Optoelectronics with Carbon Nanotubes
Optoelectronics with Carbon Nanotubes Optoelectronics with Carbon Nanotubes
commonly available metals, so the Fermi level aligns with the valence band edge, allowing for holes to tunnel easily into the conduction channel. Thus, the alignment of bandgap with the Fermi level of the contact metal determines the type of conduction that takes place. If the Fermi level aligns midgap with the CNT, both types of carriers can be injected from opposite contacts, in which case the device is defined to be ambipolar. In ambipolar devices, the back gate can be biased positive or negative to switch the majority carrier. On the other hand, in a unipolar device, either the conduction or valence band edge aligns with the Fermi level of the metal, so the current is carried mostly by electrons or holes, respectively. For such a device, only one polarity of the back gate will allow significant conduction. Figure I-8. Schematic band structure of the first conduction (Ec) and valence (Ev) bands in the “on” and “off” states. The source-drain bias (VDS) is the same in both cases. In the on-state, applied gate voltage changes the thickness of the Schottky barrier, allowing p-type carriers (denoted “h” for holes) to tunnel through. Figure I-8 illustrates how a CNTFET operates as a p-type device. In the “off” state, the p-type carriers, i.e., holes, at Fermi level cannot tunnel through the Schottky barrier into the conduction channel. By changing the band bending with an applied gate voltage (VGS), the barrier becomes thin, allowing the holes to tunnel from the contact into the channel. As is apparent from Figure I-8, the effective Schottky barrier thickness is the critical property that determines the conductivity of the device. Appenzeller, et al. demonstrated that thermally- 19
assisted tunneling through Schottky barrier is the dominant injection mechanism by comparing a simulation and measurements using carbon nanotube and boron nitride nanotube FETs 69 . Later, detailed imaging of band bending at contacts was conducted via spatially-resolved photoconductivity on ambipolar SWNT devices 70 , confirming the presence of Schottky barriers and characterizing them at different gate voltages. In this experiment, absorbed photons create excitons across the bandgap, which are separated by the electric field present at the Schottky barriers and detected as current. Current intensity as a function of position yields spatially- resolved electric field strengths, i.e. band potential profile along the channel. Figure I-9 compares ambipolar and unipolar conduction to show that ambipolar devices (i.e., exhibiting both p-type and n-type conduction) can be created by using a metal with a work function that aligns at the mid-gap of the nanotube. Furthermore, to achieve sufficient tunneling rates, it is usually desirable to use a large-diameter SWNT, i.e., a smaller-bandgap SWNT for lower Schottky barriers. Applied gate voltage can shift the band energies up and down to make electrons or holes the majority carrier, leading to a “V” shape in the current versus gate sweep as seen in Figure I-7 (b). If electrons and holes are injected simultaneously, they could recombine across the bandgap and emit light. In principle, this process could be highly efficient because, as we have already seen, SWNTs have direct bandgap. However, in a conventional ambipolar CNTFET, there is little control over position-dependent carrier concentration within the channel, which limits our ability to improve efficiency. A position-specific electrostatic doping offered by a split-gate can overcome this limitation, which will be discussed briefly below, and is the topic of Chapter IV. Figure I-9 also shows that the highest electric field regions exist next to the contacts. If injected carriers are accelerated to a sufficiently high kinetic energy, an additional inelastic scattering mechanism comes into play, i.e., creation of excitons by impact excitation. This means that even in unipolar transport, radiative recombination of excitons can occur under certain conditions. In the next section, we give a brief history of EL including both emission mechanisms in order to provide some background information for our work. 20
- Page 1 and 2: Stony Brook University The official
- Page 3 and 4: Stony Brook University The Graduate
- Page 5 and 6: diodes nonetheless show a rectifyin
- Page 7 and 8: Table of Contents Abstract ........
