Hammerhead: a compact 300 kV vacuum bushing Authors: Moein Borghei, Madeline Vorenkamp, Nicola van Moon, Robert McMullen, Robin Langtry, Brian Riordan Published: 2025 ABSTRACT High-voltage vacuum bushings are vital for particle accelerators, X-ray tubes, electron microscopes, fusion devices, and electron sources. Commercial bushings, however, are limited to ~ 100 kV, and exceeding this limit has proven to be difficult. This work reports the physics-informed design, simulation, and experimental validation of Hammerhead, a compact vacuum bushing that has been tested up to 330 kV and used reliably at 300 kV. The design utilizes a coaxial electrode configuration and a ceramic insulator that bridges the gap using a flashover-resistant geometry. Compared with the state-of-the-art, Hammerhead nearly doubles the voltage holdoff per unit of volume, does not require ultra-high vacuum conditions or sub-micron polishing, and can be manufactured in a moderate cleanroom environment. It also integrates directly with a high-voltage coaxial cable, without the need for insulating fluids or pressurized gas. Here, it is shown - over 144 h of experimental testing - that Hammerhead operates stably at 300 kV with dark current below 10 μA. Introduction High-voltage systems in vacuum are essential to a diverse array of technologies ranging from particle accelerators and fusion devices to X-ray tubes and electron microscopes. In many of these technologies, within vacuum, up to hundreds of kilovolts of potential are used to accelerate charged particle beams or to generate high-energy photons. The reliable access to such extreme voltages in a vacuum environment is known to be challenging, plagued by a lack of universal theory for failure mechanisms such as vacuum breakdown and surface flashover1,2. Unlike solids and fluids, the density of neutral particles within a vacuum environment is not high enough to enable an avalanche of ionization events. Therefore, traces of surface defects or contamination play a critical role in providing the means for the breakdown3. A high-voltage vacuum bushing (or feedthrough) is a component that allows electrical power to pass from ambient (or gas-insulated) environments into vacuum chambers. High-voltage vacuum systems have traditionally shown large dimensions to reduce the electric field intensity. Metal surfaces exposed to high electric fields emit electrons through quantum tunneling (field emission) which can lead to localized heating, X-ray emission, as well as electrical breakdown4. When an insulator is introduced, the problem becomes more complex as the insulator surface often has a lower breakdown strength relative to the breakdown strength of the vacuum gap and the bulk of the insulator5. It is commonly believed that flashover often initiates at the cathode triple junction, the junction between cathode, insulator and vacuum, where there is a significant field emission current6. When a flashover or vacuum breakdown is triggered, the resulting discharge quickly crosses the high-voltage gap. Design solutions for these problems place emphasis on the need to minimize field intensities and manage charge at crucial interfaces, which has commonly led to the use of wider, longer or multi-stage designs as well as intensive conditioning procedures. Vacuum bushings rated above 100 kV are exceptionally difficult to develop due to the aforementioned failure mechanisms. Market offering of high-voltage bushings tends to cap at ~100 kV which mandates a custom design at higher voltage needs. Figure 1 shows the cross-section of a commercially available 100 kV bushing. This design is not fully enclosed by a grounded structure. On the air side it employs a sawtooth pattern to increase creepage distance and mitigate flashover; on the vacuum side it relies on a long, slightly angled flashover path, which accounts for its substantial height of approximately 32 cm. The center conductor (cathode) has a diameter of about 4 mm, and the alumina ceramic insulator is brazed to a 15 cm-diameter ConFlat® flange. Both the anode and the cathode are typically fabricated from 304 stainless steel. Because the high-voltage portion includes external threaded connections, additional clearances and field-shaping electrodes are required, and a fully shielded cable cannot extend to the termination; the grounded shield must be cut back sufficiently far from the high-voltage connection point.Fig. 1Cross-sectional view of a commercial 100 kV high-voltage bushing.The r-z cross section of the commercially available 100 kV bushing (model EFT1C12156A, courtesy of Kurt J Lesker Co.; used with permission) is shown. Cross-sectional view of a commercial 100 kV high-voltage bushing. The r-z cross section of the commercially available 100 kV bushing (model EFT1C12156A, courtesy of Kurt J Lesker Co.; used with permission) is shown. Several examples of bushing designs for voltages exceeding 100 kV can be found in the literature7–11. Conventional methods have resorted to arrangements, including long ceramic standoffs or pressurizing the bushing with insulating gases such as sulfur hexafluoride (SF6), thus increasing size, complexity, and environmental impacts. A well-known example is the high-voltage vacuum bushing developed for the International Thermonuclear Experimental Reactor (ITER) neutral-beam injector which must provide 1 MV of insulation in vacuum. It is formed by five stages, using layered ceramic insulators and SF6 gas barriers to meet the voltage requirement; together with mechanical, cooling-channel and tritium-safety provisions, these features result in a meter-scale assembly with a large footprint7,9. After more than two decades since the initial work on the reference design12, full 1MV operation was reported in 201913. Although the ITER bushing is a specialized component, the theoretical and experimental advances made during its development offer valuable insights for next-generation vacuum bushing designs. Several development efforts have targeted the 300 kV regime. Notable examples include bushing development efforts for Jefferson Lab’s photogun and the University of Wisconsin-Madison’s inertial electrostatic confinement devices. Jefferson Lab demonstrated a compact photoemission gun using an inverted-insulator geometry capable of achieving a stable 300 kV11. This design features centrifugal barrel-polished electrodes, auxiliary electrodes for