Analysis of the formation of gap-based leakages in polymer-metal electronic systems with labyrinth seals

2.1 Materials used

In this research, hot-rolled sheets made of wrought aluminum alloy EN AW-5754 H24 (Hans-Erich Gemmel & Co. GmbH, Germany, Berlin) were used for the manufacturing of the metal inserts. Good electrical conductivity, low density and high corrosion resistance characterize this aluminum alloy, which are favorable for its usage in resilient polymer-metal electronic systems.

For the electrochemical manufacturing of an areal seal, sodium hydroxide (NaOH) with a concentration of 0.27% and HCl with a concentration of 5% were used. Both reagents were obtained from VWR International GmbH (Germany, Darmstadt).

For the injection molding, commercially available polyamide-66 granules with 30% per weight short glass fiber reinforcement (Ultramid A3EG6, BASF SE, Germany, Ludwigshafen), were used, which are commonly used in electronic and automotive applications. Following the manufacturer specifications, the granules were dried at 80°C for 12 h before processing.

2.2 Test specimen

In this study, the polymer-metal hybrid connector test specimen in Figure 1 was used for the analysis of the formation of gap-based leakages. The metal insert consists of a head part and a bottom part, which are linked via a rectangular crosspiece, Figure 1B. The crosspiece is the region of interest and resembles a flat connector, which is overmolded. The head and bottom parts are used to position and clamp the test specimen during processing and testing. The front faces of the crosspiece are 5 mm wide, and the side faces are 1 mm wide according to the sheet thickness. The area of the polymer component facing the head part will be referred to as the head side, whereas the part facing the bottom will be referred to as the bottom side. In order to investigate the formation of leakage and leakage paths, different variants of metal inserts without (reference [R]) and with various surface structures for the labyrinth seal were manufactured (Figure 2). In the following, the selection of the used structures is explained.

Figure 1:

Connector test specimen, referring to [19]. (A) Main dimensions of the metal insert and the polymer component. (B) Connector test specimen with runner system, which allows for the symmetrical filling of the polymer component along the metal insert with two gates at the flange.

Figure 2:

Schematic illustration of the insert variants 0 (A), I0 (B) and MS (C). The schematically marked material bulges of variants 0–300 (A) and I0–300 (B) are a result of the structure manufacturing and only formed by a structure depth of 0.3 mm. Detail E shows a scanning electron micrograph (produced with a scanning electron microscope of the type Ultra Plus from Carl Zeiss AG, Germany, Oberkochen) of the structured surface of variant MS, which is representative for all sides of the crosspiece.

In order to create simple leakage barriers, inserts with embossed structures across the front faces in the xy plane were overmolded. Two variants with different structure depths were investigated, labeled 0–100 for 0.1-mm-deep and 0–300 for 0.3-mm-deep structures (Figure 2A). Furthermore, the material displacement of the metal during the manufacturing of the structures on the metal insert by embossing was used for the investigation of leakage barriers at the side faces. However, only for a penetration depth of 0.3 mm (variant 0–300) a significant material displacement in the form of material bulges can be achieved (Figure 2A). A detailed description of the embossing process can be found in the next chapter. In order to realize continuous all-around grooves, variants with additional indentations and different depths were used, which are labeled I0–100 (depth of 0.1 mm) and I0–300 (depth of 0.3 mm). These variants were chosen to investigate potential effects of the side faces on the overall tightness (Figure 2B). The indentations were manufactured before the embossing process. Similar to the variants described before, only embossed structures with 0.3-mm depth (variant I0–300) lead to a significant formation of material bulges.

In order to achieve an all-over leakage barrier, the crosspiece of variant MS (Figure 2C) has a fully microstructured surface (areal seal). The microstructures feature several undercuts perpendicular to the metal surface, which, if filled with polymer during injection molding, prevent the formation of large gaps according to [18].

2.2.1 Insert manufacturing

The metal inserts of all variants were cut from sheets with a laser cutting machine (TruLaser Cell 7020, Trumpf GmbH + Co. KG, Ditzingen, Germany) to ensure smooth surface conditions at the cut surface without work hardening. Afterward, all metal inserts except for the inserts of the reference were structured by embossing or electrochemical treatment to obtain the different variants of labyrinth seals as shown in Figure 2.

