INSULATOR SURFACE FEATURES RESULTING FROM CUTTINGTECHNIQUES*

R.A. Schill1, Jr., W. Culbreth2, and R. Venkat1

University of Nevada Las Vegas1Electrical and Computer Engineering Department, Box 454026

2Mechanical Engineering Department, Box 4540274505 Maryland ParkwayLas Vegas, Nevada 89154

J.M. Elizondo, A. Dragt, and M. KroghHoneywell FM&T

Albuquerque, New Mexico

* Work supported in part by the Department of Energy under contract number DE-FG02-00ER45831.

Abstract

Surface features of similar plastic-metal laminateinsulator geometries cut with a water jet and a machinetool are examined with a scanning electron microscope.Both secondary electron emission and backscatteringelectron emission are used to image and differentiateinsulator surface structures. The insulator surface withoutand with gold coat were compared. Low level electronbeam currents and accelerating surface voltages wereapplied in the former case preventing charge up andsurface damage. General surface features appearedunaltered by the sample preparation. The surface featuresof the machine cut insulator and the water jet cut insulatorwere significantly and unexpectedly different. This mayexplain why the water jet cut piece tends to have a highervoltage hold-off as compared to the machine tooled piece.

I. Introduction

The surface breakdown on plastics has been a topic ofconcern for many decades [1,2,3]. It is well known thatthe plastic barriers located between a cathode and ananode tend to breakdown at levels below their intrinsicrated bulk breakdown levels. Various breakdownmechanisms have been identified but it appears at thistime that there is no consensus on any one dominantmechanism [2,3,4]. Further, the mechanism and surfacedamage appears to be pulse width dependent [3,5].

The microstack insulator, a plastic-metal laminate, hasshown promise in inhibiting dielectric breakdown on thesurface of barriers and insulators [6,7]. The metal platesare either flush with the surface of the laminate or extendslightly beyond the laminate. Microstacks tend to controlsecondary electron emission, desorption of species, andultraviolet effects. After damage failure, the device isresilient enough to not be damaged [6]. It is hypothesized

that the metal plates capture the electrons that initiate theflashover process thereby delaying or preventing theshorting of the anode and cathode. Depending on thefield configuration, the collected electrons are distributedthroughout the plate until a sheath is generated preventingthe further collection of charge. To aid in the voltagehold-off, a conical stack may replace the cylindrical stackwith conical angle between 0 and 450 relative to thenormal of the base of the cone located on the cathodeelectrode. In general, experiments show that conicalinsulators with a 450 angle have maximum hold-offvoltages [2,3]. The hold-off voltage of the microstack isenhanced as well.

Surface failure or breakdown often occurs as a result ofcontaminant layers, cracks and/or other surfaceimperfections [8]. Oxidation or chemical reactions, as aconsequence of the surrounding environment, may alterthe salient features of the surface. Surface roughness andpressure may influence the electrical properties of thematerials resulting in interfacial failures. The electricalperformance may be significantly affected by particulatecontaminates. Therefore, much attention must be focusedon the preparation of surface barriers and insulators.

Microstack insulators are cut using various cutting andpolishing techniques. The following two cuttingtechniques are commonly employed: 1) machine toolcutting and 2) water jet cutting. Both cutting techniquesand post surface preparation influence the overallperformance of the insulator or barrier. Experiments haveshown that the water jet cut piece tends to have a higherhold-off voltage as compared to the machine cut piece.Examining the surface structure using a scanning electronmicroscope provides some insights to the differences inthe hold-off performance of the two cutting techniques.Surprisingly, water jet cutting resulted in an unintentionalsurface grading giving the piece a directionality or ananisotropic characteristic possibly useful in the holdingoff higher voltages. Reasons for this are explained.

II. Scanning Electron Microscope &Preparation Details

A JOEL-5600 scanning electron microscope (SEM)located in the Electron Microanalysis and ImagingLaboratory at UNLV was employed to study the surfacesof two differently cut microstack insulators. The scanningelectron microscope is equipped with a bacscatteringelectron (BSE) detector and an Oxford ISIS energydispersive spectrometer (EDS) controlled by a twonetworked windows operating system.

The SEM has a resolution of up to 50 nm at 100,000times magnification. The device can scan four 1 cmdiameter samples or one 3.2 cm diameter sample. Amaximum sample height of about 1 cm is allowed.

Because samples were cylindrical in geometry and wellbeyond the limited size for placement in the SEM, thesamples were carefully cut along a chord of the cylindercross section. A hacksaw was used to perform the cut.Damage due to this cutting process was localized for themost part along the cut edge. Observations near theseedges are excluded in the discussions of this paper.

