Development of a Low-Expansion and Low-Shrinkage Thermoset Injection Moulding Compound Tailored to Laminated Electrical Sheets

Abstract

This study presents a thermoset moulding compound designed for electrical machines with high power densities. The compound reduces residual stresses induced by the difference in thermal expansion during use and by shrinkage in the compound during the manufacturing process. To reduce the internal stresses in the compound, in the electrical sheet lamination and at their interface, first the moulding’s coefficient of thermal expansion (CTE) must match that of the lamination because the CTE of the electrical sheets cannot be altered. Second, the shrinkage of the compound needs to be minimized because the moulding compound is injected around a prefabricated electrical sheet lamination. This provides greater freedom in the design of an electric motor or generator, especially if the thermoset needs to be directly bonded to the electrical sheet. The basic suitability of the material for the injection moulding process was iteratively optimised and confirmed by spiral flow tests. Due to the reduction of the residual stresses, the compound enables efficient cooling solutions for electrical machines with high power densities. This innovative compound can have a significant impact on electric propulsion systems across industries that use laminated electrical sheets.

1. Introduction

With the increasing electrification of the automotive industry, the demands on electric traction drives are rising in terms of power density and efficiency. In the field of electric machines, lower weight and higher efficiency lead to increased vehicle range and better driving dynamics.

Improving torque density generally requires a higher electrical load on the stator windings. This article focusses on the latter. The higher electrical load generates large amounts of heat. Inadequate cooling leads to increased losses and even irreversible damage to the motor [1].

Copper losses account for the majority of these losses. These losses are proportional to the following:

$$P_{L,Cu,DC}~{\displaystyle \frac{{\rho }_{Cu,amb}(1+{\alpha }_{Cu}({\vartheta }_{Cu,avg}-{\vartheta }_{amb}\left)\right(L_{Fe}+L_{EW})}{A_{Slot}·k_f}}$$

(1)

where $P_{L,Cu,DC}$ is the copper dissipation, ${\rho }_{Cu,amb}$ is the resistivity at ambient temperature ${\vartheta }_{amb}$, ${\alpha }_{Cu}$ is the temperature coefficient of the resistivity, $A_{Slot}$ is the copper cross-sectional area, $L_{Fe}$ and $L_{EW}$ are the stator plate and end winding length, $k_f$ is the copper fill factor and ${\vartheta }_{Cu,avg}$ is the average copper temperature [2].

With regard to Equation (1), increasing the copper cross-sectional area and reducing the average copper temperature are the only feasible options. Maximising the fill factor has already been achieved with the introduction of flat wires in conjunction with hairpin technology.

Improving cooling seems to offer the greatest potential for increasing the power density of electric drive machines. The potential of direct cooling compared to conventional jacket cooling is shown in Figure 1. In addition to increased power, improved cooling can also have a positive effect on service life. According to the MONTSINGER rule, a temperature reduction of 10 K doubles the service life [3,4].

Thermal management of electric motors is not only important for electromagnetic reasons, as excessive temperatures within the components are also damaging. The insulation materials of the windings, particularly the winding heads, are the most sensitive areas within the stator assembly. In the automotive sector, insulation of thermal class 155 or 180 is common. An additional constraint is the permissible junction temperature of the power electronics [3,4,6].

In addition to the stator winding, considerable heat is generated in the rotor, its lamination and, if used, in the permanent magnets. The rotor area is thermally connected to the stator area and the housing by means of convective heat transfer in the end gap and the air gap, and by means of conductive heat transfer at the bearings. Excessively high temperatures in the permanent magnets reduce the remanence and coercivity, resulting in lower machine torque and possible demagnetisation. The temperature in the magnets should therefore be kept below the maximum permissible temperature, which is 180 °C for class UH NeFeB magnets, for example [7].

The cooling of the electrical winding is a critical issue in the thermal management of electrical motors. A number of thermal analyses [7,8,9] have shown that the windings in the stator are a heat source. The challenge in cooling them lies in the fact that they are located in the stator slots within the laminated electrical sheets and are therefore difficult to access. Their location in combination with the insulation of the wires makes it difficult to transfer the heat to an external heat sink. As a result of these difficulties, conventional cooling methods such as air or jacket cooling will not yield sufficient cooling performance for future high power and torque density motors [1].

