Experimental Study on High Electrical Breakdown of Water Dielectric

Plasma Science & Technology，Vol.7, No.6, Dec.2005Experimental Study on High Electrical Breakdown of WaterDielectric *Zhang Zicheng(张自成), Zhang Jiande(张建德), Yang Jianhua(杨建华)College of Optoelectric Science and Engineering, National University of Defense Technology,Changsha 410073, ChinaA bstract By means of a coaxial apparatus, pressurized water breakdown experiments withmicrosecond charging have been carried out with diferent surface roughness of electrodes and dif-ferent ethylene glycol concentrations of ethylene glycol/ water mixture. The experimental resultsabout the breakdown stress and the effective time are presented. The breakdown stress is normal-ized to the situation that the effective time is transformed to 1 us and analyzed. The conclusionsare as fllws: (1) the breakdown stress formula is modified to E = 0.561MA-1/10tctN pl/8;(2) .the cofficient M is significantly increased by surface polishing and ethylene glycol additive; (3) itis accumulative for the capacity of improving electrical breakdown strength for surface polishing,ethylene glycol additive, and pressurization, of which pressurization is the most efective method;(4) the highest stress of 235.5 kV/cm is observed in ethylene glycol/ water mixture with an ethy-lene glycol concentration of 80% at a hydrostatic pressure of 1215.9 kPa and is about one timegreater than that in pure water at constant pressure; (5) for pressurization and surface polishing,the primary mechanism to improve the breakdown strength of water dielectric is the increase inthe breakdown time delay. Research results indicate great potential in the application of the highpower pulse conditioning system of water diclectric.Keywords: high electrical breakdown, pressurized water dielectric, polished surface ofelectrodes, ethylene glycol additive, microsecond chargingPACS: 72.20.Ht, 77.22.Jp1 Introductionstress without breakdown, W'max the maximum energystorage density, and AD the action density which isDevelopment of HPM (high power microwave) rea measurement of the capacity to keep the maximumquires a high power pulse conditioning system with aenergy storage density. On the basis of the above de-high output power, a small volume, and a wide rangescription, it is evident that pressurization, higher finishof application. Water dielectric, due to its great dielec-of the electrode surface, and ethylene glycol additivetric constant and high breakdown stress, is extensivelycan all improve the high electrical breakdown strengthused as the energy storage medium in high power pulseof water dielectric.conditioning systerns. Improving high electrical break-In order to improve the high electrical breakdowndown strength of water dielectric is an effective ap-strength of water dielectric, it is important to haveproach to the minimization of the high power pulse con-some further breakdown experiments on the effect ofditioning system. Firstly, in the early 1970's, Mirza [1pressurization on water dielectric, diferent finish levelsand Miller [2] theoretically and experimentally verifiedof the electrode surface, and different ethylene glycolthat breakdown stress of water dielectric increases alongconcentrations of ethylene glycol/ water mixture.with the hydrostatic pressure to the one eighth power.But breakdown stress given by the Miller formula at2 Experimental apparatusa constant pressure is quite different from that givenby the Martin formula [3] or the Lupton formula(3l.An experimental apparatus with coaxial electrodesSecondly, both Miller [2] and Fenneman [4 pointed outis shown in Fig. 1. The schematic experimental systemthat the surface finish level of electrodes has an effectshown in Fig.2 is mainly made of a primary energyon the cofficient M in the breakdown stress formula.storage capacitor C1, a trigger switch K, a high powerFinally, research results obtained by Fenneman[问] inpulse transformer, a test cell of coaxial electrodes, adicated that breakdown stress of water dielectric insyste中国煤化工he-r, and a breakdowncreases along with the increase of ethylene glycol addi-voltacapacitor C1 hastive, but at the expense of a decrease in the dielectrica capYHC N M H Gsformer has a trans-constant. In order to measure the high electrical