Carbon nanotubes as an electron source

This work characterizes the bulk emission properties of carbon nanotube (CNT) forest cathodes fabricated with varied geometries with the overall goal being to probe their viability as a potential replacement for thermionic cathodes currently used in accelerator systems. Geometries explored include dense nanotube forests of varying height grown on 5 mm x 5 mm silicon (Si) substrates and discrete, patterned CNT pillars fabricated using UV photolithography. Emission performance was gauged using both DC, DC IR, and pulsed IR testing. Properties explored include sample conditioning behavior, behavior throughout the IV curve, voltage turn-on levels, emission current output, emission current over time, emission area and location, self-heating behavior during pulsed and DC operation, and damage during testing. A parallel plate electron beam diode with a 100 µm A-K gap and an automated test apparatus were developed to provide a configurable experiment that gives accurate and repeatable measurements for DC, DC sweep, and timed performance testing. Testing has shown evidence of a hysteresis effect on the emission current tied to the applied field history as well as shifting of the turn-on field magnitude throughout the testing period, suggesting a conditioning effect during use. Three separate emission regions in the I-V curves during sweep testing have also been observed. In the geometric study, dense forest and patterned samples were sweep tested up to a peak applied voltage of -250 V, with the taller samples generally producing higher emission currents. Currents produced in the geometric study from the dense forest emitters ranged from 36.8 µA to 572.34 µA. Currents produced from the patterned micropillar emitters ranged from 39.4 µA to 317.51 µA. DC time testing showed a relatively stable output current over a 4.5-hour testing period. Another diode structure with a gridded anode was designed for investigating the self-heating characteristics of CNT emitters during electron emission using a mid-wave infrared (IR) camera. The custom diode setup incorporating a gridded anode was developed to enable infrared imaging of the emitter surface. Using this configuration, initial tests were conducted on two dense CNT forests and one patterned sample under DC voltage sweeps. The observed increases in IR intensity across the emitter surfaces confirmed the presence of self-heating during emission. Localized emission sites with elevated temperatures were observed across several samples of both geometries during DC sweep testing, suggesting low emission uniformity. Seven dense forest samples were tested in DC in a configuration where current measurements at the anode grid were correlated in time with increases in temperature on the emitter surface. Key parameters examined included emission area and location, the relationship between temperature and current density, total emission current, and current density. Analysis revealed a parabolic relationship between current density and temperature, with a linear trend observed in J2 vs. T, confirming the presence of Joule heating. These findings align with existing analytical models of CNT emission behavior. Infrared imaging of the emission surface revealed that nearly all emission hot spots were located at the edges of the CNT emitters, suggesting that electrostatic shielding from neighboring nanotubes is too strong in the center of the emitter. Emission area measurements ranged from 0.0085 mm2 to 0.034 mm2, showing that only a small fraction of the 25 mm2 surface is actually emitting. Peak current produced during the DC IR testing ranged from 4.69 µA to 147.21 µA. The corresponding current densities ranged from 58.57 mA/cm2 to 432.97 mA/cm2. Two dense forests samples were tested first in pulsed operation and then in DC operation in order to compare their behavior. Pulse testing showed that at the sample applied field, higher temperatures were produced at the emission sites in pulsed operation compared to DC. The relationship between current density and temperature observed in DC testing was used to estimate pulsed current density. A linear fit was performed on the J2 vs. T data collected during DC operation to produce an estimated J2 vs. T for pulsed. Estimated current densities for the pulse tests were higher than what was measured in the DC tests. Damage was observed in both DC and pulsed testing in the IR experiment. Damage was noted by a disappearance of the emission site in both DC and pulsed, with arcing being present during pulsed operation. Overall, 23 samples were tested, producing emission currents ranging from 4.69 µA to 2.41 mA. While emission currents measured in this work are below requirements for use in RF accelerator systems, measured current densities are promising. If a larger portion of the emitter can be made to emit, use in RF accelerator systems is plausible. Future work should focus on increasing the emission area of CNT cathodes to enhance emitted current, either by reducing forest density or through selective patterning of the cathode surface to mitigate electrostatic shielding effects.