Aerosol Jet Printing and Interconnection Technologies on Additive Manufactured Substrates

: Nowadays, digital printing technologies such as inkjet and aerosol jet printing are gaining more importance since they have proven to be suitable for the assembly of complex microsystems. This also applies to medical technology applications like hearing aids where patient-speciﬁc solutions are required. However, assembly is more challenging than with conventional printed circuit boards in terms of material compatibility between substrate, interconnect material and printed ink. This paper describes how aerosol jet printing of nano metal inks and subsequent assembly processes are utilized to connect electrical components on 3D substrates fabricated by Digital Light Processing (DLP). Conventional assembly technologies such as soldering and conductive adhesive bonding were investigated and characterized. For this purpose, curing methods and substrate pretreatments for different inks were optimized. Furthermore, the usage of electroless plating on printed metal tracks for improved solderability was investigated. Finally, a 3D ear mold substrate was used to build up a technology demonstrator by means of conductive adhesives.


Introduction
Digital printing technologies such as inkjet and aerosol jet printing have proven to be suitable for the assembly of complex microsystems [1][2][3][4][5][6][7][8]. The printing results are dependent on many different factors including pre-and post-processing like plasma treatment, laser irradiation [3], as well as drying and curing procedures [5]. Furthermore, the material compatibility between ink and substrate must be taken into consideration [5,[9][10][11][12]. In many applications, substrates functionalized by digital printing need to be interconnected to conventionally manufactured parts, such as batteries, sensors, microcontroller units or other electronic components. These so called hybrid printed electronics combine the advantages of printed and conventional electronics [13][14][15]. It is assumed by industry and research that hybrid electronics will play an important role in the future of printed electronics [16]. In order to connect the printed structures with other components, several methods such as soldering, conductive adhesive bonding or overprinting can be applied. In this area, the material interaction and compatibility, e.g., between solder paste or conductive adhesive, respectively, and printed ink, play a decisive role as well [5,11,13,17]. Much work has been carried out regarding the functionalization of injection-molded thermoplastic substrates [1,2,5] and paper-based circuit boards [18,19] by digital printing. Isotropic conductive adhesives (ICA) [13] and low melting solder like bismuth-based solder paste [20] showed promising results. However, as additive manufacturing methods such as Fused Filament Fabrication (FFF), Multi Jet Modelling (MJM) or Digital Light Processing (DLP) are gaining increasing importance, also the materials used in those technologies appear in the focus of interest of digital printing, as well as the desire to print directly onto the 3D objects [6,[21][22][23][24]. Solder materials and processes. All solder materials used are listed in Table 1. The low temperature stability of the 3D printed ear mold materials limited the use of standard SAC (SnAgCu) solder. Consequently, mainly low-melting solders based on SnBi were investigated. The solder paste by Nordson, S42D500A5-A, is composed of 42 wt% tin and 58 wt% bismuth with a solidus temperature of 138 °C. DP5600 by Interflux is composed of 42 wt% tin, 57 wt% bismuth and additionally contains 1 wt% silver in order to achieve a certain "presaturation". Therefore, the diffusion rate in the interface between silver ink and solder material is reduced. All solder materials were applied in paste form by dispensing with respect to the non-planar target applications using a manual nano-dispense station (Martin-SMT, Germering, Germany) and a precision placement tool (Finetech, Berlin, Germany). The soldering tests were performed in a reflow soldering system Type XXS (N2) (SMT Maschinen-und Vertriebs GmbH & Co. KG, Wertheim, Germany). Different soldering profiles inspired by tolerances according to JEDEC standard (IPC/JEDEC J-STD-020) were tested for the SnBi and SnBiAg solders for different peak temperatures between 155 °C and 180 °C as presented in Figure 2.  