Abstract
This paper demonstrates the NeXus system, a multiscale robotic additive manufacturing platform developed at the Louisville Automation and Robotics Research Institute, as a rapid prototyping tool through additively manufacturing a multilayer flexible printed circuit board (FPC) with a printed strain sensor and soldered surface mounted devices (SMD). Manufacturing of the demonstrator requires the application and curing of multiple materials with specialized properties, tools for automated assembly, and software advances to streamline the process enabling the use of industry-standardized programs to command the NeXus system. Additive manufacturing processes supported by the NeXus include aerosol jet printing (AJP) for fine feature silver conducting lines, direct write ink-jet printing for insulating materials, and intense pulsed light (IPL) for curing materials between depositions. The NeXus system transports and manipulates parts using a six-degree-of-freedom (DOF) high-precision positioner. Solder paste deposition and pick-and-place (PnP) procedures are performed by a 4DOF Selective Compliance Articulated Robot Arm (SCARA). Conversion methods between traditional printed circuit board (PCB) design software and production-ready command scripts were developed to translate basic drawings into command scripts. Multilayer structures with AJP 50-μm wide lines, an insulating bridge with a thickness of around 100 μm, and SMDs soldered to silver AJP pads were integrated within the demonstrator. An operational amplifier and other SMDs reduce the complexity of the accompanying control circuit and amplify the sensor's response by 1830 times. The successful fabrication of the demonstrator FPC highlights the rapid prototyping ability of the NeXus system.
1 Introduction
Additive manufacturing, specifically early versions of three-dimensional (3D) printing, has long been synonymous with rapid prototyping, the terms often being used interchangeably. This relationship dates back to the first stereolithography apparatus 3D printer invented by Hideo Kodama in his 1981 article [1], who was later coined as one of the fathers of modern rapid prototyping [2]. Since its invention additive manufacturing has become an integral part of new product development. It reduces time to market, reduces costs, and improves product quality. This transition along with other fundamentals of additive technologies is reviewed in the work by Negi et al. [3].
The benefits of rapid prototyping have been expanded beyond novel additive manufacturing techniques to adapt traditional manufacturing technologies for rapid prototyping tasks such as the noncontact forms of printed electronics [4,5]. Other methods include processes and techniques that introduce functional and “smart” parts as discussed by Löher et al. [6]. Examples include the fully additive, passive electronic components presented by Tan et al. [7] or chipless radio frequency identification pressure sensors from Brinker and Zoughi [8]. Additive manufacturing processes such as fused deposition model (FDM) 3D printing or inkjet printing have been validated as methods of rapid prototyping for reducing lead times on simple small-batch products or prototypes. These processes can effectively produce plastic parts, conductive wire traces, or print other specialized materials in small batches.
Another approach enabling further optimization of the manufacturing process is a combination of different printing techniques in a single workflow allowing realization of the structures with increased complexity and simultaneously reducing fabrication times. Jin et al. developed a hybrid multimaterial 3D printing method involving photocuring during the dispensing of the material [9]. Flexible circuits with tunable microheaters were fabricated with the help of direct laser machining and electrodeposition [10]. Furthermore, it was reported that aerosol jet printing (AJP) was used to manufacture 3D metal-dielectric structures to realize low-loss passives and GHz wavelength antennas with applications in wearable and Internet-of-Things (IoT) devices [11].
The development of this technology to produce an all-inclusive printed circuit board (PCB) manufacturing system stems from areas including space development, as showcased in the work of Paek et al. [12] or to reduce waste and improve the environmental impact of PCB manufacturing and development as detailed by Dong et al. [13]. As pointed out by Lall et al. [14], most developments in the additive PCB space have focused on single-layer printing in contrast with the industry standard where PCBs are double-sided at a minimum utilizing the front and the back planes of the PCB. In this work, they explored utilizing a conducting ink and insulating polymer in unison to develop a stacked multilayer approach. These techniques are similar to the techniques utilized in this study; however, they were able to utilize the same printing tool with different materials. Our study separates the materials across printing tools for collaborative printing between hardware. Similar to our method of combining printing techniques across materials, however at a larger scale, Arnal et al. [15] combined traditional FDM printing with microdispensing techniques to produce a multilayered part with several fully printed features including a CP dipole antenna, bandpass filters, and a switched-line phase shifter.