break-former ratio ot 15 whth primary and secondary induc-down strength of water dielectric, three parameters aretances of 4.13 μH and 2.694 mH. The test cell of coaxialintroduced as follows. Emax is the highest electricalelectrodes has a length L of 51.68 cm, a stressed elec-●The project supported by National 863 Project of China (No. 807-2020, 803-5051)Plasma Science & Technology, Vol.7, No.6, Dec. 2005Fig.1 Apparatus of breakdown experiment for water dielectric with coaxial electrodes”Trel20s160Fig.3 Breakdown voltage waveform for pressurized waterdielectric on surface polish and rough surface polisb of elec-trodesFig.2 Schematic experimental system of breakdown ex-periment for water dielectric with coaxial electrodestrode area A of 1780 em2, an inter eletrode separationdof 1.825 cm, an inner radius r1 of 8.34 cm, and anouter radius r2 of 10.165 cm, respectively. The test cellfor water dielectric has a resistanceRof 9.2 kS2 and acapacitance C of 8.6 nF with two eletrodes made ofFig.4 Breakdown voltage waveform for pressurized ethy-stainless seel. When the rsistivity of de ionized waterlene glycol /water mixture with ethylene glycol concentra-or ethylene glycol/water mixture exceeds 14.3 MA.cm,tion of 40% and 80%the de-ionized water purification system is stopped.Then pressurized water breakdown experiments are carUo (initial voltage on the capacitorC1):. 300 kV，ried out with the inner cylinder charged with. a neg-w (ethylene glycol concentration of the mixture): 0ative voltage and the outer cylinder grounded. The40%, and 80%,breakdown voltage of water dielectric is measured byP (hydrotatic pesure: 101.3 kPa~ 1621.2 kPa.oscilloscope through a resistant voltage divider shownin Fig. 2. The time interval between two breakdowns3.2 Experimental resultsis about 10 minutes during which water dielectric hasThe breakdown voltage waveforms for pressurizedrecovered from the breakdown itef.water dielectric with diferent electrode surface finishA hydrostatic pressure produced by highly pressur-levels and with diferent ethylene glycol concentrationsized liquid Nitrogen and controlled by means of aof ethylene glycol/water mixture are shown in Fig. 3pressure reducing valve, is applied to the water dielec-and Fig. 4, respectively. Those indicate that bothtric. The finish level of the electrodes' surface, iden-breakdown voltage and breakdown time (the time be-tifed with surface roughness Ra, is changed from atween the beginning of charging and breakdown for di-rough polish (R=1.6 pum~3.2 pum) to a finer polishelectric) increase along with the increase of hydrostatic(R.=0.4 μm~0.8 pm). The ethylene glycol concen-pressure, the improvement of the electrode surface fin-tration of the ethylene glycol/water mixture is con-ish level (rom rough polish to finer polish), and thetrolled by the value of ethylene glycol additive.increase in ethylene glycol concentration.3 Experimental results4 Analysis and discussion3.1 Option of experimental parametersAccording to Ref.[6], water breakdown initiates atBecause the hydrostatic pressure of 2026.5 kPa canthe positive electrode and breakdown stress can be ap-be easily reached in the test cell without special en-proximately expressed with the ideal formula Eex =gineering dificulties the maximum experimental hy-Uad/r21n(r2/r))17. And the general form of the break.drostatic pressure is chosen not to exceed 2026.5 kPa,lowelow)中国煤化工;上:-lectric is shown as fol-considering economy, safety, and application. Addition-ally, the experimental ethylene glycol concentration canTYHCNM H G:I/N pl18,(1)not be more than 80%, otherwise the property of thevuv“mixture, such as the dieletric constant, will sharplywhere E (kV/cm) is the breakdown stress, A (em2) isdecrease. As a result, the experimental parameters arethe stressed electrode area (defined as the area of thechosen as follows:3162Zhang Zicheng et al. : Experimental Study on High Electrical Breakdown of Water DielectricTable 1. Pressurization effect of water breakdown on good electrode surface polishP/kPa ter/us Eex/(kV.cm-T) EMartin/(kV.cm-T) △/% EMiller/(kV-cm-T) E:/(kV.cmI)101.3 1.56124.5122.41.7275.5144.43042.88141.3一343.4168.66082.92150.6373.6180.01317.2 3.20163.2 .405.3 .198.11621.2 3.24168.2-415.1204.6electrodes experiencing 90% or more of the maximum10[厂field), tefr (us) is the effective time (defined as the time0for which the voltage is above 63% of the peak value), P190}一SmutDreho(MPa) ie the hydrostatic pressure and M is the cefi-180cient depending on the quality of water (purity percentde- aeration, temperature, etc) and on the electrode'scharacteristics (geometry, surface finish, material, etc),60with 300 for a positive breakdown of pure water. If the50-.effective time te is less than or about equal to 2 pus,N=3, otherwise N=6 [2,5,8]. Q = 10 for the Martinformula or Q= 16 for the Miller formula. The Miller10有在有有15formula for a positive breakdown of pressurized waterFig.5 Breakdown stress versus pressuredielectric is shown as follows:which is in good agreement with that predicted by theE= 185A-1/184ert/sp1/8 (ter≤ 1 us),(2)Mirza theory and by the Miller formula; (3) breakdownE= 2864-16421/3p1/8 (er≈ 2 us), .(3stress is quite different quantitatively from that givenby the Miller formula. A possible interpretation is thatE = 320A-161/*p2/8 (ea> 2μus).(4the cofficient M in the Miller formula and area depen-4.1Pressurization effect on waterdence are not accurate enough on account of the infu-breakdownence of breakdown stress on the surface nature of theThe breakdown stresses of water dielectric and theelectrodes (such as surface finish); (4) the experimentalrelative diference O between the breakdown stressesresults are not only in good agreement with the Millergiven by the experimental results and by the Martin for-formula on the relationship of Et ~P, but also evenmula are shown in Table 1, where EMartin, EMiler, Eex, .more accurate than that given by the Miller formulaand E are the breakdown stresses given by the Martinquantitatively. As a result, the breakdown stress for-formula, Miller formula, experimental results, and ex-mula is modified to the following form in our experimen-perimental results normalized to the situation that thetal condition of a surface roughness of 0.4 μm~ 0.8 pum,effective time is transformed to 1 pus according to thean effective time at a microsecond scale, and a low by-relation of tetN , respectively. Those indicate that thedrostatic pressure:experimental breakdown stress Eex is in good agree-E= 0.561MA-1/14-l/N p1/8,(5)ment with the Martin formula but quite different fromthat given by the Miller formula at a constant pres-where M is equal to 168 for a positive breakdown ofsure and the diference between the breakdown stressespure water in general.shown by the experimental results and by the MillerThe mechanism for improving the electrical break-formula increases along with the increase of the hydro-down strength for pressurized water dielectric can bestatic pressure.explained by the bubble breakdown model given byThe experimental breakdown stress Et is plotted ver-Jones and K unhardt [9,101. According to their model,sus hydrostatic pressure in Fig. 5, where Eo and Er areelectron collision and ionization happen in bubbles andbreakdown stresses at a constant pressure and at ana breakdown results from the heating of field emissionapplied pressure for water dielectric when the effectivecurrent from asperities near the cathode to the bubbles.time is transformed to 1 μS. It can be seen that the ex-When a hydrostatic pressure is applied, the bubbles areperimental relationshipof Et ~ P is linear (i.e. break-compressed and the number of bubbles is reduced, thusdown stress is proportional to the hydrostatic pressurethe process of forming the bubbles heated by the fieldto the one eighth power) and in good agreement withemission current becomes difficult. Hence the electricalthat predicted by the Mirza theory and the Miller for~中国煤化工ctric is improved.mula.From Table 1 and Fig. 5, it can be seen: (1) at a con-4.finish ofstant pressure, the experimental breakdown stress EexTCH.CNMH Gr-lzdt Wateris in good agreement with the Martin formula; (2) forpressurized water dielectric, breakdown stress increasesThe breakdown stresses for pressurized water dielec-with the hydrostatic pressure to the one eighth power,tric with different electrode surface fnish levels are3163Plasma Science & Technology, Vol.7, No.6, Dec. 2005Table 2. Pressurization effect of water breakdown on rough electrode surface polishP/kPater/usEex/(kV.cm 7EMartin/(kV.cm- T )0/%B/(kV.cm2)101.31.36107.4128.1-16.2119.06082.24113.1148.0114.6.2.50117.8159.8Table 3. Breakdown stress for pressurized ethylene102t Ruptyolhed epermartal tnglycol/ water mixture00-Tmorbaln(a)w= 0100P/kPa te/4s Eex/(kV-cm) E/(kV.cm-T5 170f101.3 1.56124.5144.4三1603042.88141.3168.6u150-602.92150.6180.1317.23.20163.2198.1140-.131621.2 3.24168.2204.6(b)w = 40%P/MPa ten/us Eex/(kV-cm-) Er/(kV.cm-T)Fig.6 Breakdown stress versus pressure1013 1.36160.7178.0405.31.60170.8199.7shown in Table 1 and Table 2, where EMartin, Eex, Et .810.61.80180.8220.0and O are defined above. It can be seen: (1) at the1215.92.16183.4237.0fixed hydrostatic pressure, the experimental breakdown1418.62.20188.4245.1stress Et is about 20% higher with a finer surface fin-(c)w= 80%ish of the electrodes than that with a rough surfaceP/kPa te/us Eex/(kV-cm-) Er(kV.cm-T)finish and the experimental breakdown stress Bex is in101.3 1.52193.5222.better agreement with the Martin formula; (2) at the1.90212.8263.6fixed surface finish level of the electrodes, the break-2.00227.1286.1down stress Et increases along with the increase in the235.5306.3hydrostatic pressure; (3) it is accumulative for the ca-pacity of improving the electrical breakdown strength4.3 Influence of ethylene glycol additivefor surface finish and pressurization; (4) the increase in.on pressurized water breakdownbreakdown stress Et by 34.3% from pressurization of1013.3 kPa is higher than that by 21.3% from the sur-With different ethylene glycol concentrations in theface polish (Ra =0.4 μm ~0.8 pum), so pressurization iscondition of a good electrode surface polish, the break-more effective for improving the electrical breakdowndown stresses for the pressurized ethylene glycol/waterstrength.mixture are shown in Table 3, where Eex and Et are de-For diferent surface finish levels of the electrodes, thefined above. Those indicate that the breakdown stressexperimental breakdown stress Et for pressurized wa-Et increases with the increase of the hydrostatic pres-ter dielectric is plotted versus the hydrostatic pressuresure or the increase of the ethylene glycol concentra~in Fig.6, where the dashed line corresponds to formulation. The highest breakdown stress of 235.5 kV/cm is(5). It can be seen: (1) for different surface finish lev-observed in the ethylene glycol/ water mixture with anels of the electrodes, the experimental relationships ofethylene glycol concentration of 80% at a hydrostaticEt ~ P are both linear and in good agreement withpressure of 1215.9 kPa and is about one time greaterthose predicted by the Mirza theory and by the Millerthan that in pure water at a constant pressure.formula; (2) for surface polish of the electrodes, break-With diferent ethylene glycol concentrations, the ex-down stress is in good agreement with formula (5); (3)perimental breakdown stress Et for pressurized waterwhen the surface finish level of the electrodes changesdielectric is plotted versus the hydrostatic pressure infrom a rough polish to a finer polish, the slope of theFig. 7, where the dashed line corresponds to formulaline of Et ~ P increases by about 20% and the ceofi-(5) for pure water (w=0). It can be seen: (1) With dif-cient M rises by about 20% accordingly.ferent ethylene glycol concentrations, the experimentalAccording to the bubble breakdown model given byrelationshipsof Er ~ P are all linear and in good agree-Jones and Kuandardtl9,10, with the improved surfacement with that predicted by the Mirza theory and thefinish level of the electrodes, some explanations can be .Miller formula; (2) the slope of the line of Et ~ P in-obtained. On the one hand, the field emission currentcrease中国煤化iylene glycol concen-becomes weak, leading to an increase in the time oftratioCses with an increasenuclear forming Tnuc and an increase in the breakdownby abMHC N M H Gie glycol concentra-time lag Dd. On the other hand, the presence and for-tion of 40%0 ana 80%0， comparea with those for puremation of bubbles is inhibited and it is more difficult towater.insert bubbles on the polished surface of the electrodes.Table 4 contains three different figures of merit: theThus the probability of dielectric breakdown is reduced.highest electrical stress Emax (which can approximately3164.Zhang Zicheng et al. : Experimental Study on High Electrical