Solder materials and processes. All solder materials used are listed in Table 1. The low temperature stability of the 3D printed ear mold materials limited the use of standard SAC (SnAgCu) solder. Consequently, mainly low-melting solders based on SnBi were investigated. The solder paste by Nordson, S42D500A5-A, is composed of 42 wt% tin and 58 wt% bismuth with a solidus temperature of 138 • C. DP5600 by Interflux is composed of 42 wt% tin, 57 wt% bismuth and additionally contains 1 wt% silver in order to achieve a certain "presaturation". Therefore, the diffusion rate in the interface between silver ink and solder material is reduced. All solder materials were applied in paste form by dispensing with respect to the non-planar target applications using a manual nano-dispense station (Martin-SMT, Germering, Germany) and a precision placement tool (Finetech, Berlin, Germany). The soldering tests were performed in a reflow soldering system Type XXS (N2) (SMT Maschinen-und Vertriebs GmbH & Co. KG, Wertheim, Germany). Different soldering profiles inspired by tolerances according to JEDEC standard (IPC/JEDEC J-STD-020) were tested for the SnBi and SnBiAg solders for different peak temperatures between 155 • C and 180 • C as presented in Figure 2. Solder materials and processes. All solder materials used are listed in Table 1. The low temperature stability of the 3D printed ear mold materials limited the use of standard SAC (SnAgCu) solder. Consequently, mainly low-melting solders based on SnBi were investigated. The solder paste by Nordson, S42D500A5-A, is composed of 42 wt% tin and 58 wt% bismuth with a solidus temperature of 138 °C. DP5600 by Interflux is composed of 42 wt% tin, 57 wt% bismuth and additionally contains 1 wt% silver in order to achieve a certain "presaturation". Therefore, the diffusion rate in the interface between silver ink and solder material is reduced. All solder materials were applied in paste form by dispensing with respect to the non-planar target applications using a manual nano-dispense station (Martin-SMT, Germering, Germany) and a precision placement tool (Finetech, Berlin, Germany). The soldering tests were performed in a reflow soldering system Type XXS (N2) (SMT Maschinen-und Vertriebs GmbH & Co. KG, Wertheim, Germany). Different soldering profiles inspired by tolerances according to JEDEC standard (IPC/JEDEC J-STD-020) were tested for the SnBi and SnBiAg solders for different peak temperatures between 155 °C and 180 °C as presented in Figure 2.  Isotropic Conductive Adhesives (ICA). Compared to the established soldering technology, the use of ICA is still limited, especially considering the contact resistance to non-noble metal surfaces. Over time, the contact resistance can increase due to oxidation and galvanic corrosion. Initial reliability tests on printed conductive tracks have yielded promising results for the non-noble metal surfaces as well [31]. In this work, both ICA types specified for noble and non-noble metal surfaces were used (Table 2).  [32]. Furthermore, another electrically conductive adhesive (ICA 2) for SMD attach for non-noble surfaces like Sn, SnPb, OSP, etc., was investigated. Both adhesives were applied in a cartridge for dispensing or as a film for stamping. The curing temperature was varied between 6 h at 80 • C and 30min at 120 • C in a lab oven UF55 from Memmert GmbH, GER.
Characterization of adhesive bond strength. Shear tests according to standard DIN EN 62137-1-2 [33] were utilized to examine the adhesion strength of the ICA bonded SMD. The shear strength of the connections was measured with a Dage 4000 bond tester from Nordson, Westlake, OH, USA.

Printing on 2D Test Substrates
The inks described in Section 2 were evaluated with 2D test substrates made from the materials Luxaprint 3D mold and Fototec DLP A. Thermal sintering at the required temperatures above 140 • C partly led to warpage due to the high substrate thickness of 4 mm and was, therefore, discarded in favor of photonic curing. In Table 3, the results of the complete printing process are shown with respect to printability, photonic curing and adhesion. According to these results, the silver ink Sicrys I30 EG-1 is the most suitable ink for the investigated substrates and printing method.

Printing on 3D Ear Mold Substrates
A further challenge is printing on 3D shaped objects with uneven surfaces and slopes. The aerosol jet system used in these experiments has some 3D capability, as opposed to most inkjet printers which rely on planar substrates. Thus, it was possible to print on certain areas of typical 3D shaped ear mold substrates made from either Fototec DLP A or Luxaprint 3D mold. The printing layout was thereby adapted to the footprint of the used MEMS-based pressure sensor and applied on different 3D substrates based on typical ear molds (Figures 3a and 4a). Conductive structures with high precision and edge sharpness could be printed even on relatively rough surfaces (Figures 3b and 4b).