The NeXus is a multiscale additive manufacturing system, developed in our lab with the purpose of integrating precision robotic tools and multiple additive manufacturing methods to rapidly prototype functional structures from mircro to meso to macroscales. In comparison with the previously described multimaterial hybrid systems, the primary advantage of the NeXus system is its integration of industry-standard tools in a configuration suitable for automation and flexibility for generalized parts. The NeXus is comprised of two industrial robotic arms, a custom six-degree-of-freedom (DOF) positioner, an aerosol jetting (AJP) station, a 3D fused deposition (FDM) printing station, a PicoPlse® drop-on-demand (DOD) ink-jet deposition station, a micro-assembly station, an e-textile loom station, and a tool changer for the robotic arms which includes a custom 3D FDM end effector, auger valve, pick-and-place (PnP) tool, and robotic gripper. The 6DOF positioner acts as the motion stage servicing most of the additive manufacturing tools and the robotic arms are used for PnP procedures, as well as solder deposition to form interconnects. This study focuses on the application and integration of several processes on the NeXus manufacturing line, the AJP, PicoPlse®, auger valve, and PnP tool.
The NeXus system has been evaluated in previous work for its capacity for microsystem integration [16], hybrid sensor manufacturing using the Nordson EFD PicoPlse® [17,18], the gauge factor of silver ink printed on the Optomec [19], and various fully additively manufactured strain gauge designs [20,21]. Such designs include the strain gauge design which combines an arc and a radial pattern presented by Wei et al. [21], shown in Fig. 1. In addition to this work, extensive evaluation of the robotics tools on the NeXus system was performed and evaluated by Dr. Wei in his 2022 Master's Thesis work [22].
![Combined arc and radial strain gauge as designed and studied by Wei et al. [21]](https://asmedc.silverchair-cdn.com/asmedc/content_public/journal/micronanomanufacturing/13/1/10.1115_1.4067037/1/m_jmnm_013_01_011002_f001.png?Expires=1741625962&Signature=CbbnZpvkyHAswzh1QVj2yRsp4MKK5sNdBtMjeOfG7hH8~7kY4ekDOB96aTy2Wq2eNCrYo4Lh1XdehQ1UX3yvftz2lpkdkcsIqb7tNCwN84RUwHy9J0vr630a1MhBjJs5vdrwUFQpGSCSoC8sRXN9TyqWK2k8mhAQfQRr2H6QBiQ2GZon~G01Kx22n88thnTUWbZj90v5i7kOYMBFSLVxiL2HIZftd9QLgw~JusQQreWZ2c~zqK7tDEKJ~2XtiGFTNFed9edsLojkEMXUoRXp8arhTnkSHvZUGEP61t~l1jD5NiTyVQg~rvgD~bTmOwX9PmTXjFocUqwPxfiy69rzxg__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Combined arc and radial strain gauge as designed and studied by Wei et al. [21]
To showcase the NeXus system as a rapid prototyping tool, a demonstrator was developed to highlight its capabilities and test the processes required for this application. The successful manufacturing of this demonstrator would examine all of the necessary material interfaces and verify the technical capabilities for producing more generalized devices and products on the NeXus system. The results of the development of these systems work to demonstrate the NeXus system as a rapid prototyping tool with the ability to produce functional and accurate prototypes and custom devices in small batches. In a more general context, we demonstrated an approach for the automated multiscale manufacturing of flexible circuit devices utilizing robotic tools to enhance reliability during critical phases of the fabrication process requiring precision and repeatability, as well as introducing flexibility concerning design modifications. In this work, we develop and study the process of designing and testing an additively manufactured multilayered flexible printed circuit board (FPC) with a printed strain gauge [21] and supporting surface mounted device (SMD) based circuit as a demonstrator of the capability of the NeXus system as a rapid prototyping tool.
2 Design of the Demonstrator PCB
The demonstrator FPC was designed to highlight the NeXus' ability to emulate traditional PCB manufacturing through additive methods as well as demonstrate the novel advantages of additive manufacturing. The demonstrator FPC consists of a printed strain gauge [21] with an instrumentation amplifier (Texas Instruments, INA821IDR, Dallas, TX [23]), gain set resistor, decoupling capacitors, and set of contact pads for interfacing (Fig. 2).