screening the triple junction, and a 10−10 Pa-scale vacuum. To reach the nominal design voltage, krypton gas conditioning was employed14. Since the construction of the first inertial electrostatic confinement (IEC) devices at the University of Wisconsin-Madison in 1994, the research group has pursued operation at voltages approaching 300 kV. The initial bushing concept was based on modifying a commercial 100 kV bushing by replacing the stock conductor and insulator with a custom assembly15. Subsequently, to overcome the limitations of that arrangement, a new insulator-conductor stalk design was developed that could reliably exceed 100 kV; this design demonstrated stable operation up to 150 kV. In 2009, the group pursued a further bushing redesign aimed at eliminating the pinhole failures that had plagued the earlier device, hoping to enable 300 kV operation16. Although the redesigned bushing achieved voltages above 200 kV, the 300 kV goal remained out of reach. This work presents a compact 300 kV vacuum bushing design, Hammerhead, originally developed to address the voltage requirements of the Orbitron, a compact crossed-field fusion device17. However, this technology is applicable to a wide range of applications requiring high-energy electrons, such as X-ray sources, particle accelerators, and electron microscopes. The design concept—previously proposed by the authors18—is validated through simulations, theoretical modeling, and 144 h of experimental data, demonstrating stable operation at the target voltage in a significantly more compact form factor. Hammerhead features a coaxial geometry and utilizes a ceramic insulator to bridge the interface between vacuum and atmospheric pressure. The center electrode (cathode) is at negative potential relative to the grounded outer electrode. The cathode geometry inherently shields the triple junction without the need for auxiliary screening electrodes. By directly interfacing the high-voltage cable, the bushing is able to operate without SF6 or oil insulation. Compared to Jefferson Lab’s 300 kV bushing, which is the most compact design in its voltage class reported in the literature-Hammerhead reduces the radial dimension by roughly 20% and the axial dimension by about 25%, while avoiding the need for pressurized gas, sub-micron polishing, ultra-high-vacuum conditions or high-grade cleanroom environments. Results Hammerhead design The overall design of the Hammerhead bushing is illustrated in Fig. 2. On the air side, the high-voltage connection to the cathode is made via a custom flat connector of 20cm outer diameter. The connector houses a coaxial X-ray cable insulated with ethylene-propylene rubber (commercially marketed as C2236). The cable is potted into the connector using a two-part H1369 epoxy resin obtained from Parker Medical Inc., which also fabricated the connector. The curing process is carried out in atmospheric pressure; the temperature is slowly ramped up to ~125 ∘C followed by a slow cooling to room temperature. A compressed silicone-rubber gasket between the flat faces of the connector and the ceramic insulator eliminates any air gap and ensures a reliable dielectric seal. The cable’s grounded shield is terminated at the entry to the connector assembly, while the semi-conductive layer extends about 1cm into the potting to reduce the electric stress at the interface.Fig. 2The design and assembly of the Hammerhead compact 300 kV bushing. a An r-z cross-section of the Hammerhead bushing. b The Hammerhead assembly on the vacuum chamber. The design and assembly of the Hammerhead compact 300 kV bushing. a An r-z cross-section of the Hammerhead bushing. b The Hammerhead assembly on the vacuum chamber. On the vacuum side, the bushing comprises a central molybdenum cathode (99.9% purity), a concentric insulator machined from MACOR® machinable glass-ceramic, and a concentric 304L stainless-steel anode with a 20 cm ConFlat flange. MACOR—which consists of approximately 55% fluorophlogopite mica and 45% borosilicate glass—is selected for its excellent vacuum compatibility, high dielectric strength, and ease of machining. Its machinability allows the fins, 10–19 mm high with radii and thicknesses of 1.3 mm and 2.5 mm, respectively, to be produced with conventional tooling. A knife-edge or O-ring groove is machined into the ceramic to seal against the stainless-steel flange. Additional design details are provided in ref. 19. To limit permittivity mismatches at interfaces, the constituent materials are chosen to have comparable relative permittivities: MACOR has ϵr ≈ 6, the epoxy ϵr ≈ 4, the silicone gasket ϵr ≈ 3, and the cable dielectric ϵr ≈ 2.8. Finite-element simulations confirm that the resulting electric-field stresses remain below the dielectric strength of each material. The remainder of this section discusses key failure and loss mechanisms and their impact on the design approach. Field electron emission Through a quantum-mechanical mechanism, electrons can escape from solid surface into vacuum when exposed to strong electric fields. When the field strength reaches a critical threshold, the electrons closest to the Fermi level penetrate the surface barrier toward vacuum space. Quantitative understanding of this phenomenon stems from formulation first presented by Fowler and Nordheim20, later improved by Murphy and Good21. It incorporates image-charge effects that reduce the surface potential barrier due to the Schottky-Nordheim effect22, resulting in a formula that approximates tunneling current density as below:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$J=\frac{A{F}^{2}}{\phi \,{t}^{2}(\,y)}\,{e}^{-\frac{B{\phi }^{3/2}v(\,y)}{F}}$$\end{document}J=AF2ϕt2(y)e−Bϕ3/2v(y)Fwhere A and B represent two universal constants which depend on factors that address image-force effects and barrier rounding23. F is the intensity of the local electric field at the emitter, and ϕ is the work function of the material. Finally, t(y) and v(y) are slowly varying functions of ϕ and F. t(y) is close to unity, but v(y) takes values slightly less than unity. The current density, J, needs to be multiplied by the emitting surface area to obtain the field emission current from an emission site. Emission from broad-area electrodes originates at discrete microscopic sites associated with surface irregularities or inclusions that locally enhance the electric field. Local field enhancement