Electroless Plating on Printed Silver Structures
In order to circumvent the challenge of dissolving the printed tracks in liquid solder due to high diffusion rates of silver described in the previous section, electroless copper plating was used to cover the printed silver tracks. The catalytic behavior of silver [34] makes printed silver a candidate for a starting layer for electroless copper plating. A further advantage of electroless plating is to decrease the electrical resistance and increase the current load capacity of the printed silver tracks. First trials with a plating time of 60 min led to partial delamination of the plated structures (Figure 5a,b), but with a reduction of the plating time to 15 min no such problems occurred (Figure 5c,d). The copper plating was carried out at a process temperature of 48 °C. The copper thickness was approximately 1.4 µm for both substrate materials, which fits to the plating characteristics of the electrolyte. The plating did not visibly affect the Luxaprint 3D material. FotoTec DLP-A, on the other hand, showed a slight deformation. Tape tests after plating for first evaluation of the adhesion were passed for both materials.
By electroless copper plating of test structures on planar test substrates ( Figure 6), a significant reduction of the electrical resistance (Table 4) was achieved. Also on 3D substrates, successful electroless copper plating could be carried out (Figure 7). Using Luxaprint 3D mold substrates in combination with the Ag ink Sycris I30 EG-1, on the copper plated patterns a Ni/Au surface finish finally could be applied ( Figure 8). Ni/Au

Electroless Plating on Printed Silver Structures
In order to circumvent the challenge of dissolving the printed tracks in liquid solder due to high diffusion rates of silver described in the previous section, electroless copper plating was used to cover the printed silver tracks. The catalytic behavior of silver [34] makes printed silver a candidate for a starting layer for electroless copper plating. A further advantage of electroless plating is to decrease the electrical resistance and increase the current load capacity of the printed silver tracks. First trials with a plating time of 60 min led to partial delamination of the plated structures (Figure 5a,b), but with a reduction of the plating time to 15 min no such problems occurred (Figure 5c,d). The copper plating was carried out at a process temperature of 48 °C. The copper thickness was approximately 1.4 µm for both substrate materials, which fits to the plating characteristics of the electrolyte. The plating did not visibly affect the Luxaprint 3D material. FotoTec DLP-A, on the other hand, showed a slight deformation. Tape tests after plating for first evaluation of the adhesion were passed for both materials.
By electroless copper plating of test structures on planar test substrates ( Figure 6), a significant reduction of the electrical resistance (Table 4) was achieved. Also on 3D substrates, successful electroless copper plating could be carried out ( Figure 7). Using Luxaprint 3D mold substrates in combination with the Ag ink Sycris I30 EG-1, on the copper plated patterns a Ni/Au surface finish finally could be applied ( Figure 8). Ni/Au

Electroless Plating on Printed Silver Structures
In order to circumvent the challenge of dissolving the printed tracks in liquid solder due to high diffusion rates of silver described in the previous section, electroless copper plating was used to cover the printed silver tracks. The catalytic behavior of silver [34] makes printed silver a candidate for a starting layer for electroless copper plating. A further advantage of electroless plating is to decrease the electrical resistance and increase the current load capacity of the printed silver tracks. First trials with a plating time of 60 min led to partial delamination of the plated structures (Figure 5a,b), but with a reduction of the plating time to 15 min no such problems occurred (Figure 5c,d). The copper plating was carried out at a process temperature of 48 • C. The copper thickness was approximately 1.4 µm for both substrate materials, which fits to the plating characteristics of the electrolyte. The plating did not visibly affect the Luxaprint 3D material. FotoTec DLP-A, on the other hand, showed a slight deformation. Tape tests after plating for first evaluation of the adhesion were passed for both materials. The differences in electrical resistance between the two materials before the plating process may be due to the different substrate properties, since the printing parameters for both substrates were identical. Chemical and physical material properties of the substrates By electroless copper plating of test structures on planar test substrates ( Figure 6), a significant reduction of the electrical resistance (Table 4) was achieved. Also on 3D substrates, successful electroless copper plating could be carried out ( Figure 7). Using Luxaprint 3D mold substrates in combination with the Ag ink Sycris I30 EG-1, on the copper plated patterns a Ni/Au surface finish finally could be applied ( Figure 8). Ni/Au plating was carried out using standard plating parameters (plating time of 17 min each at 88 • C (Ni) and 86 • C (Au)).  The differences in electrical resistance between the two materials before the plating process may be due to the different substrate properties, since the printing parameters for both substrates were identical. Chemical and physical material properties of the substrates can influence the wetting behavior of the ink as well as the sintering process. Sintering of conductive inks is influenced by the substrate's coefficient of thermal expansion, and with regard to photonic sintering by the substrate's light absorption (e.g., color).