Demonstrator FPC design featuring decoupling capacitors, instrumentation amplifier, gain set resistor, and strain gauge
The demonstrator FPC was designed in kicad, as mentioned above, which displays the ability to start in a traditional PCB CAD package and utilize the pipeline for automated manufacturing on the NeXus system. The strain gauge highlights a novelty benefit of additive manufacturing by integrating sensitive sensing circuitry directly into the FPC design, eliminating the hassle of integrating a cleanroom-manufactured component or specialized additional hardware. The integration of SMD devices, in addition to multilayered trace routing, showcases the ability to emulate traditionally manufactured PCBs.
2.1 Demonstrator PCB Manufacturing Process.
The manufacturing process for the demonstration FPC is broken down in the flowchart in Fig. 3. This process consists of generating a PCB design and then translating that design to executable printing files. From there, the printing files are run in a prepared sequence: the first layer of conductive material is printed, then the insulating layer, and finally the second layer of conductive material, curing the materials between each step. This series of steps realizes the base FPC. From here, the FPC can be assembled through a process of solder paste deposition, component placement, and finally solder reflow. With the FPC complete, it is then tested for functionality. These steps are presented through the lens of our demonstration design but can be generalized to any PCB design.

Flowchart of the preplanned process of multilayer FPC manufacturing and assembly on the NeXus system
2.2 Material Interfaces.
The demonstrator PCB incorporates three material interfaces or processes that were validated, tested, or optimized for this application.
AJP conductor on Kapton substrate cured with intense pulsed light (IPL)
insulator to conductor
conductive pads to solder paste
2.2.1 Aerosol Jet Printing Conductor on Kapton Substrate Cured With Intense Pulsed Light.
Material printing directly on Kapton substrate has been extensively characterized by the work from Ratnayake et al. [19] and Sherehiy et al. [18]. For the application of rapid prototyping, the demonstrator attempts to make use of an IPL curing system capable of rapidly sintering AJP silver traces. The alternative to this process is to remove the sample from the system for treatment in an oven which can take 16–24 h or longer at 200 °C. In addition to being more efficient, IPL treatment is a surface treatment meaning the heat required for sintering does not deeply penetrate the part [24]. This improves material compatibility for alternative substrates such as FDM printed components which for polylactic acid plastic has a melting point between 140 °C and 160 °C.
2.2.2 Insulator to Conductor Interface.
In traditional multilayer PCB manufacturing, interconnects between layers are achieved through features called “vias.” These can be plated through holes which allow for soldering to through board components or buried/hidden vias which allow for layer interconnects in higher-density board applications. In all of these examples, layers of the circuit board are treated as exclusive planes, and the via is the only method of moving between them. In the demonstrator design, the multilayer effect is achieved using a simple bridge design which uses a printed insulating polymer over the conductors of the first layer. The second layer conductor is then printed over this bridge connecting the first layer to the second layer. This method reduces printing time and material by minimizing the second layer. A comparison of these two multilayering methods is shown in Fig. 4.

(a) Traditional multilayer PCB fabrication using vias for interconnects between layers and (b) additive multilayer PCB fabrication using insulating bridging for overlapping conductors
The AJP traces are printed at a consistent angle normal to the substrate. By introducing the bridging features, this affects the substrate angle. The demonstrator in this instance is able to leverage the 5 mm working space of the AJP printer to account for height differences without affecting the deposition quality; however, this did require optimization of the contact angle between the insulating material and the substrate. The contact angle was minimized to improve the printing angle of the AJP tool. Plasma treatment of the Kapton substrate was shown to be effective in improving the wetting effect of the insulating material reducing the contact angle but increasing process complexity. Plasma treatment was performed outside of the NeXus system requiring removal and recalibration of the substrate before printing.
2.2.3 Conductive Pads to Solder Paste Interface.
The solder paste selected for this study was based on the work of Juric et al. [25] who studied methods for assembling SMDs to inkjet-printed silver structures. Leaching was found to be the primary hazard of soldering ink jep printed silver contacts. Leaching refers to the solder, during reflow, causing a disturbance in the pad. Often this results in deteriorating the material or removing it entirely. This was effectively combated by ensuring the solder paste contained trace amounts of the pad material. We selected a low-cost alternative to DP 5600 solder paste. Both pastes have a composition of Sn42Bi57Ag1. The bismuth makes this a low-temperature solder paste capable of reflowing at 135 °C further improving compatibility with the silver ink which sinters at 200 °C.