factor β describes the relationship between the local electric field at the microscopic feature, F, and the applied (macroscopic) electric, Fmacro, in the form of F = β Fmacro. A small number of high-β emitters provide most of the overall emission current24. It has also been demonstrated that field-enhancement features can transform over time due to Maxwell stresses and heat-activated processes25. Material properties such as work function and surface topography, together with mechanical quality, determine both the onset and intensity of electron emission. Previous studies indicate that refractory metals such as molybdenum exhibit high work functions and successfully reduce dark current at macroscopic electric fields near 100 MV/m26. Moreover, the high thermal stability of refractory metals confers enhanced resistance to the formation of new emission sites during testing. For these reasons, molybdenum is selected as the cathode material, with a geometry designed to keep the macroscopic electric field below 40 MV/m, which is about 60% lower than the switch-on field for molybdenum reported in ref. 26. Vacuum breakdown Although an ideal vacuum gap would have infinite dielectric strength, decades of research have shown that it can rapidly transition from an insulator to a conductive plasma through vacuum breakdown. In vacuum, which lacks an ionizable medium, breakdown is triggered when neutral and charged particles are released from the electrode surfaces27. Despite extensive study, the physics of vacuum breakdown has remained an elusive subject28. Two major mechanisms have been used to describe the initiation of the vacuum breakdown: (1) vaporization of field emission sites on the cathode, and (2) breakdown through microparticle impact (clump theory). In field-emission-based breakdown, current densities exceeding 108 A/cm2 can vaporize an asperity within nanoseconds; the vapor is then ionized and initiates Townsend avalanches, ultimately producing a conductive plasma29. Beyond electric field strength, mechanical and crystallographic properties also influence breakdown thresholds; experiments demonstrate that hexagonal close-packed metals have the greatest breakdown strength, followed by body-centered cubic and then face-centered cubic structures30. When the local electric field near an emission site reaches the GV/m level required for field emission, mechanical stresses can exceed tens of MPa, which may approach the threshold for dislocation movement. The interplay between surface defects, field-emission currents (causing Joule heating and Nottingham-effect-related thermal effects), and dislocation activity may further enhance the local field, establishing a feedback loop among mechanical, thermal, and electrostatic processes31. In clump theory, microparticles detach from an electrode and are accelerated across the gap by the electric field; their impact melts or vaporizes material on the opposing electrode, triggering plasma formation32. A contaminated vacuum environment along with system deterioration promotes this breakdown method. High-speed imaging and particle detectors have observed such clumps prior to breakdown33. Anode-initiated breakdown via microparticles tends to be more prevalent in unconditioned or contaminated gaps, where many loose particles or flakes are present34. In the Hammerhead design, molybdenum is chosen in part because its high melting point reduces the likelihood that metallic asperities will grow or vaporize. To tackle the breakdown mechanism due to microparticles and contaminants, this study employs a moderate clean room followed by glow-discharge cleaning and conditioning to prepare the surfaces for operation in vacuum (more details in the Theory section). Surface charging and flashover Surface flashover in vacuum is a complex discharge phenomenon. The mechanism occurs along an insulator’s surface and typically happens at a voltage below the breakdown of the solid dielectric and the vacuum gap. Two widespread theories suggest initiation at the cathode triple junction where the cathode, insulator and vacuum meet and result in a high localized electric field strength and emission current35. The leading theory is that secondary electron emission avalanche (SEEA), posits that initial electrons emitted from the cathode triple junction strike the insulator surface, releasing secondary electrons, which in turn knock out more electrons36. The cascading process continues until reaching the anode and in the final stage, it transitions into a gas phase breakdown in the desorbed vapor or ablated material37. The other prevalent theory, electron-triggered polarization relaxation (ETPR), suggests that injected electrons are trapped in the surface or bulk trap sites of the insulator, and once a critical density is reached, rapid detrapping (or a polarization relaxation wave) initiates the discharge38. No single mechanism comprehensively explains all observations—contemporary literature recognizes multiple, often concurrent, mechanisms depending on material and conditions. Nevertheless, the fundamental stages (field electron emission, surface electron multiplication, and eventual discharge) are consistently seen across a range of vacuum dielectrics. Triple junction field enhancement and surface micro-features play a dominant role in initiating vacuum flashover. The triple junctions often undergo a strong local electric field intensification due to the discontinuity in material permittivity and electrode geometry. In addition to mismatched relative permittivities, the reduced potential barrier due to the presence of the dielectric also contributes to elevated electron emission at the cathode triple junction. With this in mind, the Hammerhead’s cathode is designed to trap the triple junction into an equipotential well where the electric field is lower than the threshold needed for electron field emission. Additionally, the sawtooth pattern of the insulator assists with surface flashover prevention and mitigation in three ways: (1) it increases the flashover path between the two electrodes to over 400 mm, (2) forces any electron propagating along surface to move against or perpendicular to the direction of applied electric field thus reducing the flashover likelihood, and (3) as the next section shows, it can capture electrons and distort the local field in such a way that the electric field on the cathode is