2.2.1.1 Embossing

The embossing process includes two identically structured punches, where the lower punch is fixed while the upper punch is moved downward. The metal insert was embossed simultaneously from both sides, while two locating pins (Figure 3) position the insert [23]. The combination of a round and a slotted hole allows for elongation of the metal insert in one direction during the forming process. The embossing was performed with a universal testing machine (RM400, Schenck Trebel GmbH, Germany, Ratingen), with a forming speed of 10 mm/min. The penetration is stroke controlled and is varied between 0.1 and 0.3 mm using the same set of punches (Figure 3).

Figure 3:

Experimental embossing setup.

2.2.1.2 Electrochemical treatment

For the electrochemical treatment with HCl with a concentration of 5%, the same experimental setup as in [22], [23] (Figure 4) was used. The main part of the test setup is an acid temperature-controlling system, which is required to avoid heating of the acid as a result of the exothermic process. It consists of a peristaltic pump (Boxer 9700 Table Top Dispenser, Boxer GmbH, Germany, Ottobeuren), a heat exchanger (custom build, Glasbläserei B+S, Germany, Erlangen) and a cryostat (Lauda Ecoline RE 112, Lauda Dr. R. Wobser GmbH & Co. KG, Germany, Lauda-Koenigshofen). The cryostat was adjusted to 23°C with water as the cooling medium. In addition, a magnetic mixer (RET-G, Janke & Kunkel GmbH & Co. KG, Germany, Staufen) was used to enable circulation of the acid in the laboratory glass. Two metal sheets and a metal insert between them were immersed in the laboratory glass, which was filled with 800 ml of HCl, and connected to a direct current power supply (Voltcraft DPPS-32-30, Conrad Electronic SE, Germany, Hirschau). The distance between the metal sheets, which were connected as cathodes, and the metal insert, which was connected as the anode, amounted to 20 mm.

Figure 4:

Experimental setup for the electrochemical treatment of the metal inserts [19].

In order to obtain the same surface starting conditions at the beginning of the electrochemical treatment, each metal insert was prepared in three steps. First, the metal insert was cleaned with isopropyl alcohol in an ultrasonic bath (Haver USC 200-80, Haver & Boecker OHG Drahtweberei und Maschinenfabrik, Germany, Oelde) for 1 min, followed by venting until the isopropyl alcohol was vaporized. Following this, the natural oxide layer was removed with a bath in 50°C warm NaOH with a concentration of 0.27% until an active hydrogen development could be observed (40–60 s). Finally, residues of the NaOH were removed by rinsing with distilled water.

After the preparation, the metal insert blanks were electrochemically treated in the HCl solution for 150 s with a current density of 5 mA/mm². Following this, residues of the HCl solution were removed with distilled water in an ultrasonic bath (Haver USC 200-80, Haver & Boecker OHG Drahtweberei und Maschinenfabrik, Germany, Oelde) over a period of 60 s. At the end, the microstructured metal inserts were covered in plastic sample bags to prevent damage or contaminations of the microstructures.

2.3 Test methods

2.3.1 Differential pressure test

Because of its high relevance in industrial serial productions [7], the differential pressure test was used to measure leakage rate as an indicator of tightness. For this purpose, the test setup shown in Figure 5A was used, which consists of a leakage test system (Integra DD6, Dr. Wiesner Steuerungstechnik GmbH, Germany, Remshalden) and a suitable clamping device.

Figure 5:

Test setups for the characterization of the leakage rate. (A) Differential pressure test. (B) Air-under-water test.