Breakdown of Water DielectricTable 4. Electrical breakdown strength for pressurized ethylene glycol/ water mixturew/% P/kPat/°C C p/M2.cm) T/ms te/us Enx7(kV.cm-T) Wmax/(kV.cmI) ADJs.m-T厂101.3 2580140.10 1.56124.5550.091317.20.10 3.20163.290.300.10 3.24168.21000.32l0 101.3 25 7200.13 1.36160.780.1140 1215.9 25 730.13 2.16183.41090.2340 1418.6 25 730.13 2.20188.4115 .0.2580 101.3 25 63250.14 1.52193.51040.1680 1215.9 25 630.14 2.20235.51550.34surface polish of the electrodes, in general. The break-down stress increases with the hydrostatic pressure tothe one eighth power. The coeficient M significantlyincreases with the improvement of the finish level of theelectrode surface and the increase of the ethylene gly-2col concentration. It is accumulative for the capacityof improving the electrical breakdown strength of water200dielectric signifcantly for pressurization, surface finish,and ethylene glycol additive, of which pressurization isthe most effective means.Research results indicate a bright future for the ap-高高Fig.7 Breakdown stress versus pressureplication of the high power pulse conditioning systembe placed by the experimental breakdown stress Eex)of water dielectric.without a breakdown, the maximum energy storageReferencesdensity Wmax, and the action density AD, where Wmax= 1/2eoErBmax; AD = Wmax .teft (the value of AD is .Mirza J s, Smith C W, Calderwood J H. Bubbles,preferably as high as possible, as it is a measurerhent ofpressure and pre- breakdown in insulating liquid. Pro-the capacity of the test cell to keep the maximum en-ceedings of the 4th Internatioal Conference on Con-ergy storage density), and the intrinsic time constant,duction and Breakdown in Dielectric Liquids. Dublin,τ= Eo∈τP in which p is the resistivity of dielectric, Eo1972.193and εr are dielectric constants in vacuum and relativeMiller A R. High energy density, low impedence ca-dielectric constant in dielectric, respectively. In general,pacitors using pressured water as a dielectric. Fifthall these figures of merit improve as the ethylene glycolSymposium on Engineering Problems of Fusion Re-search. Dublin: 1973. 471~ 474concentration or hydrostatic pressure increases, but atKristiansen M, Hatfield L L. High voltage water break-the expense of a decrease in dielectric constant with thedown studies. Lubbock, TX: Defense Special Weaponsincrease of the ethylene glycol concentration. So it isAgency, 1998accumulative for the capacity of improving the electri-4 Fenneman D B, Gripshover R J. IEEE Transactionscal breakdown strength of water dielectric with ethyleneon Plasma Science,1980, 8(3): 209 ~ 213glycol additive and pressurization. Considering the ad-5 Fenneman D B. Journal of Applied Physics, 1982,ditional magnitude of the three figures of merit, the ad-53(12): 8961 ~ 8968ditional magnitudes of Emax and Wmax are both a lttle6 Zhang z C, Zhang J D. The investigation of highsmaller, but the additional magnitude of AD is aboutelectrical breakdown for pressurized water dielectricone time higher on account of the significant additionalvith microsecond charging. Proceedings of 15th In-magnitude of the effective time from pressurization ofternational Conference on High Power Particle Beams,1215.9 kPa than that from the ethylene glycol additiveSt.Petersburg, 2004with an ethylene glycol concentration of 80%. Conse-Shen L G. Introduction of pulse power technology.quently, pressurization is more effective for improvingChangsha: National University of Defense Technol-the electrical breakdown strength.ogy, 1989 (in Chinese)Sincerny P S. Electrical breakdown properties of waterConclusionsfor repetitively pulsed burst conditions. Proceedingsof the 3rd IEEE International Pulsed Power Confer-In the condition of an effective time at a microsecondence. Albuquerque, New Mexico, 1981. 222 ~ 225scale and a low bydrostatic pressure, the breakdown9 Jones H M. Kunhardt E E. IEEE Trans. on Dielectricsstress formula is modified to the following form:中国煤化工1(6): 1016~ 102510mal of Physics, 1995,E = 0.561MA-1/10t_l/N p1/8,(5)FYHCNMHGwhere M is the coefficient depending on the quality(Manuscript received 14 March 2005)of water and on the electrode characteristics, with 168E-mail address of Zhang Zicheng:for a positive breakdown of pure water with a goodsuccesschang@163.com3165