A promising approach to improving the cooling performance of electrical machines is reducing the thermal resistance between the heat source and the heat sink. Similar to battery cooling strategies, liquid-cooled designs are superior to air-cooled designs in terms of cooling performance and are therefore better suited to high power densities [10]. New concepts in the field of liquid-based stator slot cooling are shown in Figure 2.

The small diameter of the cooling duct inside the hollow cables of concept (a) results in a significantly increased pressure drop along the duct. The higher flow rate required for the coolant results in a loss of efficiency [1].

The previously presented in-slot cooling channel concept (b) requires additional fluid routing for the coolant, which increases the thermal resistance between the heat source and the heat sink [1]. In addition, internal cooling channels can lead to an uneven cooling and thermal interaction between adjacent slots. This creates hot spots and affects the overall thermal performance of the stator.

The concept of direct winding cooling, as originally suggested by Oechslen [5,13,14] and subsequently developed, combines the advantages of the two preceding concepts. The combination of these advantages will result in a reduction of both thermal resistance and hydraulic resistance. That being said, the successful implementation of this concept is dependent on the availability of a stator seal that separates the rotor and stator chambers in a fluid-tight manner. In order to avoid induced eddy currents, this can only be achieved with a suitable plastic.

This stator seal must be very thin in comparison to its length and diameter. It is therefore in the shape of a thin liner. The production of such a liner for a larger series is, nevertheless, a challenge yet to be overcome. The most promising solution in this regard is a single-stage transfer moulding process for producing the liner. Since this is an original forming process, it also offers some degrees of freedom in designing the liner, i.e., simultaneously shaping the contour for twisting the hairpin winding. Since the electrical sheet laminations are only accessible for moulding in their axial direction, the moulding space is very limited. The absence of draft angles for demoulding within the moulding process leads to high demoulding forces, which in turn place significant stress on the materials of both the liner and the stator and their adhesion [1].

In order to guarantee the reliability of the system during operation, it is necessary to consider three potential failure mechanisms:

• Buckling of the liner due to fluid pressure;
• Detachment of the liner from the stator;
• Cracking in the liner.

Failure mechanism 1 can be countered with a suitable modulus of elasticity of the plastic. However, due to the shape factor and the relatively low pressure of the cooling fluid during operation, buckling failure plays a subordinate role in the design.

The differing thermal expansions (especially in the radial direction) between the stator constructed from laminated steel and the plastic liner result in residual stresses within the latter and shear stresses in the boundary layer. This can lead to cracking and detachment of the liner, with irreparable damage to the machine. This is precisely where the newly developed plastic for this liner comes in: its coefficient of thermal expansion has been adapted to that of the laminated steel. Consequently, the stresses in the liner and the adhesive layer are significantly reduced.

Epoxy resin-based materials are a recently emerging material group used for sealing the stator lamination. In general, epoxy resins have the advantage of a thermal conductivity superior to that of other technical plastics. Phenolic resins represent another potential solution, although their comparatively low strength renders them only competitive when combined with fibre reinforcements. Nevertheless, the aforementioned advantages in terms of design freedom are contingent upon the utilisation of isotropic moulding compounds. Consequently, the incorporation of fibre reinforcements is contraindicated in this context. A selection of commercially available epoxy-based thermoset moulding compounds, which can be used for sealing a stator lamination, is presented in

Table 1. Commercially available thermoset moulding compounds according to [15,16,17]. Key properties CTE and shrinkage in bold type.

Type	Unit/Symbol	Duresco NU 6110	Huntsman Aratherm® CW 2731	Sbhpp Vyncolit® X7530
Glass transition temperature	TG/°C	160	165	224
Thermal expansion	α/10−6/K	18	24	10‖/41┴
Shrinkage	ΔV/%	0.2–0.7	-	0.15
Tensile strength	σ/MPa	80	80	123
Elongation at break	ε/%	1.2	0.4	0.95
Tensile modulus	E/MPa	18,000	23,000	17,000

.

For any seal of a stator lamination, the seal must stay fluid tight during thermal cycling. To ensure that the selected material stays fluid tight during thermal cycling, initial tests have shown that the material must be selected on the basis of its CTE and its shrinkage [2]. HüBSCH quantifies the CTE of a stack of laminated electrical sheets as 12.6∙10⁻⁶/K in the radial direction [18].