The test sequence starts with the clamping of the test specimen into the clamping device, using toggle clamps for reproducible clamping. Afterward, a defined air excess pressure of 1 bar is applied to the reference and the test volume by the compressed air control system. After a period of 2 s, the so-called filling period, the compressed air control system is closed. In the following setting time of 3 s, the compressed air can calm down and adapt to the temperature of the test setup and test specimen. After that, the test and reference volumes are separated from each other by shutting the stop valve, followed by the measurement of the pressure difference between both volumes by a capacitive differential pressure transmitter over a period of 10 s. Finally, the pressure is released over a period of 4 s by the compressed air control system. In order to correlate the measured pressure differences with leakage rates, which do not depend on the test and reference volume, the test setup was calibrated with a tight reference test specimen and a test leak prior to each measurement series.

Within one measurement series, all test specimens of one metal insert variant were measured once. In total, each test specimen was measured five times because of a possible faulty clamping, which could lead to a measuring error not visible during the test itself. Through the comparison of the five measured leakage rates per test specimen, measuring errors can be identified, so that single measurements can be repeated. For this study, all results have been acquired with correct clamping conditions. For the calculation of the average value and standard deviation of the leakage rate of a variant, the mean value of the repeated measurements of the single test specimens was used. The tests were conducted in standard atmosphere (23°C and 50% humidity) according to DIN EN ISO 291. Immediately after these tests, all tests specimens were once again sealed in plastic foil impervious to water vapor and are stored until the air-under-water test started.

2.3.3 Microscopy

In addition to the air-under-water test for the determination of the leakage positions, cross sections of the bottom and head sides in the xy, xz and yz planes (Figure 6) of all the variants were produced to analyze the formation of leakages and the mechanism of action of the used labyrinth seals and the electrochemical manufactured areal seal by microscopy. For this purpose, a microscope (Axio Imager M2m, Carl Zeiss AG, Oberkochen, Germany) was used in combination with the software Imagic IMS from Imagic Bildverarbeitung AG (Switzerland, Glattbrugg) to measure gaps between the polymer component and the metal insert. In xz and yz planes, cross sections of the bottom and head sides of two test specimens per variant were analyzed. Because of difficulty in the preparation of the cross sections in the xy plane in the exact center of the metal insert, only the bottom and head sides of one test specimen per variant were analyzed.

Figure 6:

Measuring positions for the calculation of the characteristic gap sizes. (A) yz plane, distance between two points=250 μm. (B) xz plane. (C) xy plane.

In order to show the influence of gaps between the polymer component and the metal insert on the leakage rate, characteristic gap sizes L_lj and L_l were, respectively, calculated using Formula 2 and 3 at the defined measuring position l_ij, where the index i denotes the side of the metal insert and j, if necessary, an additional information about the position. The measuring positions l_ij (for example, a_I1) are defined in Figure 6, where index I denotes the left or upper side, and index II the right or lower side. The additional information j is only used for the characteristic gap sizes in the yz plane, where large areas are analyzed. In total, the 13 characteristic gap sizes L_aj (j=1, 2 … 21), L_bj (j=1, 2 … 5), L_c, L_d, L_e, L_f, L_g, L_h, L_k, L_m, L_n, L_o and L_p were calculated. The formula used depends on the direction of the shrinkage of the polymer and, thus, on the measuring position. Formula 2 was used for the characteristic gap sizes L_aj, L_bj, L_d, L_m, L_g, L_n and L_o, where the gaps on both sides are independent from each other and which can be more pronounced on one side than on the other. Formula 3 was used for the characteristic gap sizes L_c, L_e, L_f, L_h, L_k and L_p, where the gaps on both sides depend on each other and should theoretically be equal.

(2)Llj=∑i=12lij  or  Ll=∑i=12li , respectively

(3)Ll=12∑i=12li  

3.1 Tightness

The leakage rates of the different variants are shown in Figure 7A for both test methods: the differential pressure test and the air-under-water test. It can be seen that variant R has a high leakage rate of 0.28±0.29 ml/min, measured with the differential pressure test, which is in the same range compared to previous investigations with the same test setup, materials and almost equal process parameters [18]. The air-under-water test shows a marginally higher average leakage rate of 0.34 ml/min, but with 0.28 ml/min approximately the same standard deviation. Thus, the reference (variant R) is comparable to conventional maximum permissible air leakage rates of 0.5–12 ml/min [4], not reliable to be watertight for the most demanding requirements. The observed leakage positions of variant R and the other variants are illustrated in Figure 7B. It is noticeable that air bubbles were emitted over the entire circumference of the metal insert, with the exception of the corners and the middle sections of the front faces of the metal insert.

