11-10-2016, 02:34 PM
Self healing Metalized polypropylene Film Capacitor and its Uncoupling behaviour of current gates
1458638384-AMCCONFERENCEPAPER..docx (Size: 61.38 KB / Downloads: 6)
Abstract
Polymer dielectrics have largely supplanted other insulators such as paper, because of their excellent dielectric and physical properties, low cost and availability in very thin films. The most important of these are polypropylene, polycarbonate, polystyrene, polyethylene, and polyethylene teraphtalate (PET). Recent high power capacitor technology use thin polypropylene (PP) foils as a dielectric with 15 nm thin patterned electrodes instead of all-over metalized films. The metal electrode consists of individual segments interconnected by narrow current gates. The gates serve as fuses in case of a breakdown in one of the segments. They isolate the segment and therefore the breakdown channel from the rest of the electrode. Therefore the capacity only decreases slightly and the capacitor is protected against complete destruction. This process is called self healing. Capacitors in operation showed that in case of a breakdown not only the defect segment but also the surrounding and the distant segments are often uncoupled, leading to a higher decrease of capacity and consequently of the capacitor lifetime. The aim of the study was to understand the mechanism that breaks off distant current gates. Therefore we stressed current gates with low voltages and currents, determined the energy involved in the uncoupling process and investigated the broken gates with light microscopy. Resistance curves gave important information about the influence of structures at the PP foil surface on the uncoupling behaviour.
Keywords: Self healing; Current gate; Polyethylene teraphtalate foil; Polypropylene foil.
1. Introduction
The primary polymer foil materials used in capacitors are biaxially stretched oriented isotactic polypropylene (PP) and polyethylene teraphtalate (PET). Today high power capacitor technologies use patterned electrodes evaporated on the dielectric instead off all-over metalized films. In a capacitor usually two metalized foils are wound together to a capacitor winding. To avoid air enclosures in the capacitor rolls causing glow discharges in the winding gaps, oil or gas is used as impregnation medium. The excellent large area electrical breakdown strength of PP foils due to the high quality with which they can be made consistently and the homogenization effect taking place with oil impregnation of PP foils [1] are the major reasons why such capacitors can be operated at very high electric fields in the range of 240 V µm-1.
Fig.1 shows mosaic of metallised capacitor film. In self healing capacitors, the electrodes are evaporated onto the polymer foil as very thin films of metal. The metal used is typically aluminium, zinc or a zinc– aluminium alloy. In case of a breakdown through the dielectric, the thin electrode near the defect site is rapidly evaporated and driven outwards from the breakdown site. Thus the plasma of the breakdown arc is interrupted and the site becomes electrically isolated. These are the series of events described as self healing [2,3]. This self healing process makes the system defect tolerant since local breakdowns cause only little damages. The choices of the metallization as well as the thickness of the electrode are important parameters determining the self healing capability of the system. The thinner the layer is, the more likely the self healing will be successful. Unfortunately, if one goes too thin, the electrode resistance increases leading to losses of heat. Therefore the thermal stress of the capacitor increases and impairs the capacitor lifetime [4,5].
Fig. 1: Patch mosaic of metallised capacitor film.
Incomplete evaporation of the electrode around the breakdown channel and deposition of carbon from the dielectric, which give rise to conductive bridges in the insulating areas free from metal present a danger for continually discharging of the capacitor [6]. Therefore the patterned electrode reveals a second protection concept. The individual segments are interconnected by narrow current gates. In case of a breakdown in one of the segments the gates serve as fuses, they isolate the segment and therefore also the breakdown channel with the surrounding evaporated electrode area form the rest of the electrode. Thus the damage of the capacitor is double localized and the capacitor is prevented from continually discharging leading to large-scale damage and eventual destruction.
Fig.2: Pictures of damaged segments after a breakdown illustrating the self healing process. (a) Self healing process in which only one segments uncoupled by blowing of its four current gates. (b) Uncoupling of the surrounding and distant segments from the rest of the electrode.
Fig. 3: Cylindrical capacitors before application of schooping end–spray.