Prior tests conducted by PORSCHE AG also showed that none of the currently available moulding compounds meet the thermal cycling requirements that are necessary to implement the liner in an electric drive unit. FE calculations have demonstrated that even minor deviations in the CTE have a significant impact on the residual stress state within the stator liner sealing.

As none of the moulding compounds currently available on the market can meet the requirements necessary to implement an electric drive unit with an internal sealing liner, the objective of this project was to develop a moulding compound tailored to the requirements of a sealing liner for a stator lamination. The compound was functionalized to meet the chemical, physical, mechanical and technological requirements.

In the following paragraphs, the authors describe how such a moulding compound was specially developed for this specific application. The development took place as part of a PORSCHE AG research project in collaboration with the University of Stuttgart (Department of Lightweight Design) and the University of Bayreuth (Department of Polymer Engineering). It is characterised by minimal shrinkage and a coefficient of thermal expansion of 12.5$·$10⁻⁶/K, which is matched to the radial CTE of the stator lamination. This CTE is dominant for a liner sealing under outside pressure from the cooling fluid when attached to the stator. In this manner, the residual stresses due to thermal stress in both materials were minimized. Apart from avoiding cracks, this approach offers a higher load carrying capacity of the liner sealing against the cooling fluid pressure.

2. Material Development

In order to develop a new type of material for a functionalized polymer component in an electrical motor, first the material requirements needed to be defined. Second, a strategy for tailoring the material to the requirements needed to be developed and implemented. Both steps are presented now.

2.1. Material Requirements

To achieve the required properties of the new specific resin system, the first step is to pin down the required specification profile.

Table 2. Specifications for the definition of the requirement profile. Key properties CTE and shrinkage in bold type.

Type	Unit/Symbol	Target Value
Glass transition temperature/TG	TG/°C	≈150
Thermal expansion (CTE)	α/10−6/K	<16
Shrinkage	ΔV/%	<0.25
Tensile strength @ 23 °C	σ/MPa	>80
Elongation at break @ 23 °C	ε/%	>0.75
Tensile modulus @ 23 °C	E/MPa	>7000
Tensile modulus @ 135 °C	E/MPa	>4000

below shows the defined specifications.

Due to its use in an electrical machine, temperatures of up to 135 °C are reached for short periods [19]. Preliminary tests with a reference material identified the internal thermal stresses between the sheet metal lamination and liner seal as a key to the success of the seal. Coefficients of thermal expansion above 16∙10⁻⁶/K for the plastic lead to stress cracks and hence failure of the component. Very similar shrinkage also leads to residual stresses in plastic-metal joints, so shrinkage needs to be minimised. The required tensile strength and Young’s modulus are determined by the compressive loads imposed by the pressure of the coolant during the operation of the motor.

2.2. Material Tailoring Strategy

In order to achieve the target values of the above specifications, a new type of customized material system is being developed. Two primary issues being considered are as follows:

• Low CTE;
• Low shrinkage.

A suitable resin system must meet the requirements for processability (in particular for the incorporation of high filler contents, as well as later on for further processing in relation to component manufacture), and at the same time have a positive effect on, for example, reduced shrinkage while maintaining the required thermal properties due to its basic properties. First, a critical selection of suitable resin components is made. The main cause of shrinkage in cross-linked systems is chemical shrinkage, in addition to the influence of thermal expansion caused by temperature rise and subsequent cooling during curing. This is due to the curing reactions and is therefore unavoidable [20,21]. However, since the opening of the oxirane ring during curing leads to a reduction in the atomic spacing, the shrinkage behaviour can already be reduced by the basic structure of different types of epoxy resin.

In the field of epoxy resins, there are two basic types that are mainly used. These are the bisphenol A diglycidyl ether (DGEBA) types based on bisphenol A and the epoxy Novolak resins obtained from phenols with formaldehyde. Due to the differences in the repeating unit of the chain, there are mainly differences among the thermal properties and chemical shrinkage. As illustrated in Figure 3, combinations can be used that achieve the required glass transition temperature but reduce the shrinkage of the resin system by partial substitution.

