Differentiation of roughness and surface defect impact on dielectric strength of polymeric thin films

: Increasing the dielectric strength of polymer films has been a key theme as it is directly responsible for increasing energy density of relevant components such as film capacitors and insulation tapes. Dielectric films with higher roughness and surface defects are subject to the formation of an air gap at the interface between dielectric film and metallised polymer electrodes, which results in inaccurate dielectric strength. The air gap due to roughness was found to result in dielectric strength of 25% higher than that using depositing metal on dielectric films (integral electrode). The integral electrode method is proven to be a better way to test the genuine dielectric strength of thin and rough dielectric films. Surface defects, on the other hand, were revealed to cause lowering of dielectric strength because of their contribution to the localised electric field and charge injection. The detrimental effect of surface defects can be suppressed by submerging the film in oil or coating the film with an oxide layer.


Introduction
As one of the most critical properties for a dielectric, a higher dielectric strength represents a better quality of an insulator and is highly desirable for such applications as electrical power systems, power electronics, capacitors, cables etc. Traditionally, the shorttime method for dielectric breakdown voltage is defined according to ASTM D3755-14 Standard (under direct voltage) and ASTM D149 Standard (under alternating voltage of commercial power frequencies) [1]. Either a ball-plane test fixture or parallel metal pucks are used for breakdown voltage that are applied across specimens of thickness mostly higher than 0.8 mm. As thinner dielectric materials are being developed, these methods are still used, but subject to limitation in showing genuine dielectric strength due to smaller active test areas and the increased contribution of film surfaces [2,3]. In fact, polymer films possess three types of issues (wrinkles, roughness and defects) that complicate the measurement of dielectric strength. To minimise the area effect and the mechanical force effect for thin films, metallised polymer films were used as popular electrodes in the last decade [4][5][6][7][8][9][10]. The popularly used metallised polymer film electrode is metallised polypropylene (PP) or high-temperature polytetrafluoroethylene (PTFE) with metal facing the test samples with an active area of >2 cm 2 [11]. When a voltage is applied, the layers are drawn together by electrostatic force, creating close contact. This method has now been extended to films even thinner than 10 µm; however, the surface features such as roughness and defects have not been well-classified. The difference in their impact on the dielectric strength of dielectric thin films is not wellunderstood.
Future's more electric power applications require higher-voltage stress, thinner dielectric, higher working temperature for a dielectric material [12,13]. The contribution of film thickness variability, surface defects (e.g. contaminants, pits, dimples etc.) becomes more important with decreasing film thickness. It is a valid concern simply applying metallised polymer electrodes to a dielectric film sample without considering the effect of surface roughness and defects before obtaining a genuine dielectric strength. Fig. 1 shows the schematic contact configuration and a rough surface image of a metallised polymer electrode. Obviously, the electrode surface has quite high roughness (average roughness Sa∼0.166 µm and peak-valley roughness Sz∼1.528 µm). Even after an attractive force exerted by an electric field brings electrode and dielectric sample closer, the rough surface from both parties still leaves air gaps in between. This is similar to a metal puck electrode being well-polished down to an average roughness of 0.2 µm [14]. Films with smaller roughness are preferred to be made as a metallised polymer electrode. Indeed, the metallised polymer technology was already used in a metallised film capacitor to enable self-healing characteristics and graceful failure [15,16]. It makes more sense to leverage a similar method of depositing the metal onto the dielectric film for characterising dielectric films. However, the genuine dielectric strength test is made more complicated when both roughness and surface defect co-exist. Research so far have not shown a clear picture of how roughness influences the dielectric strength of polymer films thinner than ∼6 µm. Moreover, they also failed to differentiate the impact of roughness and surface defects [17,18].
Polarisation-electric-field relationship has also been commonly used as a method in various publications to determine the dielectric strength and energy density of a polymer film [19][20][21][22][23]. However, the test results are not reflecting the impact of surface roughness because of small active electrode areas in the tests. It can be said that the entire dielectric community has not really understood the difference in the effect of roughness and surface defects on dielectric strength. This communication will reveal the different responses of dielectric strength associated with roughness and surface defects, justify the reliable metallised film method for the characterisation of polymer films thinner than ∼6 µm, and also deepen the understanding of breakdown mechanism in polymer dielectrics.

Experimental
Dielectric films used in the work are polyimide (PI), polyetherimide (PEI), polyphenylene sulphide (PPS), PP, polyethylene-naphthalate (PEN) and PTFE polymers that were obtained from various sources such as Bollore Inc., Mitsubishi, Toray, Dupont, Saint-Gobain, DeWal Industries and Gore. Their thicknesses are in the range of 3-10 µm. A modified test fixture containing an array of 12 samples is utilised for this work [8]. The surface roughness is characterised using Keyence VK-X100 Series Laser Confocal Scanning Microscopy. Dielectric film surfaces are metallised with either gold (Au) or aluminium (Al) of at least 0.55 × 0.55 in in area. The electrodes of >100 nm in thickness are thermally deposited using a Denton evaporator.
An integral electrode method made of metallised dielectric film and metallised polymer cantilever/interposer electrode was designed in work to overcome the rough surface issue. As shown in Fig. 2, multiple independent tests can be performed on the array of samples through moving around the high-voltage electrode on the top while using metallised polymer films as a ground electrode. Ball-plane test method was also used for PEI films to understand the surface defect impact.