SMD Assembly on Printed Silver Structures by Soldering
The before mentioned high diffusion rate of Ag in molten solder during soldering leads to a quick leaching of the printed conductor tracks into the solder. Previous works [28] on inkjet-printed conductor tracks showed the influence of the sintering temperature of the inks on the dissolving behavior during the soldering process. In the case of aerosol can influence the wetting behavior of the ink as well as the sintering process. Sintering of conductive inks is influenced by the substrate's coefficient of thermal expansion, and with regard to photonic sintering by the substrate's light absorption (e.g., color).

SMD Assembly on Printed Silver Structures by Soldering
The before mentioned high diffusion rate of Ag in molten solder during soldering leads to a quick leaching of the printed conductor tracks into the solder. Previous works The differences in electrical resistance between the two materials before the plating process may be due to the different substrate properties, since the printing parameters for both substrates were identical. Chemical and physical material properties of the substrates can influence the wetting behavior of the ink as well as the sintering process. Sintering of conductive inks is influenced by the substrate's coefficient of thermal expansion, and with regard to photonic sintering by the substrate's light absorption (e.g., color).

SMD Assembly on Printed Silver Structures by Soldering
The before mentioned high diffusion rate of Ag in molten solder during soldering leads to a quick leaching of the printed conductor tracks into the solder. Previous works [28] on inkjet-printed conductor tracks showed the influence of the sintering temperature of the inks on the dissolving behavior during the soldering process. In the case of aerosol jet printed thin conductor tracks, the dissolving is even more critical. First, preliminary tests with SAC solder on the 2D test layout confirmed these phenomena on ear mold substrates. The soldering result (5s with hot air) on such printed structures is displayed in Figure 9. jet printed thin conductor tracks, the dissolving is even more critical. First, preliminary tests with SAC solder on the 2D test layout confirmed these phenomena on ear mold substrates. The soldering result (5s with hot air) on such printed structures is displayed in Figure 9. Pads fabricated by aerosol printing of two layers with a thickness of approx. 0.4 µm were dissolved by SnBi solder almost immediately ( Figure 10). Residues of the pad's edges can still be noticed. The results are similar to those obtained on inkjet-printed pads [28,31] where almost all contact pads had conductor breaks after soldering with SnBi solder due to dissolving of the tracks in the soldering paste. One possible measure to counter the effects of dissolving is to increase the Ag layer thickness by additional printed layers to delay the leaching process. However, this results in longer process times and, thus, in higher production costs. Within this work, up to 12 layers were investigated. To limit the number of layers needed, it is highly desirable to lower the diffusion rate. The rate depends, among other factors, on the process temperature. Thus, by adapting the process parameters the diffusion rate can be lowered. The following parameters had been investigated regarding the influence on the temperature:

•
Reduction of the peak temperature during soldering. A lower temperature leads to a slower diffusion rate. This is also the reason why low melting solders are promising.