To further functionalize an epoxy material, filler loading can be used to further reduce CTE and shrinkage, whilst significantly increasing mechanical properties as well. Here, the selection of suitable fillers is essential. Numerous fillers have been used, according to the literature, to influence shrinkage and CTE. The use of inorganic fibres or particulate fillers such as quartz, fused silica, cristobalite, aluminium oxide, carbon nanotubes, feldspar and other ceramic materials is common [22,23]. Fused silica is particularly suitable for reducing the CTE with a value of 0.5∙10⁻⁶/K, which shows little change even at elevated temperatures [23]. Furthermore, a toughening modification of the resin system can additionally be conducted. A block copolymer based on poly-ε-caprolactone and polydimethylsiloxane is used to counteract embrittlement of the material due to high filler loading. Various resin types have been successfully toughened with this, according to the literature [24,25]. The explicitly used material components can be seen in Section 3.1.

3. Experimental Screening

3.1. Materials and Methods

Three epoxy resins, D.E.R. 331, D.E.R. 662.e and D.E.N. 438, were procured from OLIN Corporation (Stade, Germany). The hardener, a dicyandiamide, DYHARD 100S (DICY) and the urea-based accelerator DYHARD UR400 were obtained from ALZCHEM GROUP AG (Trostberg, Germany). The toughening agent GENIOPERL W35 was procured from WACKER CHEMIE AG (Munich, Germany) and filler SILBOND FW 600 EST was obtained from QUARZWERKE (Frechen, Germany). All materials were used as received without further purification.

The resin formulations were each processed by using a standard dissolver (IKA-Werke GmbH & Co. KG, Staufen, Germany; 500 rpm). The neat resin components were first incorporated, followed by the sequential addition of the toughening modifier, hardener and accelerator. Further processing with regard to the addition of the filler was also initially carried out using a dissolver. For high filler quantities above 60 wt.-%, a three-roll mill was used. All formulations were subjected to an identical curing programme in a convection oven, comprising 2 h at 110 °C, 2 h at 135 °C, and 2 h at 180 °C. This was followed by a cooling period of 5 h to prevent the formation of residual stresses.

3.1.1. Differential Scanning Calorimetry (DSC)

DSC measurements were conducted on a DSC1 (Mettler Toledo, Columbus, OH, USA) with a heating rate of 20 K min⁻¹ under nitrogen (50 mL min⁻¹) from −50 °C to 280 °C in accordance with [26]. The sample weight was approximately 15 mg.

3.1.2. Thermomechanical Analysis (TMA)

TMA measurements were conducted on a Q400 EM (Texas Instruments, Dallas, TX, USA) with a heating rate of 3 K/min from 25 °C to 220 °C. The sample geometry was 10 × 10 × 2 mm³. To prevent potential post-curing effects, all samples were subjected to two heating cycles.

3.1.3. Thermogravimetric Analysis (TGA)

TGA measurements were conducted on a NETZSCH 209 F1 LIBRA (Selb, Germany) at a heating rate of 10 K/min from 25 °C to 800 °C under synthetic air (50 mL/min). The sample weight was approximately 15 mg.

3.1.4. Shrinkage

As illustrated in Figure 4, the shrinkage was quantified by comparing the dimensions of a tool mould of approximately 150 × 150 mm² and a cured sample plate in this mould. The dimensions of the removed sample plate in the X and Y directions of the mould were used to calculate the shrinkage in relation to the dimensions of the mould.

3.1.5. Tensile Testing

Tensile testing was conducted in accordance with the specifications outlined in DIN EN ISO 527-2, utilising the TYPE 1BA test specimen. The selected preload was 2 N, with a test speed of 1 mm/min for the determination of the modulus of elasticity and 2 mm/min for the determination of strength. In consideration of the test temperature of −40 °C, room temperature, 135 °C and 150 °C, the influence of environmental factors was evaluated.

3.2. Results and Discussion

Initially, combinations of the epoxy resin components were tested in order to develop a base system with the lowest possible shrinkage at the start. However, it is important to take the target glass transition temperature T_G (approximately 150 °C) as a boundary condition into consideration. The glass transition temperatures of the individual formulations, as determined by DSC, are presented in

Table 3. T_G determination by means of DSC for benchmark and new formulations.

	TG/°C
Benchmark	179
Formulation 1	134
Formulation 2	130
Formulation 3	125
Formulation 4	148

.