Results and discussion
Most of the past research reports paid less attention to the roughness impact on dielectric strength primarily for testing thick films. The author inspected the roughness in various polymer films of thinner than 10 µm using a laser confocal scanning microscope (LCSM) tool. Table 1 shows the details about the roughness of these films. These common dielectric films exhibit various surface roughness, particularly big peak-valley (Sz) roughness value. The metallised polymer electrodes are made of Al-deposited PP, PPS or PTFE film of about 5-25 µm in thicknesses. Fig. 3 shows the examples of representative morphologies of HCPP and PTFE films. Ridges and features can be seen clearly on these films. Higher roughness can be observed on PTFE film due to the existence of fibril and ridges and nodes (not shown). Therefore, the use of metallised polymer electrodes for dielectric strength test results in a non-conforming contact with rough dielectric samples leaving certain air gaps or pockets at the interfaces. Such a gap containing interfaces becomes part of the voltage divider to the applied electric field. Then, higher dielectric strength is expected for rough dielectric films (HCPP and PTFE) as shown by the left bar in Fig. 4.
When using the integral electrodes, the author found no statistically significant difference using one side or two sides of metallisation, and thereby the breakdown tests are conducted only by depositing a metal layer on one side of the film samples. This also provides a metal-dielectric-metal construction for simulating the actual operation environment inside a capacitor and getting the genuine breakdown information of the dielectric film. There are no air gaps involved, and thus no fake contribution from the roughness. It is hard to find the PP films of different roughness, but the same crystallinity, orientation and resin materials. Both PP and PTFE films of different roughness are used to illustrate the effect (Fig. 4). For smoother PP films, the difference in breakdown strength between using two different electrodes is smaller. In the rougher PTFE film, the bigger difference of more than 25% results as shown by the right bar in Fig. 4. Therefore, the integral electrode method is more reliable in determining the genuine dielectric strengths of thin polymers.
The removal of the air gap in the integral method was verified by measuring the capacitance change of the dielectric samples on electrical energising. Fig. 5 shows the results of PTFE films using three different electrode fixtures. The increase in capacitance with increasing bias field indicates more contact area of the electrodes with the dielectric sample. After passing a threshold field, the electrostatic force exerted by the voltages is high enough to narrow down the gap for more proximity contact in each case. The onesided gold Au electrode is still subject to the residual air gap from   the ground electrode, and thus exhibits similar climbing trend in the capacitance change. The two-side metallised dielectric film samples exhibit the highest capacitance implying the maximum contact areas. These results prove that the proximity in electrodedielectric film is critical and the roughness-caused air gap does contribute to the additional voltage withstanding capability. Integrated electrode can show the genuine dielectric strength of rough dielectric films. It is to be pointed out that the partial discharge in the air gap is also part of the breakdown process at higher voltages causing occurrence of more self-clearing events in capacitors using rough dielectric films.
On the other hand, surface defects exert different dielectric breakdown behaviour on dielectric thin films. With decreasing film thickness, the wrinkle-free polymer dielectric with similar roughness presents a transitional thickness-dielectric strength relationship as shown in Fig. 6. The dielectric strength stops increase with decreasing the thickness of polymer films below certain threshold (e.g. ∼8 µm). All thin films exhibit lowering of dielectric strength with further decrease in film thickness, which is opposite to the effect of roughness revealed by using the metallised film electrodes. This is associated with the increase in the contribution of surface defects in ultrathin films. These films are either manufactured by solvent cast subjecting to solvent evaporation or melt extrusion subjecting to slip agent and die line issue, which increases with thickness reduction. Figs. 7b and c show an example of typical defects of a 4 µ-PEI film roll (a) with roughness of Sa∼20 nm and Sz∼78 nm. These defects generated in the film production process were detrimental to its voltage withstanding capability if they are not contained. Fig. 8 shows the DC dielectric strength when the PEI films were tested in the oil and open air. When testing in the air, lower breakdown strength and wider distribution appear because the defects can be the sources of plasma erosion and charge injection under electric field. Testing in oil, however, suppresses the localised field and plasma erosion, giving rise to a higher breakdown strength and shape factor.
To effectively suppress the contribution of these defects, the field stress localised at the defects and the induced charge injection ought to be lowered. According to Fowler-Nordheim tunnelling theory, the local charge current density is proportional to the electric-field square and inversely proportional to the work function of the electrode-polymer interface [J∼E 2 /Φ exp (−CΦ 3/2 / E)]. Coating an oxide layer (>50 nm) increases the barrier thickness for tunnelling at the defects. The high-K passivation also reduces the maximum local field strength at the defects as it is proportional to the ratio of dielectric flux density and permittivity. Therefore, the electric field (E) in the equation is decreased, and lower charge injection and dielectric breakdown events are expected. The positive effect has been achieved in the author's research work on coating silicon dioxide layer [24]. With >50 nm thick coating, dielectric strength can be increased by more than 20%, which can be a promising pathway for enhanced dielectric strength.

Conclusions
Decreasing film thickness faces challenges of increased surface roughness and defects. The use of conventional metallised polymer electrodes has the limitation for revealing the genuine dielectric strength of dielectric films because of the air gaps or uneven interfaces between the rough dielectric film and the metallised polymer electrodes. The fake higher dielectric strength can be avoided by depositing metal electrodes onto the active areas of dielectric films. The metallised dielectric area may be utilised on one side or both sides of the dielectric films to exclude air-gap contribution, which is confirmed by capacitance measurements.
Surface defects results in lower dielectric strength, which is exhibited by performing tests in the air in comparison with test in the oil. The lowered dielectric strength issue was believed to be related to the localised field stress and charge injection at the defects can be avoided by a nano oxide coating on the polymer film. The inorganic coating can be a good path to increase dielectric strength and high-energy density of polymer films.

Acknowledgment
This work was supported by the Guangdong Technion Israel Institute of Technology, and Guangdong Basic and Applied Basic Research Foundation -2019A1515012056.