•
Increased heating speed and faster cooling for a shorter solder profile and shorter time over liquidus temperature.
Soldering with low-melting SnBi solder on 12 printed Ag layers showed a slightly Pads fabricated by aerosol printing of two layers with a thickness of approx. 0.4 µm were dissolved by SnBi solder almost immediately ( Figure 10). Residues of the pad's edges can still be noticed. The results are similar to those obtained on inkjet-printed pads [28,31] where almost all contact pads had conductor breaks after soldering with SnBi solder due to dissolving of the tracks in the soldering paste. jet printed thin conductor tracks, the dissolving is even more critical. First, preliminar tests with SAC solder on the 2D test layout confirmed these phenomena on ear mol substrates. The soldering result (5s with hot air) on such printed structures is displayed i Figure 9. Pads fabricated by aerosol printing of two layers with a thickness of approx. 0.4 µm were dissolved by SnBi solder almost immediately ( Figure 10). Residues of the pad' edges can still be noticed. The results are similar to those obtained on inkjet-printed pad [28,31] where almost all contact pads had conductor breaks after soldering with SnB solder due to dissolving of the tracks in the soldering paste. One possible measure to counter the effects of dissolving is to increase the Ag laye thickness by additional printed layers to delay the leaching process. However, this result in longer process times and, thus, in higher production costs. Within this work, up to 1 layers were investigated. To limit the number of layers needed, it is highly desirable t lower the diffusion rate. The rate depends, among other factors, on the proces temperature. Thus, by adapting the process parameters the diffusion rate can be lowered The following parameters had been investigated regarding the influence on th temperature:

•
Reduction of the peak temperature during soldering. A lower temperature leads to slower diffusion rate. This is also the reason why low melting solders are promising • Increased heating speed and faster cooling for a shorter solder profile and shorte One possible measure to counter the effects of dissolving is to increase the Ag layer thickness by additional printed layers to delay the leaching process. However, this results in longer process times and, thus, in higher production costs. Within this work, up to 12 layers were investigated. To limit the number of layers needed, it is highly desirable to lower the diffusion rate. The rate depends, among other factors, on the process temperature. Thus, by adapting the process parameters the diffusion rate can be lowered. The following parameters had been investigated regarding the influence on the temperature:

•
Reduction of the peak temperature during soldering. A lower temperature leads to a slower diffusion rate. This is also the reason why low melting solders are promising. • Increased heating speed and faster cooling for a shorter solder profile and shorter time over liquidus temperature.
Soldering with low-melting SnBi solder on 12 printed Ag layers showed a slightly better behavior. Still, even with increased layer thickness and reduced peak temperature of 155 • C during soldering, the pads were still dissolved as shown in Figure 11. A further reduction of the peak temperature was not feasible because no complete melting of the solder was achieved. formed on structures made from ink LAB 2271. This concerns especially the wettability that is crucial for an excellent solder connection. The contact angle is far above the limit that could be considered for acceptable wetting (Figure 12). Pretreatment with atmospheric plasma (260 V at a frequency of 21 kHz and traversing speed of 100 mm/s) slightly improved the wetting behavior of the solder on the printed LAB 2271 inks, but was still not satisfying with contact angles over 90°.  Another elementary factor is the material combination of solder and conductive layer, e.g., copper [35,36] dissolves much more slowly in tin than silver [30,37]. The copper inks showed very poor behavior on the used substrate materials, as mentioned before. Based on that, the following other measures regarding the materials were investigated:

•
Using solder with silver content using saturation effects. • Electroless copper plating of printed silver layers Using solder with silver content leads to a better soldering behavior, so that better solder connections were obtained in the preliminary tests with inkjet-printed Ag pads on 2D substrates [28,31]. On aerosol jet printed structures, the silver content in the solder leads to an improvement as well. Nevertheless, dissolving of the very thin aerosol jet printed layers hinders reproducible soldering results.
Furthermore, in the case of thick-walled substrate bodies, such as ear molds, high thermomechanical forces probably occur during the soldering process, which leads to cracks and consequently damage of the substrates, as can be seen in Figure 13 below. A reflow soldering profile with a lower temperature gradient could reduce thermomechanical stress, but at the same time extend the duration of increased temperature exposure and counterproductively affect the overall diffusion zone.
Unlike the pure Ag printed structures, conductor structures plated with electroless copper could be soldered without any significant dissolving. However, if the process time of electroless copper plating is too long (60 min), the whole structures tend to delaminate partially after soldering, as can be seen in Figure 14a. Nevertheless, if the plating time is only 15 min, the conductor structures consistently withstood the soldering process apart The solder pastes behaved differently depending on the sintering parameters and the inks used for printing the Ag structures, e.g., the meniscus of the solder joint could not be formed on structures made from ink LAB 2271. This concerns especially the wettability that is crucial for an excellent solder connection. The contact angle is far above the limit that could be considered for acceptable wetting (Figure 12). Pretreatment with atmospheric plasma (260 V at a frequency of 21 kHz and traversing speed of 100 mm/s) slightly improved the wetting behavior of the solder on the printed LAB 2271 inks, but was still not satisfying with contact angles over 90 • . formed on structures made from ink LAB 2271. This concerns especially the wettability that is crucial for an excellent solder connection. The contact angle is far above the limit that could be considered for acceptable wetting (Figure 12). Pretreatment with atmospheric plasma (260 V at a frequency of 21 kHz and traversing speed of 100 mm/s) slightly improved the wetting behavior of the solder on the printed LAB 2271 inks, but was still not satisfying with contact angles over 90°.  Another elementary factor is the material combination of solder and conductive layer, e.g., copper [35,36] dissolves much more slowly in tin than silver [30,37]. The copper inks showed very poor behavior on the used substrate materials, as mentioned before. Based on that, the following other measures regarding the materials were investigated: • Using solder with silver content using saturation effects. • Electroless copper plating of printed silver layers Using solder with silver content leads to a better soldering behavior, so that better solder connections were obtained in the preliminary tests with inkjet-printed Ag pads on 2D substrates [28,31]. On aerosol jet printed structures, the silver content in the solder leads to an improvement as well. Nevertheless, dissolving of the very thin aerosol jet printed layers hinders reproducible soldering results.
Furthermore, in the case of thick-walled substrate bodies, such as ear molds, high thermomechanical forces probably occur during the soldering process, which leads to cracks and consequently damage of the substrates, as can be seen in Figure 13 below. A Another elementary factor is the material combination of solder and conductive layer, e.g., copper [35,36] dissolves much more slowly in tin than silver [30,37]. The copper inks showed very poor behavior on the used substrate materials, as mentioned before. Based on that, the following other measures regarding the materials were investigated:

•
Using solder with silver content using saturation effects. • Electroless copper plating of printed silver layers.
Using solder with silver content leads to a better soldering behavior, so that better solder connections were obtained in the preliminary tests with inkjet-printed Ag pads on 2D substrates [28,31]. On aerosol jet printed structures, the silver content in the solder leads to an improvement as well. Nevertheless, dissolving of the very thin aerosol jet printed layers hinders reproducible soldering results.
Furthermore, in the case of thick-walled substrate bodies, such as ear molds, high thermomechanical forces probably occur during the soldering process, which leads to cracks and consequently damage of the substrates, as can be seen in Figure 13 below. A reflow soldering profile with a lower temperature gradient could reduce thermomechanical stress, but at the same time extend the duration of increased temperature exposure and counterproductively affect the overall diffusion zone. from the cracks in the massive substrate bodies, as shown in Figure 14b. After electroless copper plating of the printed Ag structures, electroless plating of a Ni/Au surface finish also can be carried out. Figure 15 shows pressure sensors soldered onto both ear mold bodies without dissolving the pads. Nevertheless, one can see clearly visible cracks in the substrate due to the effects described above.
(a) (b)  Unlike the pure Ag printed structures, conductor structures plated with electroless copper could be soldered without any significant dissolving. However, if the process time of electroless copper plating is too long (60 min), the whole structures tend to delaminate partially after soldering, as can be seen in Figure 14a. Nevertheless, if the plating time is only 15 min, the conductor structures consistently withstood the soldering process apart from the cracks in the massive substrate bodies, as shown in Figure 14b. After electroless copper plating of the printed Ag structures, electroless plating of a Ni/Au surface finish also can be carried out. Figure 15 shows pressure sensors soldered onto both ear mold bodies without dissolving the pads. Nevertheless, one can see clearly visible cracks in the substrate due to the effects described above. from the cracks in the massive substrate bodies, as shown in Figure 14b. After electroless copper plating of the printed Ag structures, electroless plating of a Ni/Au surface finish also can be carried out. Figure 15 shows pressure sensors soldered onto both ear mold bodies without dissolving the pads. Nevertheless, one can see clearly visible cracks in the substrate due to the effects described above.

