1 Introduction

Current research shows that printed sensors for technical or biomedical applications are coming into focus. Such sensors can be used to measure physical, chemical, or biological parameters [1,2,3]. In the field of polymer-based sensors, extrusion-based Additive Manufacturing (AM) processes can be used to produce housings [4,5,6,7,8,9] and sensory elements [10,11,, 12]. In addition to the possibility of printing sensory elements, it is also possible to integrate active components within the printed component area and to functionalize the components accordingly [13,14,15,16]. For the contacting of these active components, it has also been considered how conductive elements can be integrated as conductive tracks in order to electrically connect such components [14, 17]. The work presented here builds on these results and shows how conductive tracks made of a silver conductive paste can be integrated into AM parts and how inductive proximity sensors can thus be functionally manufactured and qualified.

Analog inductive proximity sensors are well known for their robustness and their wide usability in many automated industrial processes [18]. They are available in different standard cylindrical and cubic housings [19]. The motivation of this work was to show that sensors for industrial use can be manufactured in individualized housings according to the geometrical requirements of the customer or the application, i.e. for retrofit solutions or situations where standard sensors do not fit into the given space. The approach makes it possible to use standard electrical components, freely arrange them in space and integrate them into an individualized housing. This makes it possible to adapt the shape of the housing to the geometric conditions of the application and to increase the integration density.

2 Methodology

2.1 Materials

2.1.1 Material of the test specimens and the sensor housing

A polybutylene terephthalate (PBT) blend from Mitsubishi Engineering-Plastics Corp. named NOVADURAN™ 5710N1TX was used to additive manufacture the test specimens. This material is unreinforced, flame retardant and shows relatively low warpage. Table 1 lists mechanical, thermal and electrical material parameters. The PBT blend was also used to print the housings of the individualized sensors. The parameters used for processing are described in Pfeffer et al. [20].

Table 1 Properties of used PBT material NOVADURAN™ 5710N1TX [21]
Full size table

2.1.2 Conductive material

To create the conductive tracks the silver-filled paste IPC-605 from Inkron was used. The material is designed for screen printing applications and is after processing stretchable and conductive. Table 2 lists the uncured and cured (15 min @ 110 °C) properties of IPC-605.

Table 2 Properties of used IPC-605 silver paste [22]
Full size table

2.1.3 SMD adhesive

To fix the circuit board and connector pin strip in the PBT housing the SMA10SL SMD adhesive from Electrolube was used. The one-component adhesive has a high uncured green strength and it cures between 90 °C and 150 °C. The adhesive was applied manually.

2.2 Printing equipment

2.2.1 Processing of PBT

The PBT test specimens and sensor housings were printed using the Arburg Plastic Freeforming (APF) process and a freeformer 200-3X from Arburg. It has two discharge units and uses nozzles with a diameter of 0.2 mm. In addition, the freeformer 200-3X has three axes with a positioning accuracy of ± 0.022 mm. APF is assigned to the material extrusion (MEX) processes. The printing parameters are based on those of Pfeffer et al. [20].

2.2.2 Processing of silver paste

After the PBT has been printed, the build plate including the printed part (test specimen or sensor housing) is transferred to a self-built test rig of Fraunhofer IPA. This rig realizes the axis motion in three dimensions (Cartesian motion; see Fig. 1).

Fig. 1
figure 1

Simplified setup for processing the silver conductive paste; the micro dispenser moves in X and Y direction and the build plate in the Z direction

Full size image

The silver paste was then jetted with a Nordson microdispensing valve consisting of a PICO xMOD PA MV-100 piezo actuator, a PICO XMOD VS-3.0S-F0-D10 valve seat and a PICO XMOD CR-3.0S PEEK cartridge with liquid gasket. This valve was controlled by a PICO DRIVER DCON controller. The trigger signals for the controller are generated by a separate microcontroller that calculates the current position and triggers the valve when a specified distance is exceeded.

Table 3 summarizes the process parameters used to dispense lines on planar surfaces, and Fig. 1 shows the setup for processing the silver paste.

Table 3 Process parameters of the dispense valve to generate lines on planar surfaces
Full size table

2.3 Test geometries

For this paper, a silver paste was printed on different geometries to investigate if conductive tracks could be created on these structures. In addition, two contacting variants were investigated to create a connection between the silver paste and the sensor components.

2.3.1 Sloping line

To analyze how steep a slope can be to still create conductive tracks, a silver paste was dispensed on PBT samples (18,5 mm × 8 mm × 10 mm) with different slopes (40°, 50°, 60°, 70° and 80°) as shown in Fig. 2.