Uncoated samples are cleaned with tissue paper and airblown with mouth. Samples cleaned in this fashion wereexamined for gross effects and tendencies only. Finedetails such as fiber threads on the surface were alsoobserved. A 1.5 kV beam voltage and a maximum 39 µAload current allowed for examining the sample surfacewith a small working distance.

Coated samples are first cleaned with a petroleum etherthat has a low residue, high evaporation rate. A Pelcosputter coater is used to gold coat the samples after theyare thoroughly cleaned. A 5 to 10 nm thick gold coat isdeposited on the sample surface. The resolution of thecoating thickness is beyond the limits of the SEM. Abeam voltage between 10 to 15 kV is used having a loadcurrent of about 75 µA.

Excluding observed finger grease, dust and tissuecontaminants on the uncoated sample, the coated and theuncoated samples had the exact same tendencies. Thisimplies that the gold coat sputtered on the surface of theinsulator under test did not affect the surface features ofthe microstack.

III. The Microstack – Design, Cutting, andPost Cutting Processes

The microstack laminate is composed of 0.005” to0.01” of Rexolite (polystyrene) typically manufactured byC-Lec and 0.0005” thick standard 304 stainless steelannealed foil. Stainless steel 304 was used because it hasfewer sulfates and therefore is more suitable in a vacuumenvironment. A 0.001” self-supporting Pyraluxmanufactured by DuPont is used as a thermoset to bondthe Rexolite to the metal foil. The microstacks under testcontained about 20 laminate layers resulting in an overallmeasured 0.94 to 1.03 cm thickness.

After the sheets of plastic are cut, they are cleaned andplaced in a dryer or vacuum oven. The wafers are stackedin a portable clean room. The stack is then pressed with aspecial recipe taking into account the melting point of thematerial being pressed. Special care is taken when therecommended curing temperature is near the meltingtemperature of the plastic sheet.

Either a machine tool or a water jet stream is employedto cut the stack. In the former process, a kerosene orwater-soluble cooling solution is used. A kerosenecooling solution was used on the machined microstackunder study. A final clean-up cut is made when thedimension is within 0.02” in diameter. The final cut ismade to 0.01” diameter.

A company, specializing in cutting stone, performedwater jet cutting. Besides a fungicide and algae controlsolution, additives added to the water jet stream were notknown. When cutting, the jet steam was oriented normalto the top surface of the microstack. A clean-up cut is notmade using this technique.

In both techniques, the same post processing procedureis followed. The samples are polished with 600, 800,1000, 1200, and 2000 grit emery paper using as littlefriction as possible with a water based coolant. Thepolishing procedure is considered completed when thesurface yields a clear sheen and no marks or scratches areobserved with a 10x microscope. The samples are thenalcohol immersed, sprayed and dried with “Dust Free”canned air, and handled with surgical quality gloves.

IV. Surface Studies With Scanning ElectronMicroscope

The samples examined with the SEM, have been usedin experiments, labeled and shelved for a few years.

Figure 1. Machine cut microstack.

SEM pictures of the machined piece are depicted inFigs. 1-4. The bright surface is the edge of the metal foil.An overview of more than two sets of laminations in themachine cut piece shown in Fig. 1 indicates a heightdifferential among adjacent layers. The Rexolite surfaceis about 5 to 15 µm higher than the metal foil in Fig. 2. Itis believed that the mechanical rigidity and thermal

Figure 2. Magnified view of plastic – SS foil interface.

Figure 3. Surface blemish in plastic wafer and thermosetmelt between the plastic and metal foil.

Figure 4. Degassing in progress in machined cut piece.

properties of the different layers of material resulted inuneven cuts across adjacent material layers. The softthermoset appears to have melted on the right side of themetal foil in Fig. 3 where as in Fig. 2 there is substantialroughness illustrated. The left side of the metal foil inFig. 3 shows slivers of the metal foil separated or partiallyseparated from the foil itself. Figure 3 also shows aninherent imperfection in the Rexolite plastic. Striations

and pitting are apparent in Figs. 2 and 4. After over acouple of hours in the vacuum chamber, black rings beganto develop around highlighted regions with spider-liketentacles as in Fig. 4. This is an indication of degassingof material trapped in the plastic surface under the goldcoating [9]. The magnified view of Fig. 4 seems toindicate that the striations in Fig. 2 are a result of uniformpitting possibly a consequence of the polishing process.

Figure 5. Water jet cut microstack.

Figure 6. Magnified view of plastic – SS foil interface.