SMD Assembly on Printed Silver Structures by Means of ICA
Due to the observed cracks in the ear mold material after soldering, adhesive bonding using ICA was investigated. Here, cracks in the substrates do not occur after curing at temperatures around 120 • C and consequently the ear molds withstood all thermal processes for curing the conductive adhesives. Varying the parameters temperature and curing time had a high influence on the adhesion strength. This is reflected by the measured shear strengths of SMD components. The shear strengths of 0603 SMD on the used ear mold materials for both used ICAs and different curing profiles are displayed in Figure 16.

SMD Assembly on Printed Silver Structures by Means of ICA
Due to the observed cracks in the ear mold material after soldering, adhesive bonding using ICA was investigated. Here, cracks in the substrates do not occur after curing at temperatures around 120 °C and consequently the ear molds withstood all thermal processes for curing the conductive adhesives. Varying the parameters temperature and curing time had a high influence on the adhesion strength. This is reflected by the measured shear strengths of SMD components. The shear strengths of 0603 SMD on the used ear mold materials for both used ICAs and different curing profiles are displayed in Figure 16. This illustrates that the process parameters have a significant influence on the shear strengths for both substrate materials used. Yet, the fracture patterns are different, as shown in Figure 17. On the substrate made from FotoTec SL.E, the interface between substrate and printed structure partially failed. The overall bond strength on the substrate made from Luxaprint 3D mold is higher. Here, mainly the interface between printed structure and adhesive failed. The reason could be a different roughness and the assumption is that interlocking is the crucial adhesion mechanism. This illustrates that the process parameters have a significant influence on the shear strengths for both substrate materials used. Yet, the fracture patterns are different, as shown in Figure 17. On the substrate made from FotoTec SL.E, the interface between substrate and printed structure partially failed. The overall bond strength on the substrate made from Luxaprint 3D mold is higher. Here, mainly the interface between printed structure and adhesive failed. The reason could be a different roughness and the assumption is that interlocking is the crucial adhesion mechanism. This illustrates that the process parameters have a significant influence on the shear strengths for both substrate materials used. Yet, the fracture patterns are different, as shown in Figure 17. On the substrate made from FotoTec SL.E, the interface between substrate and printed structure partially failed. The overall bond strength on the substrate made from Luxaprint 3D mold is higher. Here, mainly the interface between printed structure and adhesive failed. The reason could be a different roughness and the assumption is that interlocking is the crucial adhesion mechanism.

Functionalized Ear Mold as Technology Demonstrator
For demonstrating the digital process chain using additive manufacturing of the ear mold, aerosol jet printing of the conductor tracks and assembly of the electronic components by ICA, a technology demonstrator ( Figure 18)

Functionalized Ear Mold as Technology Demonstrator
For demonstrating the digital process chain using additive manufacturing of the ear mold, aerosol jet printing of the conductor tracks and assembly of the electronic components by ICA, a technology demonstrator ( Figure 18) was built up. The assembly process of electronic components was transferred to a fully automated assembly process on 3D substrates. For automated 3D assembly on production equipment (XTec from Häcker Automation GmbH, Germany), the ear mold parts are threedimensionally aligned via a 3D substrate holder with axis system. Position detection is performed by a camera system that can reference the x and y position as well as the z coordinate with high precision. In combination with referencing the components to be assembled from their bottom side after picking, it is possible to assemble with high precision in the same way as on planar substrates. By pin transfer, very small amounts of ICA can be transferred to the 3D substrates, and SMD components with small connection pitches can be assembled. The ICA was cured thermally in a conventional oven. Up to a tilting angle of 45°, no displacement of the assembled components was observed. The SMD components could be successfully electrically contacted on both ear mold materials FotoTec DLP-A and Luxaprint 3D mold, as well as on PlastCure Rigid10500. The latter material was only intended for setting up the assembly process but both, printability and photonic curing, turned out to be very good as well. All demonstrators were realized with five printed Ag layers (ink: Sicrys I30 EG-1) at a tube temperature of 55 °C, sheath gas 70 cm 3 , atomizer 555 cm³ and impactor 520 cm 3 .