Fig. 2 displays two pictures of damaged segments after a breakdown illustrating the self healing process. In case of a breakdown a high current flows through the next current gates in direction of the defect forming the breakdown channel. Due to the high breakdown current, the current gates are evaporated or even explode. Fig. 2 a displays the self healing process in which only one segment is uncoupled by blowing one of its four current gates. This corresponds to the ideal case with minimal loss of electrically active electrode area. Fig. 2b shows the damage mostly observed in practice for capacitors operated at higher electrical fields. In this case not only the defect segment but also the surrounding and distant segments are uncoupled from the rest of the electrode, leading to an increased capacity loss and consequently to a faster decrease of the capacitor lifetime. It is of fundamental importance for an optimal self healing process that in case of a breakdown always only the defect segment and under no circumstances more segments are uncoupled from the electrode. Thus the lifetime depends decisively upon the functional ability of the current gates.
Fig. 4. Set-up to stress the current gates.
Up to now it was thought that the current gates break off by evaporation due to the high currents flowing through the gates. Based on this assumption, the size of the current gates was optimized, guaranteeing good uncoupling properties and low losses. This is justified for the current gates to the defect, where a high current is evaporating the metal film. On the other hand the mechanism leading to the uncoupling of distant current gates can be different, as the current flowing through these current gates in case of a breakdown is significantly lower. In this case beside the evaporation of current gates also mechanical stress due to electrode heating leading to detachment of the current gates is conceivable. To investigate the uncoupling behaviour of such distant current gates in detail and to find out the mechanism that breaks off the gates, we stressed current gates of different, segmented metalized PP and PET foils with low voltages and currents.
Fig..5. Resistance behaviour of a stressed current gate calculated from voltage and current curves.
Fig. 6. (a) Light microscopy picture of a PP1 current gate region after blowing of the current gate. (b) Corresponding resistance behaviour during the uncoupling process of a PP1 current gate.
2. Experimental
The investigations were made on three different commercial, segmented metalized isotactic PP foils and on a commercial, patterned PET foil. The different PP foils are designated as PP1, PP2 and PP3. The thickness of the electrode is about 15nm. The depth profile of the electrode measured by means of photoelectron spectroscopy (XPS).
Fig. 7. (a) Light microscopy picture of a PP1 current gate region showing almost no polymer surface structures after the blowing of the current gate. (b) Magnified section of the burn out channel. © Corresponding resistance curve.
The individual segments, spaced from each other by 300 mm, are connected by narrow current gates in the range of 400 mm. Two different current gate arrangements were considered. In case of the PP1 and the PET foil the current gates are located on the diagonal of the segments. The PP2 and PP3 foils have current gates located at the side of the segments.
The investigated capacitor foils reveal different surface topographies. PP1 foils show a smooth surface with 1–2 mm long polymer strings embedded at the foil surface which give the foil the necessary roughness for the oil diffusion into the winding gaps. The oil diffusion into the winding gaps is crucial for best dielectric strength results [1] because it avoids air enclosures in the capacitor rolls causing glow discharge. PP2 foils reveals a regular rough surface of about 3 mm roughness which permits an optimal oil diffusion into the winding gaps. PP3 foils show a smooth surface with a roughness below 1 mm. Such foils are equal to the PP1 foils, however without polymer strings at the foil surface. The PP3 foil was designed for gas impregnated capacitors. Finally, the PET foil reveals also a smooth surface with thin spacers in mm height range at the foil surface permitting the oil diffusion into the winding gaps.
The current gates were stressed using a set-up described in Fig. 4. The contacts are placed in a distance of 0.3 mm to the current gate. By the discharge of a 80 mF capacitor charged to VCap_40 V electrical stress was applied to the current gate. With an oscilloscope the voltages over the current gate VC as well as the current curves AC were measured. For each measurement the resistance curve and the energy curves. To investigate the current gates in similar conditions as in the capacitor, the gate was protected from atmosphere with an oil film and an impressed PP foil. The broken gates were examined with a light microscope.
