Micro Hollow Cathode Discharge (MHCD)

• Micro Hollow Cathode Discharge (MHCD) is defined as a microplasma source formed in a layered electrode–dielectric–electrode stack with a microhole that localizes glow discharges in gases such as helium and argon.
• Device performance is controlled by structural parameters like cavity geometry, cathode area, and electrode material, which affect ignition, breakdown voltage, and discharge regimes.
• Experimental and simulation studies reveal that MHCD dynamics—including steady operation, hysteresis, self‐pulsing, and metastable behavior—are key to advancing plasma processing and integrated array applications.

Searching arXiv for the cited MHCD papers to ground the article in the provided literature. Micro Hollow Cathode Discharge (MHCD) denotes a class of microplasma sources built around a small cavity or hole through an electrode–dielectric–electrode stack, typically operated in direct current or pulsed regimes and capable of sustaining localized glow discharges in helium, argon, or reactive mixtures at pressures extending to atmospheric pressure (Kulsreshath et al., 2016). In the literature represented here, MHCDs are realized most commonly in sandwich structures such as Ni/Al$_2$O$_3$/Ni or Mo/Al$_2$O$_3$/Mo, with characteristic hole diameters from tens to a few hundreds of micrometers, and their behavior is governed by a combination of cathode-fall physics, confinement, wall losses, plasma spreading, non-local ionization, and external-circuit coupling (Dufour et al., 2019). A central theme across the field is that “micro-hollow” geometry does not by itself imply a classical hollow-cathode effect; rather, steady operation, ignition thresholds, regime transitions, and array compatibility are strongly controlled by the available cathode area, sheath structure, and discharge dynamics (Dufour et al., 2016).

1. Definition, geometry, and relation to hollow-cathode physics

MHCDs are described as very small nonequilibrium dc glow discharges formed in layered structures with a microhole. Representative realizations include a Ni/Al$_2$O$_3$/Ni sandwich with holes typically 250 µm in diameter, Al$_2$O$_3$ thickness up to 250 µm, and Ni thickness about 6 µm, operated in helium over 100–1000 Torr using a 0–2500 V dc supply (Dufour et al., 2019). Closely related devices include Ni : Al$_2$O$_3$ : Ni stacks with a 250 µm dielectric and 5–8 µm nickel electrodes, with laser-drilled holes in the 135–400 µm range, as well as Mo/Al$_3$0O$_3$1/Mo or Mo/AlN/Mo tri-layers with 100 µm metal thickness and 400 µm or 250 µm holes (Dufour et al., 2016).

This family of devices shares defining structural traits: two electrodes separated by a dielectric, a through-cavity that localizes the discharge, and operation in gases such as helium, argon, and reactive mixtures. Silicon-integrated cathode boundary layer devices are presented as operationally MHCD-like variants, using a silicon substrate as cathode, a nickel film as anode, a 5 μm SiO$_3$2 dielectric, and etched cavities of 50 μm or 100 μm diameter (Dussart et al., 2016). A plausible implication is that the MHCD concept is best treated as a microcavity glow-discharge platform rather than a single fixed geometry.

The relation between MHCD structure and the classical hollow-cathode effect is treated with care. In a single-cavity microdevice with 90 $_3$3m and 190 $_3$4m openings on opposite sides and cavity length up to 400 $_3$5m, the measured Paschen curves for both polarities coincide over 10–400 Torr, and breakdown depends on the interelectrode distance, not on cavity diameter (Dufour et al., 2016). Because the cathode is only 8 $_3$6m thick in that study, the discharge is concluded not to be governed by classical pendulum-electron hollow-cathode breakdown. This distinction is important: MHCDs may be micro-hollow in architecture while not exhibiting the classical hollow-cathode effect in ignition.

At the same time, hollow-cathode theory remains relevant. The non-local ionization framework for glow discharge and hollow cathode treats the cathode as an interior singular source and formalizes ionization as an integro-differential avalanche process,

$_3$7

with incomplete wall absorption represented by a reflection factor $_3$8, $_3$9 (Gorin, 2012). This suggests that even when classical pendulum-electron breakdown is absent, MHCD operation remains deeply linked to cathode-emitted electrons, confinement, and wall-mediated non-local multiplication.