Conclusions
Digital printing enables conductive structures on materials and topographies other The assembly process of electronic components was transferred to a fully automated assembly process on 3D substrates. For automated 3D assembly on production equipment (XTec from Häcker Automation GmbH, Waltershausen, Germany), the ear mold parts are three-dimensionally aligned via a 3D substrate holder with axis system. Position detection is performed by a camera system that can reference the x and y position as well as the z coordinate with high precision. In combination with referencing the components to be assembled from their bottom side after picking, it is possible to assemble with high precision in the same way as on planar substrates. By pin transfer, very small amounts of ICA can be transferred to the 3D substrates, and SMD components with small connection pitches can be assembled. The ICA was cured thermally in a conventional oven. Up to a tilting angle of 45 • , no displacement of the assembled components was observed. The SMD components could be successfully electrically contacted on both ear mold materials FotoTec DLP-A and Luxaprint 3D mold, as well as on PlastCure Rigid10500. The latter material was only intended for setting up the assembly process but both, printability and photonic curing, turned out to be very good as well. All demonstrators were realized with five printed Ag layers (ink: Sicrys I30 EG-1) at a tube temperature of 55 • C, sheath gas 70 cm 3 , atomizer 555 cm 3 and impactor 520 cm 3 .

Conclusions
Digital printing enables conductive structures on materials and topographies other than those of printed circuit boards, which opens up a new range of material combinations with customizable properties. Therefore, in this work different DLP-capable substrate materials, nano metal inks and state-of-the-art assembly technologies were selected and investigated. The interconnection technology was conductive adhesive bonding and lowtemperature soldering by bismuth-based lead-free solder. The results have demonstrated the challenges but also new possibilities for the combination of those additive manufacturing, digital printing and assembly technologies.
Different nano metal inks (Ag and Cu) were evaluated on substrates made of Luxaprint 3D mold and Fototec DLP A. Thermal sintering at required temperatures above 140 • C led to warpage; therefore, photonic curing was identified as an alternative. The inks showed completely different behavior in terms of printability, photonic curing and adhesion strength. The silver ink Sicrys I30 EG-1 proved to be the most suitable ink for the investigated substrates and printing method. In the end, the results showed that nano metal ink and substrate need to be adapted specifically for each combination.
The possibility to reinforce aerosol jet printed Ag layers on additive manufactured substrates by electroless copper plating with optional Ni/Au surface finish has been shown. This approach did not only reduce the electrical resistance of the layers, but also showed the basic solderability by bismuth-based lead-free solder. However, the substrate materials used showed cracks after the soldering process. Although this leads to the conclusion that this approach is not feasible for both tested UV-curable resin materials, it shows the general potential for other materials and complex geometries. This is especially of interest if reduction of the electrical resistance, increasing current carrying capacity and solderability are desired.
Unlike soldering, ICA has proven to be a viable option for SMD assembly on aerosol jet printed structures on several substrate materials. The achieved shear forces of attached 0603 SMD components strongly depended on the curing profiles. The applicability of ICA by dispensing for SMD assembly on 3D-shaped substrates has been shown.
Combining additive manufacturing, functionalization by aerosol jet printing of nano silver inks and photonic sintering, and conductive adhesive bonding enables a novel digital process chain for functionalized and individualized products from different branches. Further work will focus on the reliability of the advanced functional integration of sensor components, the aforementioned printing and interconnection technologies, and the scalability of the digital process chain. More detailed research is needed to understand the cause-effect relationship between ink and substrate material.

Conflicts of Interest:
The authors declare no conflict of interest.