reduced. Lastly, similar to how field emission conditions the surface, flashover events can also condition the surface. When an insulator is subjected to repetitive flashover events at low to moderate energies, residual surface charge accumulates and weak emission sites (microscopic protrusions or contaminants) are burned off or removed, which ultimately improves flashover resistance39. Electro-quasistatic Simulation As a preliminary study of the projected electric stress on the bushing and the feasibility of the design, the Poisson’s equation are solved in COMSOL Multiphysics®. The 2-D axisymmetric model is discretized using approximately 7.1 × 105 triangular elements (~3.6 × 105 vertices) together with 15,768 edge elements, yielding an average element quality of 0.90 and a minimum quality of 0.018. The boundary mesh on the ceramic-metal interface contained 15,744 vertices and exhibited an average growth-rate quality of 0.97. A mesh-refinement study is carried out in the cathode triple junction region by successively doubling the local mesh density until the computed peak field changes by less than 1%. The electric potential and electric field distributions are shown in Fig. 3 for the cathode at the nominal voltage (−300 kV).Fig. 3Two-dimensional, axisymmetric, steady-state electro-quasistatic simulation of Hammerhead at -300 kV (r–z cross-section).a Electric potential distribution. b Electric field distribution. c Electric field near cathode triple junction. Source data are provided as a Source Data file. Two-dimensional, axisymmetric, steady-state electro-quasistatic simulation of Hammerhead at -300 kV (r–z cross-section). a Electric potential distribution. b Electric field distribution. c Electric field near cathode triple junction. Source data are provided as a Source Data file. Figure 3a shows the center conductor (the cathode) held at -300 kV with respect to the grounded outer boundary. As confirmed by Fig. 3b, the electric field is highest near the cathode surface, which is characteristic of the coaxial electrode configuration. The maximum electric field on the cathode occurs at its rounded top edges and reaches 36.7 MV/m, while the average electric field on the cathode is approximately half of that value. At 300 kV, the maximum electric field is close to the empirical breakdown limits32,40; however, the choice of molybdenum and the insulator charging behavior (discussed in the next subsection) help mitigate the risk for a well-conditioned cathode surface. Figure 3c shows the electric field distribution in the vicinity of the most critical spot, the triple junction, which undergoes less than 1.04 MV/m. This low electric field is achieved by trapping the triple junction in an equipotential well. The low triple junction field results in a reduced risk of surface flashover initiation at this point. However, the nearby cathode edge that faces the insulator sees the highest field stress, which still poses the risk of charging the insulator surface via its exposure to field-emitted electrons. The insulator fins generally experience lower electric fields compared to the cathode surface (0.1–22.4 MV/m) with the electric field decreasing radially outward. The countersink feature on the first fin acts as an additional defense feature against surface flashover through impeding electron propagation: electrons must climb uphill against the applied field direction, making surface electron avalanche less likely. Field-emitted electron trajectories In this section, a preliminary assessment of electron trajectories is performed in COMSOL Multiphysics. At an average electric field of 18.6 MV/m the cathode emits electrons toward the anode, and some of these electrons bombard the insulator surface. Figure 4a shows the electrons’ kinetic energy throughout the domain, originating from the cathode surface and eventually impacting either the anode or the insulator. Note that this simulation does not include surface charging, but as shown later in this section, it assists with providing a qualitative picture of how charging develops.Fig. 4Results from electron-tracing simulation.a Trajectories and kinetic energy of field-emitted electrons prior to insulator charging. b Normalized distribution of the kinetic energy of electrons impacting the dielectric (insulator) surface. Source data are provided as a Source Data file. Results from electron-tracing simulation. a Trajectories and kinetic energy of field-emitted electrons prior to insulator charging. b Normalized distribution of the kinetic energy of electrons impacting the dielectric (insulator) surface. Source data are provided as a Source Data file. The density of electrons born from the cathode surface follows the Murphy-Good formula (see Eq. (1)). No field enhancement factor is included given the lack of information on the magnitude or spatial distribution. Relativistic corrections are applied in the simulation, since at 300 keV the electron speed approaches a significant fraction of the speed of light. According to the simulation data, about 31.9% of these electrons emitted from the cathode strike the insulator’s surface. Since MACOR has high resistivity (>1016 Ω cm), each incident electron on the insulator surface contributes to the surface charge accumulation. To better understand the charging extent and polarity, a histogram on the count of electrons as a function of impact energy is illustrated in Fig. 4b. The minimum energy of striking electrons is 11.3 keV and the maximum 278.0 keV. The broad peak with the large width around 130 keV shows the large population of the electrons that impact the first fin’s surface facing the cathode. The next four narrow peaks at approximately 180 keV, 210 keV, 240 keV, and 270 keV correspond to the bombardment of the next four insulator fins farther away from the cathode, with higher electron energies. To understand the insulator charging extent due to electron bombardment, an illustrative secondary electron yield curve as a function of incident electron energy is displayed in Fig. 5. The secondary electron yield of MACOR for the aforementioned energy range (11–290 keV) has not been characterized experimentally yet. However, there are a significant number of reports showing that ceramic insulators have E1 < 100 eV, Emax ranging from 250 to 700eV and E2 ≈ 1–4 keV41.Fig. 5Secondary electron yield as a function of incident electron energy.Illustrative