Initially, only DGEBA-based types were used in formulations 1 to 3, which were varied according to their repeating units in the chain. Here, the proportion of the long-chain component was gradually increased in turn. With a decrease from 134 °C to 125 °C, there is only a slight decrease in relation to the T_G, but the system was not yet sufficient in relation to the requirement profile. The fourth formulation shows an elevated glass transition temperature, which was achieved through the substitution of a portion of the DGEBA component with the epoxy Novolak. This substitution results in fulfilling the required properties of the system, and therefore in a mixture of the two mentioned epoxy types as formulation 4.

Formulation 4 was the starting point for the modification of the resin, which was carried out using fused silica material as a filler. The objective was to minimise the primary required parameters (CTE and shrinkage).

As illustrated in

Table 4. Influence of filler loading on thermal expansion.

Formulation	CTE 10−6/K	Processing Method
Reference	22.5	-
Formulation 4_neat	78	Dissolver
Formulation 4_50 wt.-%	42.5	Dissolver
Formulation 4_60 wt.-% *	40	Dissolver
Formulation 4_70 wt.-%	22	Three-roll mill
Formulation 4_75 wt.-%	12.5	Three-roll mill

* Outlier due to inhomogeneous filler distribution causing change of processing method.

, the unfilled neat resin formulation has a CTE of 78∙10⁻⁶/K, which is approximately five times higher than that of potentially relevant metal-based substrates. However, this can be significantly reduced by higher filler loading. Even at a concentration of 50 wt.-% of filler, the CTE is reduced from 70∙10⁻⁶/K to 42∙10⁻⁶/K, representing a 45.5% decrease. It is noteworthy that a further increase to 60 wt.-% only results in a further reduction to 40∙10⁻⁶/K. An increase in the filler content to 70 wt.-% results in a further significant reduction in the CTE to 22∙10⁻⁶/K. This reduction may be attributed to the process method employed. The processing method used for the transition from 60 wt.-% to 70 wt.-% involved a change from a dissolver to a three-roll mill. This is due to a significant increase in viscosity resulting from the use of high filler quantities. An uneven distribution of the fillers at a filler loading of 60 wt.-% using the dissolver may explain why the CTE was minimally reduced further (α_50% = 42.5∙10⁻⁶/K, α_60% = 40∙10⁻⁶/K). If the CTE is considered at the maximum filler loading of 75 wt.-%, it is reduced by almost half again from 22∙10⁻⁶/K to 12.5∙10⁻⁶/K. Based on the unfilled pure resin system, the CTE can thus be reduced by a total of 84% and has a comparable value in relation to the metal substrate under consideration.

Due to the disparate processing methodologies and the anomalous data pertaining to the CTE, the formulations were subjected to thermogravimetric analysis (TGA).

Table 5. Comparison of the theoretical filler quantity with determination of the actual quantity using TGA.

Formulation	Theoretical Additive Content/wt.-%	Measured Additive Content/wt.-%
Formulation 4_50 wt.-%	50	49.8
Formulation 4_60 wt.-%	60	59.1
Formulation 4_70 wt.-%	70	69.8
Formulation 4_75 wt.-%	75	76.0

presents a comparison of the theoretical filler content based on the weighed quantities and the actual filler content.

A comparison of the final filler contents measured by TGA indicates that there are minimal fluctuations in relation to the theoretical quantities used, regardless of the processing method. This can also be observed in the formulation with 60 wt.-%, in which the outlier of the CTE measurements is present. This verifies the hypothesis that the fillers are unevenly distributed due to the limitations of the processing method, as there is sufficient filler in the overall composite system, indicating that an area within the outlier regarding Table 4 with excess resin was present during the production of the CTE samples. Due to the low coefficient of thermal expansion (CTE), the final formulation 4 with 75 wt.-% filler content was considered for further measurements with regard to shrinkage and mechanical behaviour.

Table 6. Shrinkage due to curing.

Formulation 4_75 wt.-%	Dimension of Tool/mm	Dimension of Sample/mm	Shrinkage /%
/%
X1	150.49	150.35	0.09
X2	150.45	150.30	0.10
Y1	149.94	149.76	0.12
Y2	149.95	149.74	0.14
Combined shrinkage			0.11 ± 0.02

shows that there is a shrinkage of 0.11 ± 0.02% in relation to the points selected at different intervals along the sample. This indicates that the optimal formulation in relation to the CTE also exhibits a shrinkage that is sufficient for the required profile.