2.3 Testing Tools.
The sensors were tested by observing the output in response to a static load. This was done through a power supply, a testing circuit, and a test bed. The testing circuit diagram is shown in Fig. 5, where V is the power signal into the strain gauge. Arc and radial resistors represent the arc and radial structures of the strain gauge. V is the calibration signal provided from V.
The supporting test circuit consists of a TPS65130EVM-839 split-rail converter, a NE5532P operational amplifier, and two 5 k potentiometers, V and V. The power supply provides 5 V to a split-rail converter, producing +5 V and −5 V. The operational amplifier uses the output of V to create a voltage follower circuit tuned to produce 100 mV for the V of the sensor. The output of V is used as the −IN of the sensor's instrumentation amplifier. The circuit is constructed on a protoboard and ±5 V, V, GND, and V are wired to the test bed. The test bed consists of a connector PCB containing leads protruding downward, which is fixed through screws to an insulated sensor bed. The sensor is placed under the leads and the connector is screwed down so that the leads are pressed onto the matching pads of the FPC. Once power is supplied through the circuits, the output can be read through a Hioki lr8431-20 memory hilogger data acquisition device. V can then be tuned to bring the output of the circuit to 0 V. The sensor bed is placed under a dial gauge stand modified for the static loading operation to apply a known amount of load onto the FPC, while the output is observed and recorded through the data acquisition device. Testing of the sensor and circuit was performed by comparing an amplified sensor with a nonamplified sensor.
3 NeXus System Manufacturing
3.1 CAD Designs to Machine Scripts.
Similar to traditional FDM 3D printers, manufacturing on the NeXus system requires a series of G-code scripts that inform the system of where, how, and what to do. Generation of these machine scripts is processed through a standardized pipeline to go from industry-standard PCB design suits to the custom G-codes used by the NeXus.
The design of the demonstrator circuit was performed in kicad, an open-source PCB design software. A PCB design suite was used for its integrated tools for SMD placement and trace routing. The necessary output of this program is a “plot” of copper traces as a .dxf file; this feature is common to many PCB design suites. This will provide a series of dimensioned shapes that represent the outline of the conductive components. This file can now be opened in solidworks or any standard CAD suite capable of drawing polylines and tangent arcs. Within solidworks, the .dxf from kicad is treated as a reference for designing the tool paths, the exact lines the printer will produce. When possible, preference should be given to designing exclusively in polylines. If arcs are required an alternative processing and printing method is used, this process is described in Sec. 3.1.1. The tool paths are printed at a fixed line width set by the tool, approximately 50 m for AJP silver conductive traces and 1 mm for DOD inkjet insulators. Lower trace resistance, for signal or voltage-carrying traces, can be achieved by printing tool paths in parallel, overlapping them slightly to produce effectively thicker traces. For the demonstrator design, the traces were printed at three lines thick, 50 m apart, to balance low trace resistance with print time. Similarly, pads should be printed as a dense series of parallel lines that will combine to create an SMD component pad, interfacing pad, or test pad. If the design requires multiple layers of conductor, then the DOD inkjet printer is used to print the insulating material as a bridge. Each printing operation should be constructed on a separate layer or “drawing” in solidworks.
With the tool paths developed, each layer is exported to a .dxf file. dxf2gcode, an open-source program, converts the DXF files to G-code. In G-code, for 3D printers or machining operations, the output is a deposition of material or a rate of rotation for a tool head controlled by the “feed” parameter F in a G1 or G2 motion command. On the NeXus, the printing tools are actuated with a binary status command, M300 S30.00 (pen up) or M300 S50.00 (pen down). These commands are custom-programmed into the configuration of the dxf2gcode program. A standard printing operation would consist of a “pen down” command followed by a series of G1 commands that define the desired printing geometry.
3.1.1 Automated DXF to Line–Arc.
The demonstrator PCB integrates a metal strain gauge sensor. The functionality of this sensor is dependent on a consistent line width throughout the design. These design constraints were formed and evaluated by the designer D. Wei [21]. The required width-sensitive designs were realized using the line–arc trajectory function from the Newport XPS-D controller's API [26]. This function moves the stage on a continuous curve at a constant speed. This ensures the printed shape has a uniform line width. As the trajectory file format is unique to the Newport controller, the generation of these files is performed by a custom python program. This program intakes a DXF file with a continuous line, consisting of only polylines and tangent arcs, beginning at the origin of the file (0,0) and translates it into a series of line and arc commands that the Newport controller can interpret.