2. Steady-state electrical regimes and the role of cathode area

A recurring result is that single-hole MHCDs in helium exhibit a breakdown voltage followed by lower-voltage sustaining operation, hysteresis, and a normal-glow regime in which the discharge voltage stays approximately constant while current increases. For a single 300 µm hole in helium at 400 Torr, three regions are identified in the $_2$0–$_2$1 characteristic: pre-breakdown, normal glow, and self-pulsing regime (Dufour et al., 2016). At 600 Torr He, the breakdown voltage is about 315 V for 130 µm and about 300 V for 300 µm, whereas the operating voltage is about 180 V and 170 V, respectively. Between about 2 and 15 mA, the $_2$2–$_2$3 curve has a slightly negative slope characteristic of a normal glow discharge.

The physical interpretation given for the normal glow regime is that, as current rises, the plasma is no longer confined to the cavity and spreads over the cathode surface outside the hollow; this increases the active cathode area available for ion bombardment and secondary-electron emission, so additional current can be sustained without a large voltage increase (Dufour et al., 2019). The qualitative scaling is stated explicitly: $_2$4 so $_2$5.

The central control parameter in several studies is the exposed cathode area. When the cathode is covered by a dielectric limiting layer, such as 50 µm Kapton with a circular opening of 500 µm to 2500 µm, or 60 µm tape with $_2$6 or 500 $_2$7m, the discharge can no longer spread freely (Dufour et al., 2019). Under those conditions the plasma is forced into an abnormal glow regime with positive differential resistance and operating voltage rising with current. For a single 250 µm hole with a 600 µm circular opening, the discharge voltage $_2$8 increases with current at 250 Torr and 750 Torr; at atmospheric pressure, the smallest opening gives the strongest positive slope, while larger openings approach constant-voltage normal-glow behavior (Dufour et al., 2019).

A detailed experimental and simulation study of a single microcavity device in helium reaches the same conclusion. With unlimited cathode area at 100 Torr He, the discharge enters a normal glow with voltage nearly constant at about 175 V while current rises from about 4 mA to 28 mA; with a limiting layer of $_2$9, the discharge ignites at about 3 mA, rises to about 9.5 mA, and the voltage increases up to about 380 V, giving a positively sloped $_3$0–$_3$1 curve (Dufour et al., 2016). The authors state that the key route to steady abnormal glow is restriction of cathode surface area rather than cavity diameter itself.

Hole diameter also affects sustaining voltage. Comparing 130 µm and 300 µm holes, the larger diameter hole operates at a smaller voltage, while the smaller hole requires a higher voltage (Dufour et al., 2016). This is interpreted through a diffusion-based normal-glow picture with

$_3$2

where $_3$3 is the ambipolar diffusion coefficient and $_3$4 the ionization rate coefficient. Smaller holes imply stronger confinement, higher wall losses, and the need for a larger electric field to maintain ionization balance.

3. Ignition, extinction, hysteresis, and self-pulsing dynamics

MHCD ignition is not a single-threshold phenomenon. A slow-ramped-voltage study of helium MHCDs using a 250 μm alumina dielectric, 8 μm nickel electrodes, and laser-drilled holes of 300–360 μm reports standard breakdown followed by a sudden voltage drop and a transition into a normal glow regime (Kulsreshath et al., 2016). On ignition, a 2 μs current pulse as high as 24 mA is observed in the abstract; in specific measurements the peak current is about 15 mA at 350 Torr and about 25 mA at 750 Torr, with the peak occurring in about 300 ns and the pulse duration about 2 μs FWHM (Kulsreshath et al., 2016). The discharge is interpreted as an RC-like structure with effective capacitance $_3$5; with ballast resistor $_3$6, the characteristic time is

$_3$7

consistent with the observed transient.

Hysteresis between breakdown and sustaining operation is reported repeatedly. In helium parallel microdischarges, a hysteresis cycle between Townsend and normal glow regimes is explicitly identified, reflecting the fact that the voltage required to ignite the plasma is significantly higher than the voltage required to sustain it (Dufour et al., 2016). In cathode-limited helium MHCDs, hysteresis is also observed, and the reduction of the voltage sweep period from 40 s to <1 s makes the hysteresis much weaker, which is taken to support a thermal contribution; long-lived helium metastables are also noted as a possible influence (Dufour et al., 2019).