Once the modification of the system was demonstrated to fulfil the required criteria with regard to the core properties of the lowest possible thermal expansion and shrinkage, it was also examined for its tensile properties. In addition to exhibiting high mechanical properties at room temperature, the system must also demonstrate resilience at elevated temperatures (e.g., 135 °C and 150 °C) and in the low temperature range (down to −40 °C). This is crucial for preventing the system from becoming subject to the problem of microcrack formations.

As illustrated in Figure 5, the tensile strength of the system at room temperature is 114 ± 8.03 MPa, which is well above the required specification profile of 80 MPa. Furthermore, under the influence of extreme environmental conditions, such as an ambient temperature of −40 °C, the tensile strength even increased to 134 MPa. Despite embrittlement due to the low temperature test, significantly higher forces are required to reach the final material failure. As anticipated, the tensile strength decreased to 85.8 MPa at elevated temperatures, such as at 135 °C in comparison to room temperature. However, it should be noted that the tensile strength remains in the range of typical high-performance epoxy resins at room temperature.

A comparable trend can be observed when examining the tensile modulus. In particular, at room temperature, the system exhibits a high modulus of 15,980 ± 1603 MPa, resulting from the high filler loading. Although the modulus decreases significantly as the temperature increases (e.g., 7254 ± 1850 MPa at 135 °C and 4042 ± 439.4 MPa at 150 °C), it still corresponds to the value defined in the requirement profile at 135 °C, even in the range of the glass transition temperature. Furthermore, the elongation at break of the system also fulfils the requirements across all ambient temperatures. Of special interest is the comparison between room temperature and −40 °C. Despite the low temperature test, the elongation at break is still equivalent to room temperature at 1.0 ± 0.1%. Here, the T_G of the toughening modifier used plays a significant role in this regard. These temperatures are −120 °C (PDMS block) and −60 °C (PCL block), and therefore the formulation does not become brittle when tested at −40 °C, continuing to contribute its mode of action to the tensile modification [19]. A complete overview of the tensile properties can be found in

Table 7. Overview of tensile properties.

Temperature /°C	Tensile Strength /MPa	Tensile Modulus /MPa	Elongation @ Break/%
−40	134 ± 9	16,980 ± 1127	1.0 ± 0.1
23	114 ± 8	15,980 ± 1603	1.0 ± 0.2
135	85.8 ± 13	7254 ± 1850	3.4 ± 0.2
150	46.3 ± 3	4042 ± 439.4	3.2 ± 0.3

.

Table 8. Overview of the core properties of the newly developed resin system in relation to the specifications and the commercial reference. Key properties CTE and shrinkage in bold type.

Type	Unit	Target Value	Reference	Novel Formulation
Glass transition temperature	TG/°C	150	175	148
Thermal expansion	α/10−6 K−1	<16	22.5	12.5
Shrinkage	ΔV/%	<0.25	0.25	0.11
Tensile strength @ 23 °C	σ/MPa	>50	80	114 ± 8.
Elongation at break @ 23 °C	ε/%	>0.75	0.5	1.0 ± 0.2
Tensile modulus @ 23 °C	E/MPa	>7000	18,000	15,980 ± 1603
Tensile modulus @ 135 °C	E/MPa	>4000	-	7254 ± 1850

presents a final comparison of the primary properties of the benchmark system and the newly developed final optimised material formulation in relation to the specifications initially defined.

4. Validation

4.1. Mechanical Behaviour

The residual stresses in the target state were calculated to define the material requirements. The entire process chain was simulated, including the thermal and mechanical process steps. Sensitivities were analysed on the basis of the simulation results.

In comparison to the reference material with a CTE of 18 ppm and a shrinkage of 0.5%, the residual stresses at room temperature were reduced from 62 MPa to 16 MPa. This represents a significant improvement, with a factor of 3–4.

The combination of the high strength of the newly developed material and the reduction in residual stress at room temperature resulted in an increase in the ratio of tensile strength to residual stress from 1.3 to 7.1. compared to the reference. This considerable enhancement in the material’s resilience renders it suitable for use in series production.