The line–arc scripting language combines lines defined by their end position in absolute coordinate space with tangential arcs defined using relative polar coordinates of radius and angle. Angle is positive or negative to indicate the direction around the radius, counterclockwise and clockwise, respectively. DXF files define lines by their absolute beginning and endpoints, and arcs by their absolute beginning and end points as well as their radius and positive or negative angle.
DXF files are a general purpose CAD data file for sharing drawing data universally across applications, unlike G-code or trajectory files which are machine scripts meant to be executed sequentially. This means DXF files do not define an order for features and store feature data abstractly. Utilizing the python package “ezdxf,” the line and arc data could be queried to produce two arrays containing line and arc data classes. As DXF files are nonsequential, the program must search for the features in sequence as the output file is produced. A “start” variable, initially (0,0), is updated with the opposite end of the found line to iterate through the connected structure. The DXF defines start and “end” points for the line and arc features. However, because these features are stored out of order, both must be searched through to build the entire structure. When a line or arc is found, its defining features are formatted and written to the output file. Unique functions are defined based on whether the feature was found by its start or end point. For lines, this defines the opposite point to be written to the output file and what to update the start variable to; for arcs, this affects if the angle is written how it is stored or inversely. If the line is approached from the end, the angle is opposite of the stored value.
Printing of these trajectory files is performed using an M98 G-code command written into a G-code printing file. M98 is a subroutine command in a standard G-code application. Because trajectory files use a relative coordinate frame, the G-code must contain the desired starting position. When printing with a trajectory file, the stage is moved to the desired starting position relative to the substrate; the printing head can then be engaged, command pen down, and then the M98 trajectory command is called in place of the typical series of G1 commands. This G-code file could be written exclusively to run the trajectory file or a series of trajectory files because each trajectory file must be a continuous line, or the commands could be integrated into a G-code script generated by dxf2gcode. This would nest the trajectory printed design into a larger G-code structure. This functionality is used in the demonstrator PCB.
To test and validate these custom files, a visualizer was developed which follows the command scripts, G-code and trajectory, line by line printing the features as the system will. This allows for the final G-code script to be visualized with the integrated trajectory commands and to validate that there were no issues in the design or conversion process.
Preference is given to G-code printing as this is a faster and more efficient process. G-code printing is performed at 10 mm/s, while trajectory printing is performed at 1 mm/s. This makes G-code designs ten times faster to print. Trajectory printing should be restricted exclusively to designs where single-line trace width is a critical constraint.
A flowchart describing the conversion process from solidworks to printing files is shown in Fig. 6.
Currently, each printing procedure is performed by the operator in sequence. This allows for curing procedures or surface treatments outside of the system to be performed.
3.2 Additive Manufacturing Tools.
Printing, curing, and inspection are performed on the NeXus system's precision manufacturing line. This subsystem is shown in Fig. 7. The inspection station, with custom microscope, is used for calibration of the 6DOF positioner relative to printing tools, substrate features, and inspection of the quality of printed or IPL sintered lines and pads. The kinematics and calibration of this subsystem are explored in depth in the work by Wei et al. [21].

(a) NeXus system precision manufacturing line and (b)NeXus pick and place station with Denso SCARA arm
3.2.1 Conductive Trace Printing and Curing.
The conductive traces for the FPC are realized by printing with a silver nanoparticle ink (Novacentrix JS-A426) using an Optomec Decathlon Aerosol Ink Jet printer (AJP), shown in box 1 of Fig. 7. The printing process of the silver lines on the Kapton surface has been evaluated extensively in previous work by Ratnayake et al. [19]. For that purpose, the Optomec AJP tool in combination with the 6DOF positioner allows the realization of the designs based on the G-code or trajectory script via the NeXus PXI control platform. Printing parameters have been calibrated to fabricate lines with a width of approximately 50 m. Patterns based on trajectory files are printed at 1.5 mm/s with a constant velocity resulting in a uniform line width. G-code based designs are printed at 10 mm/s with an acceleration of 50 mm/s2.