Figures 5-9 depicted the surface features of the water jetcut piece. Surprisingly, the water jet cut piece has auniform staircase-like cut as shown in Fig. 5. This givesthe plastic an orientation or anisotropic property. Rexoliteitself is amorphous. As a result, the straight cut along theplastic surface tends to indicate that the water jet streamwas normal to the microstack surface until deviated ateach foil edge. Figure 6 provides a magnified view of thecut across the plastic-thermoset-foil-thermoset-plasticinterfaces. It is believed that the crystalline structure ofthe stainless steel foil resulted in a material yield notalong the original trajectory of the water stream but alongone of the crystalline planes. It is apparent from Fig. 6that the water jet flows from the right to the left. This is

Figure 7. Erosion observed in the thermoset bonding theRexolite and stainless steal wafers.

Figure 8. Deposits of plastic lodged up against the foiledge on the water jet side of the foil.

further born out in Fig. 7. To the left of the foil edge,turbulent effects are easily identified. This erosion ofmaterial is an indication of Strouhal eddies formed in thewake of the water stream because of the softness of thethermoset relative to the foil and plastic materials. BSEwas used to identify the material composition ofroughness on the “leeward” side of the foil edge. Thestainless steel edge appears to be sharp or corrugated withmetal fragments imbedded in the plastic. Figure 8illustrates on the water jet side (right side of Fig. 6) thatthe foil edge is smoothed over. The impact of the waterstream on the foil side (not edge) tends to buckle the foil.Plastic fragments tend to become lodged up against thefoil side. Although not illustrated in these figures, BSEscans indicate that a foreign obstacle with pointedgeometry appear to be randomly lodged in the surface ofthe microstack. Although not verified, it is possible thatsome fine cutting additive may have been incorporatedinto the water jet stream to aid in the cutting process orthe fine grit of emery cloth was deposited on themicrostack surface during the polishing process. Similarto the machine cut piece, striations formed by pitting on

the plastic surface, Fig. 7, and on the foil edge, Fig. 8, arepresent and perpendicular to the flow of the water jetstream. This is a good indication that the post polishingprocess tends to modify the surface feature. Althoughinfrequent, material imperfections in both the plastic andthe foil edge were observed.

V. Conclusion

Different cutting techniques and post cutting processescan significantly alter the surface features of themicrostack insulator due to the varying material propertiesof the different laminate layers. This may significantlyaffect the electrical performance of the insulator. Theunintended stair-cased structure of the water jet cut piecerounds off and exposes a metal conductor to aid incapturing primary electrons before and electron avalanchecondition results. Further, appropriately orientating theinsulator allows one to capitalize upon the angledependence of the cone. In comparison, the machined cutmicrostack tends to hide the electrode thereby inhibitingits ability to collect charge efficiently.

VI. References

1. Martin, J.C., J.C. Martin on Pulsed Power, eds. T.H.Martin, A.H. Guenther, and M. Kristiansen, PlenumPress, NY, 1996, Chapter 8.

2. R. A. Anderson, "Surface Flashover: Three Decadesof Controversy", in XIVth International Symp.Discharges and Electr. Insul. in Vacuum (Santa Fe,New Mexico, USA, Sept. 17-20), 1990, pp. 311-317.

3. I.D. Smith, “Pulse breakdown of insulator surfaces ina poor vacuum,” Proc. I Intl. Symp. on Inusl. of HighVolt. in Vac., M.I.T., 1964, pp. 261-280.

4. T.P. Hughes and T.C. Genoni, “Yield calculations fora krypton plasma radiation source,” Mission ResearchCorp., Albuquerque, NM, Tech. Report MRC/ABQ-R-1997, chapter 2, 2000.

5. A. Watson, “Pulsed flashover in vacuum,” J. AppliedPhysics, vol. 38, pp. 2019-2033, Dec. 1966.

6. J. Elizondo, and W. Money, “Microstack insulator forflashover inhibition, Phase II,” Defense NuclearAgency, Alexandria, VA, Tech. Report DNA-TR-91-86, Oct. 1992.

7. S.E. Sampayan, G.J. Caporaso, D.M. Sanders, R.D.Stoddard, D.O. Trimble, J. Elizondo, M.L. Krogh,T.F. Wieskamp, “High-performance insulatorstructures for accelerator applications,” LawrenceLivermore National Lab., Report No. UCRL-JC-127274, May 1997.

8. K.N. Mathes, “Surface failure measurements,” inASTM Special Technical Publication 926,Engineering Dielectric, vol. IIB, Electrical PropertiesPart. B, 1987, pp. 221-312.

9. A.R. Cambell, S.A.W. Lundberg, and N.W. Dunbar,“Solid inclusions of halite in quartz: Evidence for thehalite trend,” Chemical Geology, vol. 173, pp. 179-191, 2001.