J. B. Javedani

Insulator surface flashover due to UV illumination

The surface of an insulator under vacuum and under electrical charge will flashover when illuminated by a critical dose of ultra-violet (UV) radiation - depending on the insulator size and material, insulator cone angle, the applied voltage and insulator shot-history. A testbed comprised of an excimer laser (KrF, 248 nm, ∼16 MW, 30 ns FWHM,), a vacuum chamber, and a negative polarity dc high voltage power supply (≤ −60 kV) were assembled to test 1.0 cm thick angled insulators for surface-flashover. Several candidate insulator materials, e.g. High Density Polyethylene (HDPE), RexoliteR 1400, Macor™ and Mycalex, of varying cone angles were tested against UV illumination. Commercial energy meters were used to measure the UV fluence of the pulsed laser beam. In-house designed and fabricated capacitive probes (D-dots, ≫12 GHz bandwidth) were embedded in the anode electrode underneath the insulator to determine the time of UV arrival and time of flashover. Of the tested insulators, the +45 degree Rexolite insulator showed more resistance to UV for surface flashover; at UV fluence level of less than13 mJ/cm2, it was not possible to induce a flashover for up to −60 kV of DC potential across the insulators surface. The probes also permitted the electrical charge on the insulator before and after flashover to be inferred. Photon to electron conversion efficiency for the surface of Rexolite insulator was determined from charge-balance equation. In order to understand the physical mechanism leading to flashover, we further experimented with the +45 degree Rexolite insulator by masking portions of the UV beam to illuminate only a section of the insulator surface; 1) the half nearest the cathode and subsequently, 2) the half nearest the anode. The critical UV fluence and time to flashover were measured and the results in each case were then compared with the base case of full-beam illumination. It was discovered that the time for the insulator to flash was earlier in time for the cathode-half beam illumination case than the anode-half illumination case which led us to believe that the flashover mechanism for the UV illumination is initiated from the cathode side of the insulator. Qualitatively stated, the testing revealed that the shielding of the cathode triple point against UV is more important than the anode triple junction in the design of vacuum insulators and electrodes.

UV Induced Insulator Flashover

Insulators are critical components in high-energy, pulsed power systems. It is known that the vacuum surface of the insulator will flashover when illuminated by ultraviolet (UV) radiation depending on the insulator material, insulator cone angle, applied voltage and insulator short-history. A testbed comprised of an excimer laser (KrF, 248 nm, ~2 MW/cm2, 30 ns FWHM,), a vacuum chamber (low 1.0E-6 torr), and dc high voltage power supply (<60 kV) was assembled for insulator testing to measure the UV dose during a flashover event. Five in-house developed and calibrated fast D-Dot probes (>12 GHz, bandwidth) were embedded in the anode electrode underneath the insulator to determine the time of flashover with respect to UV arrival. A commercial energy meter were used to measure the UV fluence for each pulse. Four insulator materials high density polyethylene, Rexolitereg 1400, Macortrade and Mycalex with side-angles of 0, plusmn30, and plusmn45 degrees, 1.0 cm thick samples, were tested with a maximum UV fluence of 75 mJ/cm2 and at varying electrode charge (10 kV to 60 kV). This information clarified/corrected earlier published studies. A new phenomenon was observed related to the UV power level on flashover that as the UV pulse intensity was increased, the UV fluence on the insulator prior to flashover was also increased. This effect would bias the data towards higher minimum flashover fluence.

Progress on simulating the initiation of vacuum insulator flashover

Vacuum insulators are critical components in many pulsed power systems. The insulators separate the vacuum and non-vacuum regions, often under great stress due to high electric fields. The insulators will often flashover at the dielectric vacuum interface for electric field values much lower than for the bulk breakdown through the material. Better predictive models and computational tools are needed to enable insulator designs in a timely and inexpensive manner for advanced pulsed power systems. In this article we will discuss physics models that have been implemented in a PIC code to better understand the initiation of flashover. The PIC code VORPAL [1] has been ran on the Linux cluster Hera at LLNL. Some of the important physics modules that have been implemented to this point will be discussed for simple angled insulators. These physics modules include field distortion due to the dielectric, field emission, secondary electron emission, insulator charging, and the effects of magnetic fields. In the future we will incorporate physics modules to investigate the effects of photoemission, electron stimulated desorption, and gas ionization. This work will lead to an improved understanding of flashover initiation and better computational tools for advanced insulator design.