Silver-printed lines were cured using two methods: IPL sintering and oven sintering. IPL sintering is achieved with the help of the Xenon Corp S2210 system by repeatedly flashing a capacitively charged Xenon, four arc, discharge lamps (LH-150) onto the substrate. The IPL system is shown in Fig. 7. IPL sintering is a developing sintering technology that greatly improves curing time and can be performed without removing the sample from the NeXus system. The user can control the intensity of the light (voltage), frequency of the pulses, length of the pulse, and number of pulses. The recipe used for curing is shown in Table 1.
Process parameters of Xenon IPL system for curing of Novacentrix JS-A426 silver nanoparticle ink on Kapton substrate
Duration (s) | Pulses | Delay (ms) | Voltage (V) |
---|---|---|---|
350 | 10 | 100 | 1500 |
800 | 10 | 300 | 1500 |
Duration (s) | Pulses | Delay (ms) | Voltage (V) |
---|---|---|---|
350 | 10 | 100 | 1500 |
800 | 10 | 300 | 1500 |
Oven sintering is achieved by curing the printed sample in an oven for 20 h at 200 °C. These curing parameters were evaluated in previous work by Ratnayake et al. [19].
3.2.2 Insulator Printing and Curing.
Insulating material is printed by the Nordson EFD PicoPlse®, a piezoelectric drop-on-demand ink-jet printer, shown in Fig. 7. The PicoPlse® is used for more viscous materials or application of larger quantities of materials than would be efficient with the Optomec. Two different materials were used as potential insulators, Bondic® UV Adhesive and Total Boat® UV Clear Cure Epoxy. Both materials were printed using the 100 m nozzle on the PicoPlse® resulting in a line with approximately 0.6–1+ mm in width depending on surface treatment and printing settings. The settings used to print both materials are documented in Table 2.
PicoPlse® printing parameters for Bondic® and Total Boat® UV Curing Polymers
Material | Stroke (%) | Pressure (psi) | Temperature (°C) | Cycle (ms) | Plasma |
---|---|---|---|---|---|
Bondic® | 90 | 60 | 45 | 85 | 5 min-high |
Total Boat® | 90 | 40 | 32 | 85 | N/A |
Material | Stroke (%) | Pressure (psi) | Temperature (°C) | Cycle (ms) | Plasma |
---|---|---|---|---|---|
Bondic® | 90 | 60 | 45 | 85 | 5 min-high |
Total Boat® | 90 | 40 | 32 | 85 | N/A |
Prior studies have used Taguchi design of experiment methods to characterize the deposition of the Bondic® adhesive [18]. These principles have been utilized to produce usable printing settings with the Total Boat® adhesive. The plasma treatment referenced in the table is a basic oxygen plasma cleaning treatment that improves the surface-wetting effect.
Due to the viscosity of both materials, approximately 5000 centipoise (cP) and 1000 cP for Bondic® and Total Boat®, respectively, the process of loading the material into the PicoPlse® compatible syringes introduces air. This is counteracted by using a vacuum chamber to perform degassing overnight before the syringe is used.
Curing of the materials was done using a UV lamp for a minimum of 5 min.
3.3 Assembly Tools and Methods.
Assembly of SMD components is automated through the application of the 4DOF Selective Compliance Articulated Robot Arm (SCARA) DENSO (HM-40A04) industrial robotic arm. This arm is equipped with a PnP tool and a Nordson EFD auger valve. These tools allow it to perform PnP tasks as well as solder paste deposition. Application of this arm for PCB assembly has previously been evaluated by Nimon et al. [27]. Sequentially, the tools are used to deposit solder onto the desired pads around the PCB, then the SMD components can be placed appropriately. The PCB assembly station within the NeXus is equipped with an IR heater which is then able to reflow the solder paste finalizing the assembly of the PCB.
In the study by Nimon et al. [27], authors validated the accuracy of the SCARA for PnP tasks in an open loop configuration, as compared with the tolerances expected of a traditional Cartesian PnP system. This meant the tasks could be completed within an expected 200-μm tolerance without the application of visual servoing dead reckoning methods. This effectively reduced the subsystem's complexity and assembly time. The same open loop method was deployed for the assembly of the demonstrator circuit as well as for control of solder deposition.
Solder deposition was characterized utilizing recommendations from the Nordson manuals for the EFD auger valve and valvemate controller [28,29]. This defined the back pressure applied to the syringe and the voltage applied to control the auger motor. Three different nozzle sizes were tested on the auger valve: 0.32 in, 0.16 in, and 0.08 in diameters. The final parameter explored defining the characterization of the solder deposition was deposition time which was tested from 1/8th of a second up to 2 min.