Fig. 2
figure 2

CAD models of the PBT test specimens with different slopes (from 40° to 80°)

Full size image

In order to keep the printing frequency and the drop spacing constant, the parameters’ printing speed (speed of movement of the microdispenser) and the dot spacing had to be adjusted in the software. The adjusted parameters are shown in Table 4. For a better understanding of drop spacing and dot spacing, the difference is visualized in Fig. 3.

Table 4 The adjusted printing parameters for the sloped structures to maintain the same drop spacing and printing frequency
Full size table
Fig. 3
figure 3

Difference between drop spacing and dot spacing

Full size image

The conductivity of the printed lines was tested by hand with a multimeter (METRAHIT 29S by Gossen Metrawatt) with measuring tips.

2.3.2 Vias

In addition to slopes, vias were investigated as a feature to interconnect two Z-planes. Therefore, PBT test geometries (20 mm × 8 mm) were printed with twelve 1 mm × 1 mm square cavities (see Fig. 4). The size of the vias is relatively large compared to vias in conventionally manufactured PCBs (currently industrially relevant dimensions are approximately 0.1 mm–0.7 mm [23, 24]), but had to be chosen in this order of magnitude because the volume and the placement accuracy of the generated drop required this cavity size. To generate the vias, the silver paste was dispensed into these cavities. Two of them (top left and bottom right) were used to check the positioning in the test rig. Two depths were investigated: 3 mm and 5 mm. Conductivity was tested by hand with a multimeter with measuring tips.

Fig. 4
figure 4

CAD models of PBT test specimens with different cavity depths

Full size image

Two contacting strategies were investigated to connect the various sensor components (coil, circuit board, and connector pin strip) with the silver paste. The first option was to insert the component into the housing and then dispense the silver paste onto the area to be contacted. The second was to first print the silver paste into the housing and then place the component into the housing and onto the paste. The variations are shown in Fig. 5.

Fig. 5
figure 5

CAD models of two connecting types; left: the connection from above; right: the connection from below

Full size image

2.3.3 Curing process

All PBT specimens and PBT sensor housings including the dispensed silver paste structures and the applied SMD adhesive were placed in a universal oven UF55plus (Memmert, Germany) at 110 °C for 30 min. Curing of the silver paste took place during this process.

2.3.4 Overview of the test specimen process chain

In Fig. 6, the process flow for manufacturing the test specimens is summarized.

Fig. 6
figure 6

Flow chart of the test specimen manufacturing process

Full size image

2.4 Demonstrator design

The overall approach within this activity is to integrate electrical components for sensors and actuators into a polymer housing. Ultimately, an individualized sensor can be manufactured, where the individual components such as coil, circuit board or connector can be freely arranged and application specific. The proposed sensor design is shown in Fig. 7. Cavities are created within the PBT housing to integrate the electrical components and conductive tracks.

Fig. 7
figure 7

Schematic view of an individualized sensor design using standard sensor elements, which are arranged application specifically. The transparent part represents the PBT housing, which was printed using APF

Full size image

2.5 Hybrid manufacturing process

In order to manufacture such an individualized sensor, several manufacturing steps have to be performed, which ultimately form a hybrid process chain. The manufacturing steps of the hybrid process chain are listed below (shortened):

  1. 1.

    Additive manufacturing of PBT housings and PBT inserts.

  2. 2.

    Integration of sensor coil and PBT insert into PBT housing.

  3. 3.

    Imprint the second PBT housing part.

  4. 4.

    Production of silver conductive tracks between the sensor coil (“connecting from above”, cf. Figure 5) and the circuit board (“connecting from below”, cf. Fig. 5).

  5. 5.

    Insertion of the connector element and fixation with SMD adhesive.

  6. 6.

    Integration of circuit board, fixation with SMD adhesive and production of silver conductive tracks between circuit board, and connector (“connecting from above”, cf. Fig. 5).

  7. 7.

    Integration of PBT insert and enclosing of PBT housing.

  8. 8.

    Potting of the cavities with polyurethane potting compound to ensure watertight encapsulation of the coil, board, and connector.

The term circuit board is used to refer to the assembled PCB (Printed Circuit Board) that carries the active and passive electronic components. The PCB drives the sensor coil, performs the signal processing, and provides the output voltage to the control unit via the connector.

