single emitters): (a) (9,7) nanotube probed at 293 K and (b) (9,4) nanotube at 6 K. The bottom panels show emission spectra as acquired at the absorption maximum of PL peaks. Excitation (PLE) satellites are lettered with S. The ($n$, $m$) assignment corresponds to that of Ref. 15. Note that the emission intensity scales in the 2D maps (color bars) are not linear.

Figure 2

Photoluminescence map in the $E_{11}$-excitation region of HiPco nanotubes dispersed in ${\mathrm{D}}_2\mathrm{O}/1$ wt.% SDS. The white stripe of high intensity is due to Rayleigh scattered excitation light. As a result of uniformly small Stokes shifts, $E_{11}$ excitation and $E_{11}$ emission peaks overlap on this stripe (indicated as ($n$, $m$) and assigned from PL peaks determined in the characteristic $E_{22}$ excitation region [${\lambda }_{\mathrm{exc}}\sim 550–900\mathrm{nm}$, (not shown)] [17]. The $G$ and $G^{\prime }$ lines indicate Raman signals (weak narrow stripes) and relatively broad and intense PL peaks corresponding to ($E_{11}+G$) phonon-assisted excitation and $E_{11}$ emission. The PL peaks denoted as ${\mathrm{DE}}_1$ are attributed to emission from low-energy dark excitonic states excited via $E_{11}$. Arrows indicate correlations between PL peaks for (9,5) nanotubes.

Figure 3

Energy gap, $\Delta E$, between the first bright ($E_{11}$) and low-energy dark (${\mathrm{DE}}_1$) excitonic states for different ($n$, $m$) species as determined from the PL maps of HiPco and PLV nanotubes dispersed in ${\mathrm{D}}_2\mathrm{O}/\mathrm{SDS}$ (SDBS) and excited in the $E_{11}$ spectral region (see Figs. 2, 3). Experimental accuracy for $\Delta E$ is $\pm 5\mathrm{meV}$.

Figure 4

Schematic diagram of manifolds ($n=1$, 2) of excitonic states (horizontal lines) and electron-hole continuum bands (shaded areas) in semiconducting SWNTs. One-photon optically active excitons with formation energies $E_{11}$ and $E_{22}$ are denoted as ${\mathrm{OE}}_1$ and ${\mathrm{OE}}_2$. Weak emission satellites observed in this work are ascribed to dark states ${\mathrm{DE}}_2$ and ${\mathrm{DE}}_1$ photoexcited via ${\mathrm{OE}}_1$ or ${\mathrm{OE}}_2$ and lying $\sim 40$ and $\sim 100–140\mathrm{meV}$ below ${\mathrm{OE}}_1$, respectively. The presence of numerous other possible excitonic states (i.e., dark states not detected in this work) is indicated by dashed lines.