Figure 7:

Leakage rates, measured with both differential pressure and air-under-water test (A) and the observed leakage positions (B) during the air-under-water test of all the variants.

In contrast to the leakage rate of variant R, all other variants can also fulfill the most demanding requirements (Figure 7A). In addition, the different test methods show almost equal results, which is why they are not discussed separately in the following. The best results can be achieved with the microstructured variant MS, where no leakage rate could be measured with the used leakage tests. Considering the measurement resolution, the leakage rate of variant MS has to be less than 0.005 ml/min and therefore beyond measuring resolution. Furthermore, only at one test specimen, a single and tiny air bubble was observed at the middle section on one side of the metal insert (Figure 7B). The variants 0–100, I0–100 and I0–300 also yielded very good results with low leakage rates of approximately 0.01±0.01 ml/min. Only variant 0–300, with a structure depth of 0.3 mm, shows a higher leakage rate compared to variant 0–100, with a structure depth of 0.1 mm. In all four embossed variants, the leakages were concentrated at the corner sections of the front faces of the metal insert. However, no air bubbles were observed at the corners of the metal inserts.

Based on the results of the leakage tests and the observation of the leakage positions, the following conclusions can be drawn for the embossed structures. First, all variants with embossed structures have a positive effect on the leakage rate compared to the untreated reference. Since variants I0–100 and I0–300, with lateral indentations, show similar leakage rates, it can be concluded that the structure depth has no relevant impact. Furthermore, the design of the additional leakage barriers at the side faces has an influence on the leakage rate, but only in a small range (Figure 7A), variants 0–100, 0–300, I0–100 and I0–300. Here, the distinct material bulges of variant 0–300 seem to have a negative effect on the leakage rate. The reasons for these effects are explained by the microscopy results in the next chapter.

3.2 Microscopy

In Figure 8B and D, the characteristic gap sizes of all variants are shown for all positions for the head (B) and bottom (D) sides in the yz plane. Only for variant MS were no gaps measured, which is in accordance with the good results during the leakage test and corresponds to [18], where the formation of no gaps is explained by the polymer-filled microstructures with undercuts perpendicular to the metal surface. Furthermore, it can be seen that the gap profiles of variant R and those with embossed structures, which are respectively built by the characteristic gap sizes, are very similar. This is valid for both the head and bottom sides. Based on these gap profiles, especially at the bottom side, four representative sections can be defined: three for the front faces and one for the side faces of the metal insert (Figure 8B and D). The average values a1, a2, a3 and b1 of the respective characteristic gap sizes of the representative sections A1, A2, A3 and B1 are shown in Figure 8C and E, which facilitate the comparison of the variants.

Figure 8:

Characteristic gap sizes and average values a1, a2, a3 and b1 of the respective characteristic gap sizes of the representative sections A1, A2, A3 and B1 in the yz plane. (A) Measuring positions for the yz plane as orientation. (B) Characteristic gap sizes of two test specimens for the head side and definition of the representative sections. (C) Average values of the characteristic gap sizes of L_aj and L_bj for the representative sections of the head side. n.m. denotes that no gap could be measured. (D) Characteristic gap sizes of two test specimens for the bottom side and definition of the representative sections. (E) Average values of the characteristic gap sizes of L_aj and L_bj for the representative sections of the bottom side. n.m. denotes that no gap could be measured.