4.2. Flow Path Length

The utilisation of the recently developed epoxy resin for a functionalized, original formed component in an electrical machine requires the validation of the flow behaviour. This validation is conducted through spiral flow tests. The adjustable test parameters include the material portion temperature (temperature of the pre-plasticised material quantity), the injection pressure, the mould temperature and the holding pressure phase. The processing is executed using a transfer moulding process. This process is suitable for the processing of highly filled epoxy resins (EPs) and is used in the production of a stator liner sealing for direct cooling of the windings in electrical motors. The minimum required flow path length of the newly developed material can be derived from the target mould design and is 600 mm with a safety margin of 1.2.

A test plan was devised to systematically analyse the influence of process parameters on flow path lengths. The mould temperature was maintained at 135 °C. Initial tests demonstrated that lowering or raising the mould temperature does not result in any improvement in flow path lengths. Furthermore, interactions in relation to the mould temperatures were also not observed. The mould temperature was set to 80 °C, 100 °C and 120 °C. The transfer pressure was set to 70 bar, 100 bar and 150 bar. The holding pressure was set to 180 s, 500 s and 1200 s. The test for normal distribution shows no abnormalities, nor do the residual plots. Outliers did not occur at a significance level of 5%.

The flow path lengths vary from 100 mm to 1386 mm. The longest flow path lengths are achieved at a material portion temperature of 100 °C, injection pressures of 100 bar and a maximum holding pressure duration of 1200 s. The contour diagram of the flow path length as a function of the material portion temperature and the holding pressure duration is shown in Figure 6. Of particular interest is the significant influence of the holding pressure duration on the flow path length. This indicates a slow cross-linking reaction, which means that sufficient material is still fed into the mould over a long injection period.

The high sensitivity of the material concerning the holding pressure duration is striking. Slight changes to the process parameters result in greatly altered flow path lengths. In particular, injection pressures of more than 150 bar lead to flow path lengths of less than 50 mm. This is due to the high filler content, which leads to clogging of the feed point of the mould. The hypothesis of the filler being responsible for the clogging was confirmed by the observation in the area of the feed point. Low-viscosity components were observed to escape below the feed point, while the fillers were observed to clump together at the feed point. However, the flow behaviour is deemed to be sufficient for the production of a liner sealing, given the minimal necessary flow path lengths that were initially defined.

5. Conclusions

This paper presents a newly developed, highly filled EP injection moulding compound that fulfils all requirements listed in paragraph 2.1. These requirements are derived from the use case to which the material is functionalised, which is a liner sealing for a stator lamination of an electrical engine. In particular, the two primary requirements of low thermal expansion (12.5∙10⁻⁶/K) and low shrinkage (0.11 ± 0.02%) should be mentioned. This leads to significantly lower thermal residual stresses and thus to an increased load-bearing capacity of the stator seal, which provides the application with higher safety margins or the possibility to increase the pressure of the cooling fluid. Concurrently, the high tensile strengths, which can be achieved through the formulation, also result in a considerable increase in load-bearing capacity. This increase in tensile strength at high temperature combined with the achievable glass transition temperature of 148 °C promise a safe short-term use of the seal at these elevated temperatures.

The investigation of the flow behaviour using flow spiral tests has demonstrated a significant influence of the holding pressure time on the flow path lengths. This influence has also been observed in relation to the mould temperature and the injection pressure.

The production of a stator liner sealing in a series application requires the use of materials that facilitate stable manufacturing processes. The developed material does that. It also does not limit the design of the injection mould because the material properties are isotropic.

Current steps for manufacturing liner seals on an industrial scale include the reduction of the duration of the pressure holding phase. This is necessary to reduce the cycle times, to match production in larger quantities, and consequently to reduce production costs. In this and other aspects, the newly developed material will be optimized for production. The objective of these investigations is to enhance the processability of the material and reduce cycle times. In order to achieve this goal, the use of various additives and a variation of process parameters will be researched in greater detail. As soon as this last part of the research is completed, the prototypes will be produced. These prototypes are scheduled to be assembled into fully functional electric drive units.

6. Patents

The results of this research were deemed worthy of protection. Consequently, the moulding compound was submitted to the German Patent Office on 24 April 2024 under the number DE10 2024 111 526.2.