- Page 9 and 10: Chapter I List of Figures Figure
- Page 11 and 12: Figure V-3 Source-drain electric
- Page 13 and 14: My parents have been steadfast supp
- Page 15 and 16: This work explores both fundamental
- Page 17 and 18: In fact, we typically have no knowl
- Page 19 and 20: (a) (b) Figure I-2 (a) Energy dispe
- Page 21 and 22: (a) (b) metallic Figure I-4. One-di
- Page 23 and 24: optical absorption peaks for differ
- Page 25 and 26: elax rapidly to lower non-radiative
- Page 27 and 28: The disorder-induced band (D-band)
- Page 29 and 30: The saturation limit for semiconduc
- Page 31: (a) (b) Figure I-7. (a) A schematic
- Page 35 and 36: majority carriers that gain enough
- Page 37 and 38: conventional ambipolar emission. Th
- Page 39 and 40: on quartz, which remains a signific
- Page 41 and 42: Chapter II Methods 1. Materials One
- Page 43 and 44: diameter (< 2 nm) SWNTs at IBM T. J
- Page 45 and 46: devices were annealed in vacuum at
- Page 47 and 48: 3. Experimental set-up The optical
- Page 49 and 50: off wavelengths of 2150 nm, 2000 nm
- Page 51 and 52: Chapter III Unipolar, High-Bias Emi
- Page 53 and 54: Figure III-1. Semi-log plot of drai
- Page 55 and 56: wetting with CNTs 57 and a relative
- Page 57 and 58: Figure III-4. (Main panel) Electrol
- Page 59 and 60: By plotting the current as a functi
- Page 61 and 62: phonon temperature in broadening. S
- Page 63 and 64: found multiple tubes bound together
- Page 65 and 66: where i is the phonon mode, Tsub is
- Page 67 and 68: The main panel of Figure III-10 sho
- Page 69 and 70: the effect following Perebeinos’
- Page 71 and 72: the optical phonon population is no
- Page 73 and 74: (a) Figure III-13. (a) Spectra from
- Page 75 and 76: DOP = I║ / (I┴ + I║) = 0.77.
- Page 77 and 78: inding energy for perpendicular exc
- Page 79 and 80: 3. Conclusions We have examined the
- Page 81 and 82: In a split-gate scheme, a new level
commonly available metals, so the Fermi level aligns <strong>with</strong> the valence band edge, allowing for<br />
holes to tunnel easily into the conduction channel.<br />
Thus, the alignment of bandgap <strong>with</strong> the Fermi level of the contact metal determines the<br />
type of conduction that takes place. If the Fermi level aligns midgap <strong>with</strong> the CNT, both types of<br />
carriers can be injected from opposite contacts, in which case the device is defined to be<br />
ambipolar. In ambipolar devices, the back gate can be biased positive or negative to switch the<br />
majority carrier. On the other hand, in a unipolar device, either the conduction or valence band<br />
edge aligns <strong>with</strong> the Fermi level of the metal, so the current is carried mostly by electrons or<br />
holes, respectively. For such a device, only one polarity of the back gate will allow significant<br />
conduction.<br />
Figure I-8. Schematic band structure of the first conduction (Ec) and valence (Ev)<br />
bands in the “on” and “off” states. The source-drain bias (VDS) is the same in both<br />
cases. In the on-state, applied gate voltage changes the thickness of the Schottky<br />
barrier, allowing p-type carriers (denoted “h” for holes) to tunnel through.<br />
Figure I-8 illustrates how a CNTFET operates as a p-type device. In the “off” state, the<br />
p-type carriers, i.e., holes, at Fermi level cannot tunnel through the Schottky barrier into the<br />
conduction channel. By changing the band bending <strong>with</strong> an applied gate voltage (VGS), the<br />
barrier becomes thin, allowing the holes to tunnel from the contact into the channel. As is<br />
apparent from Figure I-8, the effective Schottky barrier thickness is the critical property that<br />
determines the conductivity of the device. Appenzeller, et al. demonstrated that thermally-<br />
19