schematic showing a typical secondary electron yield curve as a function of incident electron energy. Secondary electron yield as a function of incident electron energy. Illustrative schematic showing a typical secondary electron yield curve as a function of incident electron energy. This suggests that electrons with an striking energy greater than 11 keV cause the impacted areas of the insulator to become negatively charged, since the quantity of electrons reaching the surface is greater than that of the secondary electrons leaving it. This negative charge slows down incident electrons until their energy is similar to that of E2, resulting in a state of equilibrium. Whether the electron flux is sufficient to reduce the energy to E2 cannot be predicted without knowledge of the emitter distribution on the cathode. However, post-experiment observations show that the insulator remains nearly intact, which suggests that the negative charging is not extreme enough to cause local field concentrations exceeding the dielectric strength. Furthermore, if the insulator near the cathode (i.e., the first fin) accumulates negative charge due to secondary electron emission, this negative potential will locally reduce the electric field strength at the cathode surface, thereby weakening field-emission current. Thus, it can be hypothesized that a state of equilibrium can be reached in which a slight degree of charging is maintained by a decreased level of field-emission current, resulting in enhanced flashover resistance. Experimental setup Figure 6 shows the experimental setup used for characterizing the Hammerhead bushing. The DC power supply, rated to 500 kV, is wired to the bushing assembly through an interconnect system consisting of a low inductance coaxial cable and a flat connector mated with ceramic insulator in an entirely grounded enclosure. The chamber is pumped and evacuated with turbomolecular pump (primed by a roughing pump) to achieve a base pressure below 10−6 Pa comprised primarily of residual H2, H2O, N2, and O2.Fig. 6An overview schematic of the experimental setup.The pumping system—including turbomolecular pump and roughing pump—is not shown. An overview schematic of the experimental setup. The pumping system—including turbomolecular pump and roughing pump—is not shown. The power supply has a voltmeter and ammeter with respective accuracies of ± 1 kV ± 0.5% of and ±10 μA ± 1% of reading. Fast measurements of the applied voltage are taken on the power supply side with a voltage probe with a 2.5% accuracy. The low-voltage output from the probe is connected to a digital oscilloscope for real-time waveform tracking with a 60 MHz bandwidth. For visual diagnostics, a high-speed camera (Chronos 2.1 by Kron Technologies) imaging at upwards of 5000 frames per second, is located at the bottom of the chamber to capture events on the insulator and the cathode through an optical window. The recording is triggered with anomalies in the voltage probe signal corresponding with conditioning events and vacuum arcs. Furthermore, a Geiger-Müller radiation detector is installed near the air side of the bushing for monitoring X-ray emission corresponding with field-emission currents, serving as a sensitive indicator of conditioning efficacy as well as consequences of vacuum breakdown events. A residual gas analyzer monitors the chamber pressure and the composition of the gas in real-time. During the gas conditioning procedure (described in the Theory section), 99.999%-pure krypton gas is fed into the chamber from a pressurized reservoir through a mass flow controller. At the end of the gas conditioning, the krypton flow is stopped, and the vacuum pumps return the chamber to high vacuum. The stored energy in the setup comprises the contributions from the power supply, the coaxial cable, the high-voltage probe, and the bushing assembly. At 300 kV the power supply stores approximately 167 J. The 6-m-long coaxial cable, with a capacitance of about 102 pF/m, yields a total capacitance of ~612 pF. The high-voltage probe and the bushing assembly add 20 pF and 28 pF, respectively, giving a combined capacitance of ~660 pF, which corresponds to ~29.7 J at 300 kV. As a result, the total stored energy within the setup at 300 kV is about 197 J. Experimental results The I–V behavior of the Hammerhead bushing under both gas and current conditioning is depicted in Fig. 7a, revealing important information about their effectiveness and the progress of the test as the applied voltage increases. The two solid lines show the mean cathode current as a function of voltage for gas-conditioned (blue) and current-conditioned (orange) operation. The shaded band around each line indicates the standard deviation of the measurements from the mean. In these experiments, gas conditioning is applied first to bring the system to a target voltage, followed by current conditioning at the same voltage set point. At voltages below 150 kV, both methods result in negligible dark current, indicating minimal field emission. However, at higher voltages, gas-conditioned samples tend to exhibit elevated dark-current levels. This increase is attributed to enhanced ion-induced secondary electron emission as described in the previous section. Furthermore, gas conditioning is intentionally conducted first to improve the cathode surface through sputtering and ion implantation, effectively removing high-β emission sites. As a result, by the time current conditioning is initiated, many of the most prominent emitters have already been passivated, leading to the generally lower emission activity during the latter phase.Fig. 7Results from the conditioning tests.a I–V characteristics during gas and current conditioning. The solid lines represent the mean cathode current versus voltage for gas-conditioned (blue) and current-conditioned (orange) operation, while the shaded bands denote ±1 standard deviation. b Time spent at each voltage range for both gas and current conditioning. Source data are provided as a Source Data file. Results from the conditioning tests. a I–V characteristics during gas and current conditioning. The solid lines represent the mean cathode current versus voltage for gas-conditioned (blue) and current-conditioned (orange) operation, while the shaded bands denote ±1 standard deviation. b Time spent at each voltage range for both gas and current conditioning. Source data are