At the head side, gaps can be measured at each position for variant R and for those with embossed structures between L_a1 and L_a21 as well as L_b1 and L_b5 (Figure 8B). At L_a1 and L_a21 as well as L_b1 and L_b5, which represent the corner areas of the metal insert, only very small or no gaps could be measured. The formation of these gaps at the front faces and side faces of the metal insert can be explained by the thermal contraction and crystalline shrinkage. At the corners of the metal insert, the thermal contraction of the polymer component, which is bigger than the one of the metal insert, leads to the closure of potential gaps during the cooling down while the predominant part of the polymer is still molten. Parallel to this, the crystalline shrinkage of the polymer, the base of which is a contour of the test specimen between the injection mold and the metal insert, leads to a gap formation between the corners. On closer consideration of variant R and those with embossed structures, a1–a3 are similar except for variant R, where a1 and a3 are significantly larger than a2 (Figure 8C). This difference can be explained by the embossed structures, which also influence the gap formation in areas without a structure. The explanation is supported by the comparison of the variants with embossed structures among themselves. Two trends can be identified here. First, the deeper the structure, the smaller the average values of the characteristic gap sizes. The only exception is b1 for variant I0–100/I0–300, which can be explained by the dominant indentations at the side faces, which may conceal effects of the material bulges. Second, through additional indentations at the side faces, the average values of the characteristic gaps increase at each representative section, with the largest increase located at section A2.

In contrast to the head side, gaps cannot be measured at each position for variant R and those with embossed structures between L_a1 and L_a21 as well as L_b1 and L_b5 at the bottom side (Figure 8D). However, at the head side, only very small or no gaps could be measured in the corner areas. The reason for the gap formation at the corners is already explained for the head side. On closer consideration of variant R and those with embossed structures, it is conspicuous that the gap profiles and, thus, the average values of the characteristic gap sizes of the representative sections are significantly different to the ones of the head side, especially the differences between a1/a3 and a2. The reason why a1 and a3 are greater than a2 for all variants with the exception of variant MS is not yet clear. One theory is that the metal insert is very warm over an extended period of time, because the contact area of the metal insert with the injection mold is, due to its design, very small. Furthermore, the mold wall at the end of the flow path is 4 mm small and not actively cooled. These conditions in combination with the high thermal mass of the polymer and the high temperature of the injection mold lead to a slow cooling down of the polymer, especially in the vicinity of the corners of the metal insert. The result is a high crystallization degree at the corner areas, which results in a higher crystalline shrinkage at these positions. In addition, differences of the spherulite size between the corner areas and the areas between the corners may also have an influence on the crystalline shrinkage. Because of the long distance to the gate, this crystalline shrinkage presumably cannot be compensated by the holding pressure period. Furthermore, the local fiber orientations may also have an influence. The potential higher crystalline shrinkage in combination with the thermal contraction also leads to a higher overall shrinkage, which may be responsible for the small values of b1 with the exception of variants I0–100 and I0–300, where the additional indentations also have a big influence. The influences of the crystallization degree, differences of the spherulite size and fiber orientation on the formation of gaps and the gap profile will be investigated in detail in further studies.

When comparing the average values of the characteristic gap sizes of the representative sections of the bottom side with the head side of variant R, a significant decrease in the average values can be seen for all sections (Figure 8C and E). This is in line with the expectations, because the polymer dimensions around the metal insert are significantly smaller than those at the head side with its flange (Figure 2), and thus, the effects of the crystalline shrinkage on the gap formation are lower. However, when comparing a1, a2, a3 and b1 of the bottom side with those of the head side of the variants with embossed structures, no clear trend can be seen (Figure 8C and E). In addition, the comparison of the average values of the characteristic gap sizes of the representative sections at the bottom side (Figure 8E) only shows that the additional indentations have a big influence on the gap formation at section B1. These results show that there must be a complex relationship between the structures and shrinkage, which cannot be seen at the head side, where the flange prevents dislocations of the polymer on the crosspiece of the metal insert caused by shrinkage. However, the average values of the characteristic gap sizes of the representative sections of all variants are in good accordance with the observed leakage positions. Only in variants I0–100 and I0–300 were no air bubbles observed at the side faces, although large gaps were measured. The reason for this is analyzed in the xy and xz planes in the following. Furthermore, the average values of the characteristic gap sizes of the representative sections of variants R and MS are also in good accordance with the observed (R) or non-existent (MS; only at one test specimen a single air bubble was seen) leakage positions.