Understanding high voltage vacuum insulators for microsecond pulses

High voltage insulation is one of the main areas of pulsed power research and development since the surface of an insulator exposed to vacuum can fail electrically at an applied field more than an order or magnitude below the bulk dielectric strength of the insulator. This is troublesome for applications where high voltage conditioning of the insulator and electrodes is not practical and where relatively long pulses, on the order of several microseconds, are required. Here we give a summary of our approach to modeling and simulation efforts and experimental investigations for understanding flashover mechanism. The computational work is comprised of both filed and particle-in-cell modeling with state-of-the-art commercial codes. Experiments were performed in using an available 100-kV, 10-μs pulse generator and vacuum chamber. The initial experiments were done with polyethylene insulator material in the shape of a truncated cone cut at +45° angle between flat electrodes with a gap of 1.0 cm. The insulator was sized so there were no flashovers or breakdowns under nominal operating conditions. Insulator flashover or gap closure was induced by introducing a plasma source, a tuft of velvet, in proximity to the insulator or electrode.

Measuring helical FCG voltage with an electric field antenna

A method of measuring the voltage produced by a helical explosive flux compression generator using a remote electric field antenna is described in detail. The diagnostic has been successfully implemented on several experiments. Measured data from the diagnostic compare favorably with voltages predicted using the code CAGEN [1], validating our predictive modeling tools. The measured data is important to understanding generator performance, and is measured with a low-risk, minimally intrusive approach.

The application of Kiuttu's formulation to study coaxial Flux Compression Generators

A class of flux compression generators (FCGs) is based on the compression of the cross-sectional area of a coaxial geometry where the current flows along the outer conductor and returns through the inner conductor. This compression causes an increase in current since magnetic flux must be conserved. Kiuttus inductive electric-field formulation is a powerful tool for the conceptual design of coaxial FCGs. The usefulness of this formulation is demonstrated in this paper for a simplified geometry using a finite-element partial differential equation solver (FlexPDE) for calculation of the inductive electric field. A time-varying applied current or a moving surface creates the nonconservative electric field. Losses due to diffusion of magnetic flux into conducting surfaces can also be accounted for and modeled in this setting. This analytical-computational approach serves as an important step in validating the magnetohydrodynamic (MHD) portion of the complex multiphysics parallel Lawrence Livermore code, Arbitrary Lagrangian-Eulerian (ALE3D). The nonintuitive boundary conditions involved in solving the otherwise straightforward partial differential equations are described in detail and illustrated in a simple model. The physical parameters used in the simulations are not based on a specific design.