The extinction phase depends strongly on pressure. At 350 Torr, current decreases smoothly from a few mA to about 60 μA and then drops to zero without oscillations; the inferred plasma impedance is about 1.6 MΩ, much larger than the ballast resistance, so the effective RC time is around 100 μs (Kulsreshath et al., 2016). At 750 Torr, by contrast, a clear oscillatory regime appears a few tens of microseconds before extinction, with current and voltage oscillating together with period about 14 μs. The plasma impedance near the beginning of this sequence is about 360 kΩ, giving

$_3$8

slightly larger than but comparable to the oscillation period. Above about 400 Torr, extinction instabilities therefore emerge as a circuit–plasma interaction rather than a simple monotonic decay.

Metastable helium dynamics were tested directly by tunable diode laser absorption spectroscopy on $_3$9 at 1083 nm. The metastable density reaches about $_2$0 after ignition at 750 Torr and about $_2$1 at 350 Torr for a current of about 2 mA (Kulsreshath et al., 2016). During oscillatory extinction at 750 Torr, metastable density also oscillates, but with a delay of about 2 μs relative to the current oscillations. The reported conclusion is that metastable atoms cannot be at the origin of the observed instabilities.

Self-pulsing constitutes a distinct dynamical regime. In helium MHCDs, a self-pulsing regime may appear after breakdown before transition to the normal glow (Dufour et al., 2019). With a ballast resistor of 1 MΩ, self-pulsing is obtained easily at mean currents typically < 1 mA, giving pulses of about 28 mA with duration 1–2 μs and charging intervals about 140 μs at 200 Torr and 380 V (Kulsreshath et al., 2016). A numerically studied micro thin cathode discharge in atmospheric-pressure argon, while not a conventional symmetric MHCD, is placed in the same broader microdischarge context and shows self-pulsing in the expected MHz frequency range. In that case, a glow-like core forms near the cathode edge, a second localized density maximum appears near the dielectric, propagates, and merges with the core, producing quasi-periodic but not strictly periodic current pulses with chaotic features (Wollny et al., 2011). This suggests that localized plasma–surface ionization cycles can dominate high-frequency self-pulsing in thin-cathode microdischarges.

4. Cathode sheath, plasma structure, non-local ionization, and thermal state

Simulation and diagnostic studies converge on the cathode sheath as the dominant control region in many MHCD operating regimes. In a 2D cylindrical fluid model of helium microdischarges with limited cathode area, the principal effect of cathode-area reduction is on the cathode sheath rather than on the axial field in the cavity (Dufour et al., 2016). For a 5 mA discharge at 100 Torr, the cathode sheath voltage is about 160 V with no limitation, about 185 V for $_2$2, and about 330 V for $_2$3. By contrast, the axial voltage profile along the cavity center is nearly unchanged, with about 20 V across the cavity interior, corresponding to an on-axis field of about

$_2$4

or

$_2$5

The same study reports a clear sheath–bulk structure in the radial density profile. For $_2$6, the quasineutral bulk extends from the axis to about 70 $_2$7m, the bulk plasma density is about $_2$8, the sheath thickness is about 60 $_2$9m, and the ion density remains about $_3$0 in the sheath (Dufour et al., 2016). For $_3$1, the plasma is slightly more confined and the sheath thickness increases to about 65 $_3$2m. This is presented as direct evidence that cathode-area limitation raises current density and sheath voltage rather than fundamentally altering the cavity interior.

Broader hollow-cathode measurements in pure hydrogen provide a sheath-resolved comparison class. Spatially resolved electric-field measurements in an abnormal glow hollow cathode discharge, interpreted with Rickards’ and Wroński’s models, show convex cathode-fall field profiles, cathode-fall length decreasing with increasing pressure and current, and ion energies near the cathode surface of roughly 35–70 eV (Gonzalez-Fernandez et al., 2020). Although this discharge is not a microdevice, the reported implication for MHCDs is that the sheath can occupy a large fraction of the cavity and strongly control scaling, ion bombardment, and secondary-electron generation.