The characteristic gap sizes of the xz plane are shown in Figure 9 for the head (C) and bottom sides (E) of all variants with embossed structures. It can be seen that at the head side, all variants have gaps at each position of the embossed structure, so that no leakage barrier is generated. The main reason for this is the position of the structure, which is located in the area of the flange of the polymer component. At this position, the flange prevents a dislocation of the polymer component in the direction of the crosspiece of the metal insert, which is required for the leakage barrier. Thus, the polymer shrinks only perpendicular to the metal surface in the direction of the polymer base as described before, which leads to a gap formation at the structure ground, represented by L_d. However, because the structure flanks are not perpendicular to the structure ground, gaps also formed at these positions, represented by L_c and L_e. At the bottom side, where a dislocation of the polymer is not prevented, the L_c of all the variants is very small or cannot be measured. Furthermore, the shrinkage of the polymer in the direction of the flange leads to an opening of L_e, when comparing the proportion of L_d to L_e at the head and bottom sides. When comparing L_d of the bottom and head sides of all variants, a significant decrease in the values can be seen. This can be explained by the crystalline shrinkage perpendicular to the metal surface in the direction of the polymer base, because the polymer dimensions around the metal insert at this position are significantly smaller than those at the head side with its flange (Figure 2). The measured characteristic gap sizes are in good accordance with the measured leakage rates of these variants. Two main conclusions can be drawn from this. First, structures across the crosspiece of the metal insert, which should be used as a leakage barrier, must be deeper than L_d in the xz plane plus the maximum of L_aj in the yz plane. Second, these structures have to be placed at areas in which the shrinkage is not hindered by the injection mold and which have a maximum distance to the center of the shrinkage.

Figure 9:

Characteristic gap sizes in the xy and xz planes for the head and bottom sides. n.m. denotes that no gap could be measured. (A) Measuring positions for the xz plane as orientation. (B) Measuring positions for the xy plane as orientation_. (C) Characteristic gap sizes in the xz plane for the head side. (D) Characteristic gap sizes in the xy plane for the head side. (E) Characteristic gap sizes in the xz plane for the bottom side. (F) Characteristic gap sizes in the xy plane for the bottom side.

The characteristic gap sizes of the xy plane are shown in Figure 9 for the head (D) and bottom sides (F). At the head and bottom sides of variant 0–300, L_g is not measurable or measures only 1 μm. This is in rough accordance with the small values of b1 in the yz plane (Figure 8C and D), which describes the gaps at similar positions at the side faces of the metal insert, but outside of the structured areas. For the other characteristic gaps, L_h can only be measured at the head side and L_f only at the bottom side, which is in line with the expectations of the shrinkage behavior. However, all characteristic gaps at both sides are very small or not measurable, which is in contrast to the characteristic gaps in the xz plane with similar geometrical conditions. The only main difference between these two planes is the considered dimension of the metal insert, 1 mm in the z direction (xz plane) and 5 mm in the y direction (xy plane). Thus, it can be concluded that the dimensions of the metal insert in combination with the thermal contraction probably are responsible for the differences of the sizes of the characteristic gaps. However, even if the characteristic gaps are very small in the xy plane, they give a clue about the gap formation at the flanks of the material bulges.

For variant I0–100, the L_n of the head and bottom sides shows the same trends as the L_g at variant 0–300, but the difference is more distinct. The reason for this may be the indentations of the metal insert at the bottom side, which act as wedges during the shrinkage of the polymer in the x direction and, thus, open small gaps. In addition, the high internal stress in the polymer, which would be a result of the hindered shrinkage, could explain the value of b1 at the bottom side (Figure 8E), where a complex relationship between the structures and the shrinkage is already inferred. The characteristic gap size L_k at the bottom side, which is not measurable in contrast to L_p, supports this argumentation. However, in contrast to L_d and L_e of the embossed structures in the xz plane, which describe similar geometrical conditions, L_n and L_p are not measurable at the head side. These differences can be explained analogous to variant 0–300 by the dimensions of the metal insert in combination with the thermal contraction.