2.6 Qualification tests of the whole sensors

For this paper, twelve individualized sensors (produced as explained in Sect. 2.5) were sent to the test campus of Balluff GmbH to be subjected to the standard tests for series-produced sensors. Qualification was carried out there in accordance with the product standard DIN EN IEC 60,947–5-2, which is generally used for conventionally manufactured inductive proximity sensors. The standard recommends testing the sensor's characteristics concerning heating, insulation properties, making and breaking capacity of the switching elements, behavior under conditional short-circuit current, material requirements, protection classification, switching distances and frequencies, electromagnetic compatibility, shock resistance, and vibration resistance. Before and after the environmental tests, the output characteristic of the individualized sensors has been measured. A linear output voltage of 0 V to 10 V is specified in the linear range of 0.2 mm to 3.5 mm of a metal object (falling on approach). The detailed test conditions are presented in Sect. 3.4.

2.7 Micrograph preparation and microscopy

All microscope images were taken with a digital microscope DVM6 (Leica Microsystems, Germany). For the micrographs, some specimens or housings were embedded in epoxy resin and prepared with the grinding/polishing machine LaboPol-20 (Struers GmbH, Germany).

3 Results

3.1 Sloping lines

The conductive structures on the 60° slopes were conductive on all three test specimens. At 40° and 50°, two out of three lines of silver paste were continuously conductive (R < 0.5 Ω), and at 70°, one out of three lines was continuously conductive. At an angle of 80°, no continuous conductive lines could be produced. The results are shown in Table 5 and an example is shown in Fig. 8.

Table 5 Overview of the results of the sloping line test
Full size table
Fig. 8
figure 8

PBT test specimens of the sloping line test with the dispensed silver structures

Full size image

3.2 Vias

Of the 3 mm deep vias, 85% (34 out of 40 samples) are continuously conductive and of the 5 mm deep structures, 74% (37 out of 50 samples) are continuously conductive. The results are shown in Table 6 and an example is shown in Fig. 9.

Table 6 Overview of the results of the via test
Full size table
Fig. 9
figure 9

PBT test specimens of the via test with the dispensed silver paste

Full size image

3.3 Contacting of the sensor components

Figure 10 shows a functional, but not overprinted and encapsulated sensor with all conductive tracks and connection elements in polished cut images. A planar conductive track is shown in section A. The connection from below was used to connect the circuit board to the tracks coming from the coil (section B). The vias were realized in section C to connect the coil to the conductive tracks, which are located a few millimeters higher. The connection from above was used to connect the conductive tracks to the circuit board on the pin strip site (section D) as well as to connect the conductive tracks to the pins of the pin strip (section E).

Fig. 10
figure 10

The not overprinted sensor with all electronic components; cross-sections of the main features printed with silver paste: A conductive track; B connection from below; C via; D connection from above; E connection of the pin strip

Full size image

3.4 Sensor qualification

Qualification was performed in cooperation with the Balluff test campus using standard test methods and equipment. The qualification tests performed, the test conditions applied and the results obtained are presented in Table 7.

Table 7 Test conditions and results of the qualification tests performed
Full size table

4 Discussion

4.1 Sloping lines

With the Fraunhofer test rig, the Nordson dispensing system and the selected dispensing parameters, conductive tracks could be produced on slopes ranging from 40° to 70°. On the 40° to 60° slopes, seven of the nine tracks were conductive. In general, it can be assumed that the droplets are not always fired perfectly vertically from the nozzle. Slight accumulations of material at the nozzle, which were observed during the tests and are shown in Fig. 11, are very likely to increase this non-vertical ejection. When the droplets fly a few millimeters deep, as in these tests, the drop spacing can vary even more, which can explain the inconsistency at 40° and 50°.

Fig. 11
figure 11

Accumulations of material at the nozzle after dispensing

Full size image

For the 70° and 80° geometries, only one of the six tracks could be generated continuously. The variation in drop spacing must also be taken into account. In addition, it is likely that above a certain inclination, the drops slip slightly and do not stick at the desired height.

To reliably dispense conductive structures on slopes the process parameters need to be further optimized. Dot spacing could be further reduced to compensate for the variation in drop spacing. In addition, reducing the dot spacing means increasing the amount of material dispensed for the same track length. In the case of slopes steeper than 60°, the additional material can act as a support structure, which could make it possible to realize slopes of 70° or 80°.

In any case, material accumulation at the nozzle should be avoided. This will allow a straight drop flight and thus increase reproducibility. Further optimization of the printing parameters may be useful. In addition, the choice of a different dispensing system (e.g. nozzle or piston geometry) or a different silver paste could help to avoid this problem.