3. Results and discussion
3.1. Self healing
One of the first investigators to study self healing was Heywang [7]. He characterized the self healing process by measuring the discharge current pulse during self healing events and by calculating the energy involved. The results show that the self healing events and by calculating the energy involved. The results show that the self healing of a breakdown takes place within 1–5 ms for operational voltages around 150 V. Reed et al. [2] showed that the time for a clearing to take place at 400 V is roughly 20 ms. Therefore the time is strongly dependent on the voltage at which the self healing occurs. At high electric voltages of 3000 V applied on the metallized foils self healing times up to several 100 ms are extrapolated. This is probably because of the higher discharge energy and the longer demetallized path at higher voltages. This view is supported by the observation that the demetallized areas are noticeably larger at higher voltages. During the self healing process high currents flow through the current gates causing the blow of the gates next to the damaged segment. Moreover current gates of surrounding and distant segments are uncoupled from the rest of the electrode, leading to a faster decrease of the electrode area. To minimise the loss of electrically active electrode area in case of a breakdown the strength of the current gates has to be that high, so that distant current gates get over the electrical stress during self healing.
Fig. 8. (a) Light microscopy picture of a PP2 current gate region after blowing of the gate. (b) Resistance curve of the stressed current gate.
3.2. Resistance behavour of current gates
To simulate the conditions prevailing for distant current gates in case of a breakdown, current gates were stressed with low voltages and currents. The resulting resistance curve allows determining the breakdown behaviour in detail. The electrode heating , the loss of current gate area due to detachment or evaporation as well as the behaviour with regard to the heat conductibility of each electrode-polymer system affect the shape of the resistance curve and can therefore be investigated. In addition, light microscope measurement allow to perform damage analysis of the broken current gates. Fig. 3 shows a typical resistance behaviour of a stressed current gate of a PP foil. The resistance curve shows two phases, which are described by a linear behaviour in the first part of the curve and a power dependence for higher uncoupling times. The two phases will be discussed in the following:
3.2.1. Phase I: warm up phase
Due to the current gate heating the resistance over the current gate is increasing. This linear increase of the resistance is described by the relation R_(1_aDT)R0
where R0 is the initial resistance over the current gate at room temperature and a is the temperature coefficient. For the Zn_Al electrode a amounts to 4.2_10_3 1 K_1 [8]. The slope of the resistance curve allows to determine the temperature of the current gate. This implies a first indication whether the current gates are broken by evaporation or not. At the resistance RU (Fig. 3) the linear behaviour of the resistance curve changes into a power dependence. Therefore RU indicates the beginning of the uncoupling process and the end of phase I. The calculated temperature increase of the current gate during the warm up phase is T_60°C.
For the metallized PP foils R0 amounts to 36 V, while the effective resistance of the current gate measured directly at the border amounts to 15 V. Considering this value the effective temperature of the current gate metal layer at RU amounts to 170°C. This value is far too low to evaporate the current gates. Based on vapour pressure curves for Zn a temperature of 400°C is needed to evaporate the gates [9]. For Al this temperature is even higher. This result represents a first indication that the uncoupling of the current gates is not caused by evaporation.
3.2.2. Phase II: uncoupling phase
This phase describes the uncoupling process of the current gates. During the formation of the uncoupling channel the cross section of the current gate is strongly decreasing. As the resistance is inversed proportional to the cross-section of the current gate, the resistance curve over the current gate is strongly increasing with decreasing cross-section. Finally, the current gate is broken and the two segments are coupled from each other. The uncoupling time t* indicated in Fig. 3 defines the period from the beginning of the current stress until the uncoupling of the current gates and represents therefore a degree for its strength.
Fig. 9. (a) Light microscopy measurements of a damaged current gate on a PET foil. (b) Magnified section of the second damaged current gate.© Corresponding resistance curve of a stressed current gate.
3.3. Uncoupling time t* of current gates
As the uncoupling time t* represents a degree for the strength of the current gate we determined t* for all investigated metallized foils. In Fig. 10 the uncoupling times t* results are presented. The presence, the nature and the location of surface structures decisively affects the uncoupling time t*. For the PP1 foil we distinguished between current gates showing numerous polymer strings and current gates showing less or almost no strings. Depending on the number of polymer strings present in the current gate region the uncoupling time can differ up to the multiple of a magnitude. The average uncoupling time for PP1 foils amounts to 230 ms and reveals therefore lowest t* results. Individual measurements of current gates showing a great number of polymer strings yield uncoupling times down to 40 ms. Due to slight differences in current gate size, PP1 foils with current gates over the edges of the electrode segments reveal much higher uncoupling times (t*_ 1.34 ms) as PP1 foils where the individual segments are connected on the side by narrow current gates (t*_469 ms).