Kinetic theory adds a complementary perspective. The non-local ionization source theorem for glow discharge and hollow cathode formalizes ionization as a finite avalanche in energy space, with kernel nilpotency leading to the finite Neumann expansion

$_3$3

and with energy support constrained by

$_3$4

The physical meaning given is that hollow cathode sustainment is a non-local, energy-limited avalanche organized by geometry, field, and wall reflection rather than a purely local Townsend process (Gorin, 2012). A plausible implication for MHCDs is that their small cavity dimensions make this non-locality particularly consequential for ignition and sustainment.

Gas heating is substantial but does not imply arc transition in the reported helium devices. In direct-current parallel microdischarges in helium, gas temperature determined from the rotational structure of the nitrogen second positive system,

$_3$5

with line intensity

$_3$6

reaches roughly 300–500°C for $_3$7–$_3$8 mA, and about 500°C at 10 mA (Dufour et al., 2016). The result is explicitly identified as much lower than an arc temperature and consistent with normal glow rather than arc. In cathode-limited helium microdischarges at 100 Torr, measured gas temperature reaches about 540 K at 18 mA for unlimited cathode area, and about 600 K at only 4 mA for $_3$9; simulations give maximum $_2$0 K, 430 K, and 650 K for unlimited, 1 mm, and 500 μm openings, respectively (Dufour et al., 2016). The stated reason is increased power deposition by ion current in the sheath when current is forced through a smaller area.

5. Parallel operation, arrays, and integrated implementations

Parallel operation is a major practical challenge because the first cavity to ignite usually lowers the device voltage below the breakdown threshold of neighboring cavities. In direct-current parallel microdischarges in helium, multi-hole devices with 7, 12, 20, or 31 holes and spacing 175–380 µm do not generally ignite all holes simultaneously (Dufour et al., 2016). The discharge begins in one hole, the voltage drops, additional holes may ignite sequentially as current increases, and the number of lit holes can increase up to about five in the reported 12-hole device. At higher current, the discharge can collapse back into a single hole. The interpretation given is that the normal-glow plasma extends along the cathode surface until it encounters another cavity with sufficiently favorable conditions.

Cathode-area limitation provides a route to ballast-free parallel operation. In a seven-hole helium MHCD with holes in a line and 260 µm center-to-center spacing, no cathode limitation leads to only one cavity igniting first and remaining the only active hole even up to 25 mA (Dufour et al., 2019). When the cathode is covered so that the exposed area is 600 × 2700 µm$_2$1, the breakdown voltage increases to about 245 V from 185 V, the discharge enters an abnormal glow regime with positive differential resistance, and additional holes ignite as current increases; the other cavities can be ignited at about 27 mA without individual ballasts. Small voltage jumps of about 5 V at approximately 9, 13, and 19 mA correlate with ICCD observations showing ignition of additional cavities. The emission intensity from different holes is not identical, and current is not equally shared among all cavities. Microscopic differences in hole geometry and edge roughness are identified as important sources of ignition variability.

Integrated silicon implementations generalize MHCD operation to microfabricated arrays. In silicon micro-reactors designed as cathode boundary layer devices, 10 × 10 arrays of 100 μm cavities operated in He at 500 torr achieve total current around 20 mA, with 83 of 100 cavities igniting because 17 were blocked by epoxy; by about 10 mA, all 83 usable cavities were lit, and current per microdischarge could reach about 300 μA average (Dussart et al., 2016). For 50 μm anisotropically etched cavities, microplasmas do not ignite at lower pressure, even at 700 torr, but at 1000 torr, nearly all cavities ignite. The paper also states that 100 microdischarges of 50 m diameter could be ignited in parallel at 1000 torr, while noting nonuniform emission intensity and the appearance of parasitic edge sparks at high current.

Polarity plays a decisive role in these silicon devices. When the silicon substrate is the cathode, plasma is strongly linked to the cavities; when polarity is reversed so that nickel becomes the cathode, the plasma expands more broadly over the cathode surface, does not significantly enter the cavities, and more current is needed to light the whole array (Dussart et al., 2016). This supports the broader MHCD result that cathode area confinement, rather than cavity geometry alone, governs whether a discharge remains cavity-localized and abnormal-glow-like.

