Skip to content

Demonstration of a sheet electron beam production from a UNCD field emitter array

FieldValue
JournalOpen MIND
Published2026-01-29
AuthorsWonjin Choi, Dongsung Kim, Heather Andrews, Chun-Chieh Chang, Muhammed Zuboraj
DOI10.18429/jacow-napac2025-wep016

A nitrogen-doped ultra-nanocrystalline diamond array of 1 by 81 pyramids produced a sheet electron beam at 20 kV. The measured divergence angles were 0.23° in the x direction and 1.2° in the y direction. Beam current ranged from 5.5 µA to 7.3 µA during the test.

The cathode used 81 pyramids with a 25 µm base and a 50 µm pitch. The array extended along x, and the beam image was fitted in both transverse directions.

The authors selected this geometry because sheet beams can carry more current than conventional pencil beams. A high-aspect-ratio sheet can also interact efficiently with planar structures. These are design expectations from the introduction, not a current comparison reported in this experiment.

The measured current varied by about ±14%, and the authors judged space-charge effects on electron trajectories to be negligible. The test established sheet-beam production at low current, without establishing high-current performance.

The authors cite mechanical and chemical stability, high thermal conductivity, and low electrical resistivity as reasons to use UNCD for field emission. Its very small grain size permits emission from a flat deposited surface.

Pyramidal emitters also help characterize the emission area and emission properties. The introduction says that the pyramid shape can reduce turn-on voltage.

Unlike thermionic and photocathodes, field-emission cold cathodes do not require heating or laser systems. The paper cites compact sources for table-top accelerators, dielectric laser accelerators, and terahertz vacuum tubes.

The introduction states that diamond field emitter arrays can operate under poor vacuum above 10^-5 Torr. Their resistance to contamination and low outgassing were given as reasons. The reported experiment operated at 10^-7 Torr or lower, so that cited capability was not tested here.

Fabrication began with dry oxidation of a silicon wafer. Electron-beam lithography wrote the desired array pattern on the wafer using a multipass method. Inductively coupled plasma etching removed the developed SiO2. Tetramethylammonium hydroxide then etched the exposed silicon anisotropically, forming inverted pyramids.

A UNCD layer was deposited over the silicon mold by chemical vapor deposition. The diamond layer was brazed onto a molybdenum substrate before the silicon and SiO2 were etched away. The finished cathode was cleaned and inspected by scanning electron microscopy.

Most pyramids had one sharp tip, although some tips were blunt or formed double tips. One blunt tip had a diameter of approximately 40 nm. The extracted beam showed nonuniform emission.

The test stand operated in a chamber at 10^-7 Torr or lower. Its power supply could apply more than 40 kV, but the reported run used a -20 kV cathode. The cathode was mounted on a fixed holder.

The grid and screen sat on independent motorized stages. The cathode-to-grid gap was approximately 4 mm. The grid-to-screen gap was varied from 6.5 mm to 35 mm. The grid used 20 µm diameter wires with a 254 µm pitch.

The screen consisted of aluminum-doped zinc oxide deposited on sapphire. A camera outside the chamber recorded the beam image. Beam current was inferred from the voltage across a 33 kΩ resistor. This arrangement measured current, transverse beam growth, and image shape over several screen distances.

Six images were averaged at each grid-to-screen distance to reduce flickering pixels. The authors fitted each averaged image with a two-dimensional Gaussian. The fit described horizontal and vertical spreads, tilt, center, offset, and amplitude. Least-square slopes of the fitted spreads gave the reported divergence angles. The result was 0.23° in x and 1.2° in y.

Grid wires appeared brighter than the surrounding beam instead of casting shadows. The authors suspect that secondary electrons from the wires added intensity. They also propose that local electric fields around the wires affected the image. They proposed a grid with a different material for a later test.

Screen colors changed near beam edges from 8 mm onward. The authors attributed this to damage from earlier tests, which produced brighter pixels. They did not consider the damage a major problem for the analysis. They plan to repeat the test with a new, undamaged screen.

The smaller x divergence was consistent with the array being extended only in x. The authors note that higher divergence in the narrow dimension is typical of a space-charge-dominated sheet beam. They state that space-charge dominance did not apply to this result.

Their statistical estimate used RMS beam size, array pitch, and tip count to predict directional divergence. With b = 0.6 mm, M = 81, and p = 50 µm, the estimated y-to-x divergence ratio was 2.19. The model predicted a larger y divergence, consistent with the measured ordering.

Large fluctuations in fitted x spread indicated relatively high fit errors. Nonuniform emission, bright grid-wire features, and different fit tilts caused the fluctuations.

The authors compared the images with a CST Particle Studio simulation of the cathode and beam transport. Unstructured tetrahedral meshes resolved features spanning nanometers to millimeters. The simulated emitting tips had a diameter of 100 nm.

The analysis applied the same two-dimensional Gaussian fits to simulated particle positions. Maximum particle positions produced divergence angles of 5.4° in x and 2.9° in y. RMS particle positions produced 1.1° in x and 1.5° in y. The experimental pair was 0.23° in x and 1.2° in y.

Neither particle-position measure reproduced the experimental pair exactly. The maximum-position result also reversed the measured ordering of the two directions. The comparison shows that the position statistic changes the inferred divergence.

The authors identify two unresolved limitations: nonuniform emission and grid patterns. The origin of the grid pattern requires further investigation. Improved fabrication and cathode conditioning may reduce nonuniform emission. The Gaussian fit errors were high enough to produce large fluctuations in fitted x spread.

The present cathode was used only for low-current testing. The next planned work is to measure emittance in both transverse directions with a pepper-pot method. High-current evaluation will follow that measurement. A new screen will remove the earlier-test damage from the next image set. These steps will test whether the sheet remains uniform as current rises.

The study supports UNCD field-emitter arrays as a route to shaped electron sources for planar structures and high-frequency vacuum tubes. For device selection, the demonstrated operating point was 20 kV, with low-microampere current and anisotropic divergence. The paper does not establish high-current operation, low divergence in both directions, or stable grid-free imaging. Those properties must be established before a sheet source can be engineered for an accelerator application.

View Original Abstract

Ultra nanocrystalline diamond (UNCD) is a promising material for field emitters because of its mechanical and chemical stability, high thermal conductivity, and low electrical resistivity. We proposed to demonstrate fabrication of a special shape field emitter array to produce a sheet electron beam for high frequency vacuum tubes. At Los Alamos, we established a Field Emitter Array Test Stand (FEATS) where we can apply voltages up to 40 kV to test field emitter arrays in a vacuum level of 10^-7 Torr or lower. At this test stand, we can take beam images, measure beam current and study beam divergence. We fabricated diamond cathodes in form of arrays of 1 by 81 pyramids and used them to demonstrate production of a sheet electron beam. This talk will present details of the emission tests and analyses of the produced sheet beam.