PP3 foils showing current gates on the side of the
segments, reveal an average uncoupling time t* of 2.6 ms. Such high values for t* are also obtained for PP1 foils showing no polymer strings in the current gate region. In contrast to this, the PP2 foil yields an average uncoupling time t* of 630 ms. Finally, the PET foil reveals an average uncoupling time of 1.1 ms, comparable to the average uncoupling time t* of PP1 foils PP1 foils in operation on a TGV for over one year reveal an average uncoupling time that is about 40% lower (0.85 ms) than that of PP1 foils without any treatment (1.34 ms). Therefore the uncoupling behaviour of the current gates is significantly influenced by the operational conditions in the capacitor. Oil diffusion, heating during operation as well as the applied electrical stress on the PP foil affect the functional ability of the current gates. The influence of temperature on the uncoupling behaviour was investigated for PP1 foils. The measurements reveal at room temperature (RT) an uncoupling time t* of 1.34 ms. Foils stored at 50 and 100°C show average uncoupling times t* of 980 and 550 ms, respectively. These results point to a significant dependence of the uncoupling behaviour on the surrounding temperature, what is not very surprising due to the fact that the fuses are broken by current gate heating.
Fig. 10. Uncoupling times t* of current gates under applied stress for different capacitor foils.
As discussed, the uncoupling process of current gates takes several 100 ms at high electric voltages of 3000 V applied on the metallized foils. The uncoupling time results show that even for the low stresses we applied to reproduce the conditions prevailing for distant current gates in a capacitor in operation, the determined self healing times are long enough to break the current gates of PP1, PP2 and PET foils. Only the PP1 foil with no polymer foils at the current gates as well as the PP3 foil reveal uncoupling times t* for distant current gates that are long enough to get over the stress during self healing. We conclude that surface structures which are necessary for the oil diffusion have the disadvantage to strongly reduce the uncoupling strength of current gates.
3.4. Uncoupling energy
The theoretical value for the energy Eeva needed to evaporate the electrode was calculated on the basis of the enthalpy of vaporization values for Zn and Al found in Refs [8,11]. Eeva for the Zn_Al electrode with 15 nm thickness amounts to 0.8 mJ mm_2. To compare the theoretical value with the experimental energy involved in the uncoupling phase Eu, we stressed current gates of segmented metallized PP2 capacitor foils with different voltages VCap in the range of 40–240 V corresponding to currents AC of 0.3–2 A, respectively. Higher applied stress allows to study current gates closer to the segment showing a breakdown through the dielectric. By light microscopy the area of the uncoupling channel was measured and by means of the calculated energy curves we determined the energy involved in the uncoupling phase of the current gates. The light microscopy measurements of the current gates stressed at higher voltages and currents are presented in Fig. 11. The corresponding calculated uncoupling energy values and the uncoupling times of the current t* gates are presented in Table 1.
Higher applied voltages reveal an increased area of the uncoupling channel. At the border of the uncoupling channel detached electrode pieces are visible. For higher applied stresses the uncoupling time t* decrease from ms down to values in the ms range. The energy Eu involved in the uncoupling phase is strongly decreasing for increasing electrical stress. Therefore the uncoupling process is not adiabatic and cooling of the current gates is of significant importance during the uncoupling process. The measurements show that for an applied stress of 240 V and 2 A the experimentally determined energy to separate the segments (Eu_0.66 mJ mm_2) is lower than the theoretical value Eeva. These results represent a third indication that distant current gates are not broken by evaporation.