4.2 Vias

It was possible to create via depths of 3 mm and 5 mm. However, the results show that the reliability of continuous contacting decreases with increasing via depth. There can be several reasons for this. 1 mm × 1 mm cavities are relatively fine features for the PBT printing process [19]. Figure 12 shows that in some places small unwanted protrusions or fine PBT strings extend from an outer wall into the cavity. When the silver paste drops hit these unwanted structures, they can get stuck and prevent through-hole plating. The occurrence of such unwanted structures on the cavity wall increases with increasing component height, as reflected in the results of this paper.

Fig. 12
figure 12

Micrograph plane of a via specimen in which small unwanted protrusions or fine PBT strings can be detected

Full size image

The issue of drops not being fired perfectly vertically mentioned in Sect. 4.1 also plays a role here, increasing the likelihood of drops sticking to the cavity wall and thus preventing through-hole plating. The positioning of the nozzle above the cavities could also play a role. Even if cavities #1 and #12 were used for positioning, it cannot be ensured that the nozzle was positioned exactly over the cavities. Accordingly, this can also be a reason why the drops are more likely to stick to the cavity walls.

It has been shown that 5 mm deep cavities can be filled and thus serve as vias. However, it has also been shown that this method of producing vias is not entirely reliable. The easiest way would be to slightly enlarge the cavities, e.g. from 1 mm to 1.2 mm, so that the drops no longer hit the cavity walls. Another way to ensure the continuity of printed cavities would be to optimize the build strategy or the process parameters. For example, the sequence or number of contour lines can be changed or the deposition speed of the contour lines can be varied. However, for industrial production of such individualized sensors, the positioning of the nozzle and the cavities should rather be optimized, e.g. by using optical sensors. On the one hand, matching the APF parameters to the PBT material minimizes/eliminates small unwanted protrusions or fine strings. On the other hand, as discussed in Sect. 4.1, a better match between the dispensing system, process parameters, and material to each other would improve the quality of the drop ejection. Future studies should consider the influence of particle size distribution as well as temperature and shear rate-dependent viscosity of the silver pastes used on the ability to fill the vias should be considered.

4.3 Qualification tests of the sensors

4.3.1 Temperature shock test

One of the six sensors subjected to the temperature shock test did not pass. This sensor had a maximum reading of 9.28 V before the test. After the test, there was no measurable voltage. It is likely that one or more connections were destroyed. The exact location and cause of the electrical failure could not be determined. However, one factor may be a mismatch in the coefficients of thermal expansion of the materials used.

4.3.2 Humidity-heat test

Four sensors were subjected to this test, one of which did not pass. This sensor had a maximum measurement voltage of 9.83 V before the test and a maximum measurement voltage of 10.54 V after the test. Only the linearity curve showed a non-standard characteristic, so the test was rated as failed. The cause could not be pinpointed. However, the mismatch of thermal expansion coefficients is expected to be a major reason for this electrical failure.

For both the temperature shock test and the humidity-heat test, the use of other silver pastes and/or other curing parameters (see Sect. 4.4) could also be of interest for future testing.

4.3.3 Vibration and shock test

Both tested sensors passed the vibration and shock test, indicating that the materials and connections can withstand mechanical stress. However, the number of two samples is too small to draw any valid conclusions.

4.4 Curing

The curing temperature and time of the silver paste were not varied and investigated in this work. The lowest curing temperature (110 °C) given in the data sheet [21] was chosen to avoid unnecessary heating of the PBT and possible deformation. A curing time of 30 min was chosen, as this proved to be suitable for the somewhat thicker structures, compared to screen printing, in pre-tests. Higher temperatures up to 130 °C and longer or shorter curing times may have an influence on the curing mechanism and thus on the final properties of the silver paste and thus also on the adhesion properties and quality at the connections with the standard electrical components. However, since this is not evident from the silver paste data sheet, it is considered unlikely. However, it could be investigated in future work.

5 Conclusion

The test results have shown that the connection types described in this paper (slopes, vias, connection from above and connection from below) can be used to produce individualized inductive proximity sensors that pass industrial standard qualification tests. Due to some failures and the small number of samples, further investigation should be done.

The current prototype process involves several manual steps, including the transfer from the freeformer to the Fraunhofer test-rig where the conductive paste is dispensed, the insertion of the coil and the sensor board as well as the potting. In the future, automated processes will be required for more cost-effective production. Nevertheless, it was possible to produce geometrically individualized inductive proximity sensors within the scope of this work. In addition, limits or design guidelines for the microdispensing of the silver paste have been determined. This proof of concept enables new innovative solutions for customers in the field of automation which cannot be addressed with conventional sensors.