provided as a Source Data file. The conditioning procedure continues relatively smoothly until 165 kV, after which, longer wait times are needed to remove stubborn field emission sites. Figure 7b displays the time spent for each conditioning method over 15 kV voltage ranges between 30 kV and 330 kV. The gas conditioning period for 165–180 kV is about 8 h which is longer than entire period leading to the 165 kV operating point. Throughout the conditioning process, discharge events occur; they appear as fluctuations in voltage along with spikes in cathode dark current and X-ray emission. By triggering diagnostics such as fast-frame imaging on these events, insight into the underlying physics can be gained. Figure 8 shows a characteristic vacuum breakdown between the cathode and anode. There is a 200-350 ns delay between the external trigger and the moment the sensor begins exposure. The camera operates in circular-buffer mode and is configured to save footage whenever a rapid transient in the cathode voltage is detected. Since each frame integrates light over 200 μs at 5000 frames per second, the image captures not only the discharge initiation but also its subsequent evolution.Fig. 8Fast-frame still of a characteristic vacuum breakdown between the cathode and anode.The camera’s 200μs integration time at 5000 frames per second mean the image captures both the initiation and the evolution of the discharge. Fast-frame still of a characteristic vacuum breakdown between the cathode and anode. The camera’s 200μs integration time at 5000 frames per second mean the image captures both the initiation and the evolution of the discharge. Occasional de-conditioning events are observed during the process. They appear as spikes in current and X-ray emission followed by a more stable increase in dark current and X-ray level. While most of these events are relatively benign, extreme cases can require tens of hours of conditioning to revert the damage. An example of this can be seen on Day 3 in Fig. 9. This figure displays the voltage and current ranges of experiments for each of the 23 days of testing. The depth of color for each point denotes the cumulative hours spent conditioning at that voltage; darker points indicate longer conditioning durations. On Day 3, a local maximum in voltage is observed; it takes nearly six days of conditioning to return to that voltage again on Day 9. It is worth noting that the current on Day 9 is less than that on Day 3, indicating that the cathode is more thoroughly conditioned after recovering from the de-conditioning event than it was prior. These de-conditioning events are hypothesized to stem from two main mechanisms: (1) explosive vaporization of field emitters on the cathode surface, and (2) cathodic contamination via sputtered particulates-metallic or insulating-that are dislodged during breakdown and transported across the vacuum gap. Upon redeposition, these particulates can generate new high-field enhancement sites (high β) or exacerbate existing ones, leading to renewed field emission. In either case, the result is a temporary degradation in conditioning efficacy and a setback in voltage holdoff, necessitating further emitter suppression and surface reconditioning to restore stability.Fig. 9Cathode voltage and current over the course of the experiment.Each data point is plotted with low opacity; overlapping points appear progressively darker, indicating more time spent conditioning at that voltage. Source data are provided as a Source Data file. Cathode voltage and current over the course of the experiment. Each data point is plotted with low opacity; overlapping points appear progressively darker, indicating more time spent conditioning at that voltage. Source data are provided as a Source Data file. Between Days 9 and 13, conditioning proceeds nominally with no major de-conditioning events, and an average dark current of approximately 80 μA until 300 kV. While no major setback is observed in this period, the dark current increases from 40 μA at 250 kV on Day 11 to 120 μA at 270 kV on the next day. Upon reaching the target voltage of 300 kV, transitioning from gas conditioning to current conditioning proves challenging. On Day 14, the system reaches 300 kV under krypton backflow, but the same voltage cannot be maintained when the system is returned to high vacuum without krypton. This observation indicates that the effect of gas conditioning is more complex than simply being a tool for removing tough emitters. A plausible explanation is that the increase in background pressure due to krypton backflow enhances scattering of electrons and ions as a result of shorter mean free path, thereby reducing the probability that they traverse the gap with energies near the full potential. Reversible removal of implanted krypton upon returning to high vacuum is unlikely, as such desorption typically requires bake-out temperatures that are not reached during these tests. Next, the krypton backflow is reintroduced and the voltage is raised to 330 kV, 10% above the nominal operating voltage. The cathode is held at 330 kV under gas conditioning for 1 h (Day 19), before the voltage is then reduced to ~250 kV. From there, current conditioning resumes and proceeds for the subsequent two days until the cathode is stable at 300 kV. Hammerhead’s cathode is then held stably at 300 kV for over 2 h under high vacuum, with dark current remaining below the ±10 μA resolution of the measurement system. The average dark current during this period is approximately 8 μA, indicating a successful recovery and stabilization. The success of krypton conditioning at higher voltage can be attributed to two effects: (1) the longer conditioning duration allows sputtering of stubborn emission sites until the emitter flux becomes too low to initiate breakdown, and (2) the higher energy of Kr+ ions increases the likelihood of implantation, leading to greater emitter passivation through an increased work function. Post-experiment microscopy and surface analysis Following conditioning completion, the cathode is inspected with Optical Imaging, Optical and Laser Profilometry, Scanning Electron Microscopy (SEM), as well as Energy Dispersive Spectroscopy (EDS). Fig. 10 displays a characteristic clump of a copper contaminant that has impacted the cathode. It is observed that copper (Cu) and silver (Ag) particulates are embedded in the cathode