3.6. Mechanical tensions between the electrode and the dielectric
PP as well as PET foils revealing uncoupling times higher than several 100 ms show along the burn out channel large detached electrode areas. In this regions the metallized electrode is completely separated from the polymer films (TM(PP)_170°C, TM(PET)_130°C [17]) compared to the Zn_Al metallization mechanical stresses are assembled in the current gates. Investigations of mechanical properties on biaxially stretched PP foils have shown that the capacitor foil after heat treatment at 100°C reveals mechanical relaxations [18]. Due to the heat treatment the foil reveals in one direction shrinking of 10% and in the other foil direction expansion of 10%. Thereby the differences of expansion for a 400 nm large current gate at the interface between the PP foil and the electrode (aAl_1.7_10_5 K_1 [8]) at a temperature of 100°C amount to 43 and 37 nm, respectively [19]. In Section 3.2 we determined an average uncoupling temperature of 170°C for PP foils. This temperature corresponds exactly to the melting temperature of the investigated PP foil [1,17]. At 170°C the PP foil begins to melt and therefore adhesion between the metallization and the dielectric is strongly affected. For the metallized PET foil we determined an uncoupling temperature of about 130°C. As well as for the PP foil this temperature corresponds exactly to the melting temperature of PET. We conclude that the considerable mechanical tensions at the electrode-polymer interface in case of applied electrical stress.
Table 1
Calculated uncoupling energy values and uncoupling times t* of PP3 current gates stressed at higher voltages and currents.
PP3 foil
Eu (mJ Mm_2)
t*
Vcap = 40V, Ac = 0.3A
90
3ms
Vcap = 60V, Ac = 0.5A
10
320µs
Vcap = 150V, Ac = 1.3A,
2
5µs
Vcap = 240V, Ac = 2A
0.66
2µs
4. Conclusions
The uncoupling process of distant current gates of high power self healing capacitors was investigated in detail by the resistance behaviour of current gates in case of low applied electrical stress and optical damage analysis for different, segmented metallized capacitor foils. The resistance curve of stressed current gates can be assigned into two phases which describe the different steps during the uncoupling process. Light microscopy measurements as well as the resistance behaviour of current gates show that the uncoupling process is strongly influenced by the polymer surface structures in the current gate region. If over the current gates showing surface structures at the foil surface electrical stress is applied, mechanical tensions are formed along the surface structures due to the electrode heating. Therefore the uncoupling channel is following these weak points, reducing drastically the uncoupling time t* of the current gates. Uncoupling times down to 40 ms are observed for low applied stress corresponding to the conditions as they prevail in distant current gates. In case of provoked breakdowns at higher applied voltages self healing process times up to several 100 ms result. These times are long enough to break the current gates of the surrounding and even distant current gates. Thus numerous segments are uncoupled from the capacitor in case of a breakdown, leading to a faster decrease of the capacitor lifetime. Further measurements point to a significant dependence of the uncoupling time on the capacitor operational temperature. At higher temperatures the uncoupling time t* of current gates is strongly reduced. The functional ability of current gates during self healing is decisively affected by adhesion failure along the polymer-electrode interface. Polymer foils with surface structures show that due to the electrode heating in case of breakdown mechanical tensions are assembled mainly along the surface structures. Therefore loss of adhesion occurs at the interface and the uncoupling channel is formed along the surface structures. If no surface structures are observed in the current gate region, the differences of expansion at the interface between the dielectric and the electrode as well as the low melting point for polymer foils determine the uncoupling behaviour during self healing. Thus during electrode heating strong mechanical tensions are assembled along the interface, affecting detrimentally the adhesion between the dielectric and the electrode. Such foils reveal along the uncoupling channel a large electrode area which is completely detached from the dielectric. In addition, the formation of weak boundary layers due to oil diffusion gives rise to adhesion failure at the interface.
Damage analysis of capacitor foils in operation have shown two types of broken current gates. First, due to the high electric field over the current gates next to the damages segment an arc is formed during self healing, whereby the current gates are evaporated. Second, due to the lower electric field prevailing for current gates connecting segments located far from the damaged segments, no arc is formed over distant current gates. Our investigations have shown that distant current gates are broken by detachment of the electrode from the polymer foil due to temperature induced mechanical stresses along the surface structures and the polymer electrode interface. From the results presented in this work, guidelines can be taken to improve the strength of current gates and therefore to minimize the loss of electrically active electrode area in case of a breakdown. A possibility to achieve higher current gate strength may consist in connecting the individual segments by larger current gates, whereby the resistance, the electrode heating and consequently also the mechanical stress in the current gates is considerably reduced.
Acknowledgements