Development of the vacuum power flow channel for the Mini-G

The Mini-G explosive pulsed power system is a two-stage helical-coaxial FCG that is geometrically a half-scale version of LLNLs FFT device. The generator is capable of delivering 60 MA currents and 10 MJ of energy to suitable inductive loads. The Mini-G is presently used in high-energy-density physics experiments that require efficient current delivery through a vacuum power flow region to the load. As with the FFT device, the Mini-G system requires a compact, high-voltage gas-to-vacuum insulator and low-inductance vacuum power flow channel to achieve high performance and maximum energy delivery. In designing the Mini-G system, we followed the successful approach used in developing the FFT device. This included shaping the electrodes and insulators to manage electric field enhancements, applying coatings to cathode surfaces to suppress electron field emission, introducing baffles to the power flow channel to block UV, and applying coatings to electrode surfaces to absorb UV. This paper describes the design of the Mini-G vacuum interface and power flow region, and results of modeling and simulations that were done to evaluate and optimize performance. Appropriate codes were used to examine electric field enhancements, magnetic insulation, flashover inhibition and UV ray tracing in the channel. In this paper, we also present results of laboratory testing on`and shapes, UV induced insulator flashover, along with measurements of HV thresholds for electron emission. We also report on UV reflectance data for some of the coatings considered. To date, there have been eight experiments performed using the Mini-G system. For the first two tests, the power flow channel had an extremely low vacuum inductance of 0.9 nH. On the second Mini-G test it appeared that a partial shorting occurred in the power flow channel, limiting full energy delivery to the load. The design was modified to reduce electrical stress, improve UV attenuation, and incorporate additional diagnostics. This increased the inductance of the power flow channel to 1.5 nH. On the third Mini-G test the partial shorting reoccurred and the new diagnostics (inner Bdot probe) helped to identify the location at the vacuum insulator surface - about 10% of total current of 41 MA was diverted into the short. Further design modifications were incorporated to decrease electrical stress across the insulator and reduce UV illumination of the insulator surface. This increased the inductance of the power flow channel to 1.9 nH. On subsequent Mini-G experiments full current delivery to the load has been achieved with no occurrence of shorting.

Evaluation of magnetic insulation in SF 6 filled regions

The use of magnetic fields perpendicular to quasi-static electric fields to deter electrical breakdown in vacuum, referred to as magnetic insulation, is well understood and used in numerous applications. Here we define quasi-static as applied high-voltage pulse widths much longer than the transit time of light across the electrode gap. For this report we extend the concept of magnetic insulation to include the inhibition of electrical breakdown in gases. Ionization and electrical breakdown of gases in crossed electric and magnetic fields is only a moderately explored research area. For sufficiently large magnetic fields an electron does not gain sufficient energy over a single cycloidal path to ionize the gas molecules. However, it may be possible for the electron to gain sufficient energy for ionization over a number of collisions. To study breakdown in a gas, the collective behavior of an avalanche of electrons in the formation of a streamer in the gas is required. Effective reduced electric field (EREF) theory, which considers the bulk properties of an electron avalanche, has been successful at describing the influence of a crossed magnetic field on the electric field required for breakdown in gases; however, available data to verify the theory has been limited to low gas pressures and weak electronegative gases. High power devices, for example explosively driven magnetic flux compressors, operate at electrical field stresses, magnetic fields, and insulating gas pressures nearly two orders of magnitude greater than published research for crossed fields in gases. The primary limitation of conducting experiments at higher pressures, e.g. atmospheric, is generating the large magnetic fields, 10s Tesla, and electric fields, ≫100 kV/cm, required to see a significant effect. In this paper we describe measurements made with a coaxial geometry diode, form factor of 1.2, operating at peak electrical field stress of 220 kV/cm, maximum magnetic field of 20 Tesla, and SF6 pressure of 760 torr.

Vacuum Surface Flashover of Insulators for Microsecond Pulses

Summary form only given. A critical component to a high energy pulsed power system is the insulator. The insulators function is to standoff high voltages between two electrodes of opposite polarity. Although the intrinsic dielectric strength of the insulator is rather high, relatively low electric fields -several times less-is withstood by insulator surfaces that is exposed to vacuum. To investigate surface flashover for the widely used truncated conical insulator configuration, a vacuum chamber was fabricated where we installed a positive 45 degree conventional HD polyethylene insulator in the electrode spacing (nominal 1.0 cm). With applied 100 kV pulses of 5 microsecond duration the insulator held off the applied voltage -but experienced flashover if a source of plasma/electrons in the form of a small piece of velvet (1.0 mm dia.) was introduced in the vicinity of the insulator on either of the electrodes surfaces. Plasma expansion velocities of 1.4 to 2.7 cm/microseconds are inferred from voltage collapse and current drawn signals. It appears that the actual flashover occurs when the plasma that is launched from the cathode reaches either the insulator or the anode electrode. The breakdown mechanism and improvements in the insulator design strictly speaking can be made independently with the more advanced computational means. We also report on the progress that has been made with our PIC code modeling and its agreement with our experimental observation.