surface. It is hypothesized that during the disassembly and reassembly procedures, contaminants enter the chamber, originating from the bushing’s Cu seals and Ag-plated fasteners. These contaminants are likely then clumped onto the cathode as dictated by Clump Theory32, in which microscopic particles or clumps become dislodged from one electrode and traverse the vacuum gap, leading to a disruptive discharge upon impact on the opposite electrode32,42.Fig. 10A series of SEM images of a characteristic Cu contaminant.a SEM image. b Composite EDS map showing concentration of molybdenum (red) and copper (green). c EDS map of molybdenum. d EDS map of copper. A series of SEM images of a characteristic Cu contaminant. a SEM image. b Composite EDS map showing concentration of molybdenum (red) and copper (green). c EDS map of molybdenum. d EDS map of copper. Post-experiment surface analysis shows that clumps of different materials (e.g., Au and Cu) are clustered at discrete sites, suggesting either pre-aggregation before acceleration or electrostatic focusing to specific impact points by local field enhancements (see refs. 43–46 for discussions of field-enhancement features). As shown in ref. 7, the trajectory of a charged particle released from rest in a static electrostatic field is governed by the field geometry (with the direction set by the sign of the charge), and thus particles follow field lines. In contrast, if some clumps are ejected with nonzero initial velocities or possess different charge-to-mass ratios (q/m), inertial effects can modify their paths; combined with electrostatic focusing, this can produce the observed clustering. Further analysis of these hypotheses will be presented in a future publication. Discussion The presented experimental study validated the design and simulation of a compact 300 kV bushing, Hammerhead. The presented bushing is approximately 20% smaller in diameter and 25% shorter in height compared to the most compact 300 kV bushing previously reported11. Over a total of 144 h of testing, the time required for effective conditioning is found to vary significantly with cathode voltage and prior surface condition. Notably, some of the most persistent surface irregularities—such as those arising from the breakdown on Day 3—do not occur at the highest voltages but still required a significant recovery period. This emphasizes that emitter’s stubbornness, not just field strength, governs reconditioning effort. Additionally, the effect of krypton gas on voltage holdoff seems to be somewhat reversible: once the gas is removed, some of the field emission sites re-activate and reduce the voltage holdoff. However, extending gas conditioning with krypton to 110% of the nominal voltage appears as a practical and effective approach that works through deeper ion implantation and more robust emitter suppression. The presence of copper and silver particulates on the cathode surface helps to explain the observed stagnation in the conditioning process. As these particulates impact the cathode, they likely result in the addition of sharp emission sites which would require additional time to process off. In order to prevent such delays moving forward, the clean room technology and surface cleanliness can be improved to prevent particulates from entering the chamber in the first place. It is generally understood that low-energy systems condition upward, whereas high-energy systems condition downward: the former lack sufficient energy to sputter appreciable material into the vacuum assembly or to cause substantial damage, while the latter can produce sputtering and damage. Given the upward trend of the conditioning procedure, the ~195 J of stored energy appears sufficient without being excessive. Moreover, with the stable dark current below 10 μA at the nominal voltage, the total power dissipation is less than 3 W; therefore, the temperature throughout the anode-cathode gap should remain essentially uniform during standard operation. The principal contributions of this work can be summarized as follows: (1) demonstration of stable 300 kV operation with <10 μA in a bushing of 20 cm radial and 25 cm axial length, using only solid-state insulation and vacuum-compatible materials, (2) implementation of a field-suppressed triple junction design, verified through electrostatic and electron tracing simulation to limit local field intensities to sub-emission levels, and (3) verifying stability through a dual-phase conditioning method to achieve emitter suppression and surface passivation, followed by observation of conditioning traces on the electrode surface via surface analytical techniques. The demonstrated increase in voltage-holdoff per unit volume represents a significant step toward the miniaturization of high-voltage infrastructure, with implications for a wide range of scientific and industrial applications where space, reliability, and environmental compatibility are critical. Future investigation will focus on (1) quantifying the long-term stability and fatigue behavior of conditioned surfaces under extended high-voltage operation, (2) modifying the cathode structure to accommodate multi-conductor geometries that may be useful for electron sources, and (3) improving electrode surface conditions and clean room technology to reduce the conditioning time, and (4) evaluating Hammerhead with alternative ceramics such as Al2O3 which is currently under fabrication. Methods Cleaning procedure The cathode and insulator are manufactured from bulk materials using computer numerical control (CNC) machining. The cathode is polished to a surface roughness equal to or less than 1 μm through manual sanding using 2000 grit sandpaper. The cathode material is molybdenum (Type 361) with a purity of 99.9%, the insulator is manufactured from MACOR ® Machinable Glass Ceramic, and the anode is a commercially available vacuum chamber constructed of 304L Stainless Steel. Following fabrication, the samples are subjected to a cleaning process to remove hydrocarbons and other contaminants that remain from the manufacturing and handling processes. Initially, the parts are hand-washed with soap and water to remove bulk contaminants. Next, each sample is subjected to a series of ultrasonic cleaning baths. The first cycle is 15-min in a mixture of detergent and heated deionized water. Following this cycle, each sample is rinsed with deionized water