Array operation remains limited by reliability and parasitic phenomena. In integrated silicon arrays, transient sparks and parasitic microdischarges appear especially near sample edges once all nominal cavities have ignited or when average current exceeds about 10 mA, and samples are described as fragile with lifetime only about 10 minutes for injected currents between 1 and 20 mA before a short circuit eventually forms (Dussart et al., 2016). These observations suggest that scalability in MHCD arrays depends not only on ignition physics but also on materials robustness, unintended cathode-area access, and defect control.

6. Diagnostics, applications, and current research directions

Optical and electrical diagnostics are central to MHCD research because regime identification, gas heating, cavity-to-cavity variability, and reactive-species production are not accessible from static $_2$2–$_2$3 curves alone. ICCD imaging in helium shows that the luminous region extends well beyond the hole as current increases, directly confirming plasma expansion over the cathode surface in the normal glow regime (Dufour et al., 2019). In extinction studies, PMT emission oscillations are in phase with current oscillations, confirming a plasma origin for the instability (Kulsreshath et al., 2016). TDLAS on helium metastables, high-resolution rotational spectroscopy, and synthetic-spectrum fitting provide quantitative access to metastable density and gas temperature.

Recent work on a DC Ar/N$_2$4 MHCD plasma jet emphasizes that rotational thermometry in microplasmas is probe-dependent under strong non-equilibrium conditions. In a Mo/Al$_2$5O$_2$6/Mo device with 400 µm hole, 20–80 mbar high-pressure chamber, 10 mbar low-pressure chamber, and operating currents 0.5–3.5 mA, high-resolution spectra of N$_2$7(C–B), OH(A–X), NH(A–X), and NO(A–X) are analyzed with Boltzmann plots and synthetic spectra (Stefas et al., 8 Jan 2026). The relevant expressions include

$_2$8

and

$_2$9

The study finds that $_3$0 of N$_3$1(C) is strongly influenced by argon metastables and NH overlap, whereas OH(A) is the most reliable estimate of $_3$2, typically around 450–650 K. This sharpens an important diagnostic caution: isolated rotational-temperature measurements can be misleading in metastable-rich MHCDs.

MHCDs are also being used as reactive microplasma sources for materials synthesis. A ns-pulsed N$_3$3/Ar MHCD with 100 µm Mo electrodes, 750 µm AlN dielectric, 250 µm hole, $_3$4 mbar, $_3$5 mbar, negative pulses of $_3$6 kV, 500 ns width at 10 kHz, and a +200 V DC biased substrate holder 4 cm away is used for plasma-enhanced chemical vapor deposition of h-BN on Si (Menacer et al., 7 Jul 2025). The discharge energy is reported as

$_3$7

with equivalent MHCD capacitance about 45 pF. In-situ OES detects N$_3$8 second and first positive systems, N$_3$9 first negative system at 391.4 nm, and multiple Ar I lines; the authors highlight argon-metastable-assisted nitrogen dissociation pathways. ICCD imaging shows a bell-shaped profile with intensity highest along the MHCD axis and more uniform near the substrate than in the bulk. Raman spectroscopy reveals the h-BN $_2$0 mode at about 1367 cm$_2$1, AFM gives average thickness about 33 nm after 90 min, and XPS shows average B/N = 1.5 ± 0.05 in the central map, with oxygen and compositional nonuniformity attributed to plasma and thermal inhomogeneity. The use of AlN rather than Al$_2$2O$_2$3 is explicitly motivated by mitigation of oxygen contamination from the dielectric.

Across these studies, several misconceptions are addressed directly by experiment. MHCD operation is not identical to the classical hollow-cathode effect in thin-cathode microdevices (Dufour et al., 2016). Stable high-current operation in helium does not imply arc formation, because measured gas temperatures remain consistent with normal glow (Dufour et al., 2016). Metastable oscillations during extinction are not the origin of the instability, because they lag the current oscillations (Kulsreshath et al., 2016). A plausible synthesis is that MHCDs are best understood as sheath-dominated, geometry-sensitive, externally coupled microglow systems whose behavior spans normal glow, abnormal glow, self-pulsing, sequential array ignition, and reactive-plasma processing, with cathode-area control providing one of the most effective levers for both physical understanding and device engineering (Dufour et al., 2019).