For variant I0–300, L_g, L_m and L_o of the head and bottom sides show the same trends like L_g and L_n at variants 0–300 and I0–100. Analogous to variant I0–100, the reason for this may be the indentations of the metal insert at the bottom side. As mentioned for variant I0–100, the high internal stress in the polymer as a result of the hindered shrinkage could explain the value of b1 at the bottom side (Figure 8E). Furthermore, in contrast to L_f and L_o, L_k and L_h are not measurable at the bottom side, which is in accordance with the gap formation of variant I0–100 and, thus, supports this argumentation. However, as for variant I0–100, no or only small values can be measured for the characteristic gaps at the head side. The probable explanation is the same, which is already given for variant 0–300.

When evaluating the characteristic gaps in the xy plane, it must be noted that only one test specimen per variant was analyzed, and thus, the results only show trends. Nevertheless, the results are in good accordance with the observed leakage positions. Especially, when looking at Figure 8E, the results of variants I0–100 and I0–300 in Figure 9F show why no air bubbles could be observed at the side faces (Figure 7B). Furthermore, based on the measured leakage rates and the characteristic gaps in the xy and xz planes, it can be concluded that the design of the structures has an effect on the leakage rate. If the structures at the front faces and side faces are designed differently, for example, through material removal at the front faces and application of material on the side faces, the leakage rate is increased. The reason for this is that the leakage barrier is not closed, because the edges on which the gaps are closed due to the shrinkage of the polymer are different, for example, variant 0–300 (Figure 9E and F). If the structures at the front faces and side faces are designed in the same way, for example, through material removal, the leakage rate is constantly low. In order to analyze the interactions between the front faces and the side faces in detail, further investigations with more test specimens are necessary. Furthermore, metal inserts with other ratios of the side faces to the front face are needed. In addition, for the analysis of the effects shown for variant 0–300, larger structures should be used.

In summary, the microscopy results show where the gaps are formed at the different variants at the head and bottom sides in the xy, xz and yz planes. Through the combination of these results, the formation of leakage networks can be explained for the investigated variants. At variant R, gaps at the side faces and front faces run from the head side to the bottom side, forming a leakage network without any leakage barriers (Figure 10A). In contrast, no gaps could be measured at variant MS, where the polymer-filled microstructures prevent any gap formation in the different planes. For the variants with embossed structures, the leakage networks are more complicated. In the range of the polymer flange, structures at the front faces cannot form leakage barriers and only extend the leakage path. However, at the bottom side, structures at the front faces straighten the leakage paths but cannot completely avoid the formation of a leakage network. At the side faces, structures can prevent the formation of leakage paths. However, only at one structure can flank gaps be closed (Figure 9D and F). Hence, if the structures are designed unfavorably, they can interact with those of the front faces, because there are also very small gaps at the edges of the metal insert (Figure 10B). If the structures at the side faces are designed favorably, which means that they are designed in the same way as the structures at the front faces, they can form, in combination with those of the front faces, a closed leakage barrier around the metal insert (Figure 10C). The fact why no air bubbles could be observed at the side faces during the air-under-water test of all variants with embossed structures can be explained by the interaction of the gaps behind the last structure at the bottom side (in the test direction). At variants 0–300, I0–100 and I0–300, there is only a small leakage path at the front faces (Figure 9E and F). In addition, the gaps after the last structure are larger than those in the structure, and the air pressure is significantly reduced, so that the air flows through the gaps on the front faces out of the test specimen. The very small gaps at the corners of the metal insert in combination with the reduced air pressure and the small gaps at the side faces of the metal insert seem to be too small to form leakage paths under the given conditions but are also potential leakage paths. For variant 0–100, the gaps at the side faces and the corners of the metal insert are also very small, so that potential leakage paths may be closed completely at a minimum of one position or they lead to the front faces of the metal insert.

Figure 10:

Formation of the leakage network of variants R (A), 0–300 (B) and I0–300 (C) on the surface of the metal insert at the bottom side of the test specimen. The material bulges and structure flanks are illustrated disproportional.