before being transferred to the second cycle, 2-min in a deionized water bath for a second ultrasonic cleansing. Next, the sample is rinsed again before entering a third cycle, this time 15-min in isopropyl alcohol. Lastly, each sample is brought to a class-10,000 cleanroom for hand drying with a lint-free cloth. After installing the Hammerhead bushing on the vacuum chamber and achieving a 5 × 10−6 Pa pressure, the entire vacuum chamber is the heated gradually to 80 ∘C over the course of 8 h, held at temperature to be baked out for 24 h, and then gradually cooled down over an additional 8 h. The vacuum pressure after bake-out is settled at 4 × 10−6 Pa. Prior to testing, glow discharge conditioning is implemented as one final cleaning measure for the bushing’s electrodes. Ar/O2 gas mixture with 9/1 ratio flows into the chamber according to the procedure described in ref. 47. By reversing the polarity of the power supply, it is possible for both electrodes to be exposed to ions. This is done with a target current density of 25 μA/cm2. The gas mixture cleans the surface employing several fundamental mechanisms. Ar leverages sputtering to remove impurities from the surfaces, while the addition of O2 facilitates surface reactions producing CO and CO2, which accelerates the removal of surface carbon from the chamber48. With a total pressure within the 13–133 Pa range and applied voltage magnitude between 300 and 600 V, the targeted current density was achieved for the outer and center electrodes with 20 mA and 1.5 mA, respectively. Current conditioning Despite the careful design of Hammerhead, the compact geometry of the bushing leads to an electric field strength on the cathode surface that is sufficiently strong for field emission. These electrons can bombard the insulator and the anode eventually trigger breakdown phenomena discussed, under certain conditions. In the high voltage testing, two phases of controlled conditioning discharges are used to remove surface irregularities and field emitters, current conditioning and gas conditioning. In total, the two methods are implemented over 144 h to achieve stable operation at 300 kV in high vacuum. The leading method of conditioning, current conditioning (also known as spot-knocking conditioning), involves raising the applied voltage in small increments and holding it until no discharge or current/X-ray spike is observed. The mechanism by which current conditioning improves the voltage hold-off is due to the burn-off of the emitter through Joule heating of the field emission sites with high current densities49. While effective, current conditioning can often be time-consuming and sometimes unreliable. One alternative, gas conditioning can more efficiently burn off the emitters, and is heavily employed in this work. Gas conditioning In the gas conditioning method, an inert gas is introduced to the chamber at relatively low voltages; the gas is then maintained as the voltage is ramped up. The voltage is raised in 2 kV increments and held until the cathode current and X-ray emission stabilize before the voltage is increased again. During this process, some of the gas molecules are ionized by the field-emitted electrons from the cathode. These newly formed ions improve voltage hold-off through two mechanisms: sputtering and ion implantation. In the case of sputtering, the newly formed ions are attracted to the field-emission sites (often metallic protrusions or insulating particulates) because of the high local electric field (high-β sites). As the ions impact these sites, they blunt the protrusions; this process is known as sputtering. In the other mode, ions are implanted into the cathode surface, ultimately increasing the work function of the metal and thereby reducing the likelihood of electron tunneling from the cathode, as shown in Eq. (1). In addition to sputtering and implantation, the ionization of krypton molecules by electron impact and their subsequent collision with the cathode can liberate more than one electron per incident ion, depending on the impact energy. This secondary emission can increase the dark current. The interplay between the raised work function and the enhanced secondary emission creates a situation in which, depending on the operating parameters, the krypton backflow may either increase or decrease the total dark current. In this work, krypton (Kr) is used while various other gases, including He, Ar, N2, O2, and H2, have also been used for gas conditioning in the literature50. The flow is introduced at a rate ranging between 8.33 × 10−9 m3/s and 1.67 × 10−7 m3/s, which at the highest, raises the background pressure to ~0.15 Pa. The benefits of using Kr for gas conditioning have been discussed in the works done at Jefferson Lab11,51. In the case of Hammerhead, heavy Kr+ ions bombarding the cathode have a similar atomic mass to molybdenum (84 vs. 96 amu), so at energies above ~ 1 keV a single ion can eject multiple Mo atoms. This efficient sputtering rounds off micro-tips and smooths nanoscale asperities-features that concentrate field emission. Lighter ions such as He+ have smaller sputtering effect. Krypton ions that do not sputter away are implanted within the upper tens of nanometers of the cathode (e.g., a 300 keV Kr+ ion has a projected range of ~ 63 nm), and these implanted noble gases can raise the local work function on the order of 1 eV51. The combination of controlled sputtering and shallow implantation reduces tunneling currents and suppresses field emission without significant bulk modification, although excessive ion bombardment can roughen the surface. Gas and current conditioning serve complementary roles. Introducing a low-pressure krypton backfill allows field-emitted electrons to ionize krypton atoms; the resulting ions sputter sharp asperities and implant into the cathode, raising its work function and passivating high-β emission sites. Once these prominent emitters have been suppressed, current conditioning in high vacuum can more effectively blunt the remaining sites through Joule heating and provide a smooth transition to steady high-vacuum operation at the nominal voltage. Supplementary information Description of Additional Supplementary Files Supplementary Software Transparent Peer Review file Description of Additional Supplementary Files Supplementary Software Transparent Peer Review file Source data Source Data Source Data