4 Results
Each functional component of the FPC demonstrator is evaluated individually and then combined as a whole after the final assembly. The following experimental tests have been performed:
comparison of curing methods
evaluation of printed insulating materials
evaluation of printed silver pad soldering
evaluation of operating amplifier functionality in FPC
4.1 Comparison of Sintering Methods.
After printing and sintering the first layer, each circuit's resistance was evaluated using 11 test points across the FPC. A multimeter was used to test resistance and continuity. Three of the test points assess the success of the strain gauge sensor, and the remaining points evaluate the resistance of signal or voltage-carrying traces which will affect the amplifier's performance. The major difference between these two trace types is that the strain gauge diaphragm part is printed with single 50 m wide lines, while the remaining traces have a width of around 150 m (three combined AJP lines). The 11 test traces are numbered and highlighted in Fig. 8.

The 11 traces and testing points used for evaluating continuity and sintering success of demonstrator FPC
The two sintering methods, oven and IPL, were evaluated based on the number of samples with continuous lines and adequate resistance (rated as success) for the strain gauge and circuit sections. Sensor success means that the three tests of the sensor showed it had continuity and was functional, while circuit success implies all traces achieved continuity, had a sufficiently low resistance, and the sensor was functional. If the criteria were met, the circuit moved forward to the next manufacturing step. Results showed that the oven sintering method is more reliable (Table 3), consistent with our observations indicating that IPL thermal treatment is more effective for shorter printed sections, <10 mm. For the larger areas of Kapton substrate with printed AJP lines of lengths more than 10 mm, oven sintering is more effective. The average resistance per mm for the eight signal traces and the standard deviation of /mm is shown in Table 4. These data do not account for traces that did not achieve continuity, thus showing some bias toward IPL sintering. The sensors that produced full functionality across all testing points were transferred to the next step in the process flow.
4.2 Evaluation of Printed Insulation Materials.
The Bondic® adhesive consistently shrunk during silver sintering in the oven, changing its morphology—shape and dimension. This resulted in the discontinuity of printed conducting lines (Fig. 9(a)). This change in morphology is apparent from the profilometry data shown in Fig. 9(b) depicting the profile of the Bondic® adhesive before and after the standard 20 h, 200 °C, curing period for AJP silver lines.

(a) Microscopy image of defect of AJP line on Bondic® adhesive and (b) profilometry comparison of Bondic® before and after thermal curing
The Total Boat® resin has a lower viscosity and a stronger wetting effect on the Kapton substrate. When optimized for printing, Total Boat® resin produced a smaller line with a lower contact angle without plasma treatment. Total Boat also produced greater printing consistency as a benefit of the simpler printing process. Profilometry characterization reflects these observations as shown in Figs. 10(a) and 10(b). However, the material experienced microbubbling (trapped air bubbles in the solidified UV adhesive), which after oven curing distorted the AJP printed lines seen in Figs. 10(c) and 10(d).

(a) Profilometry comparison of Total Boat resin without plasma-treated substrate and Bondic with plasma-treated substrate, (b) repeatability demonstration of Total Boat® resin printed with PicoPlse, (c) and (d) Microscopy images at 90 deg and 45 deg of Total Boat® resin containing microbubbling features which have affected AJP lines

(a) Profilometry comparison of Total Boat resin without plasma-treated substrate and Bondic with plasma-treated substrate, (b) repeatability demonstration of Total Boat® resin printed with PicoPlse, (c) and (d) Microscopy images at 90 deg and 45 deg of Total Boat® resin containing microbubbling features which have affected AJP lines
4.3 Evaluation of Printed Silver Pad Soldering.
Once the insulating bridge with the printed conducting traces was formed, our device was ready to assemble the amplifying circuit using SMD components and the low-temperature solder paste.
Low-temperature solder paste was evaluated with respect to the resistance change tested across a 100 resistor soldered across two printed silver pads 1.5 mm × 1.5 mm and 1.2 mm apart. The exact resistance of the SMD component was measured prior to soldering and then again after from pad to pad across the component to produce a change in resistance measurement. This method allowed us to determine average resistance and detect imperfections, such as leaching, in the solder paste/AJP Ag pad interface. Table 5 describes the average resistance change and standard deviation for the two sintering methods applied to the testing pads. These results indicate a sufficiently low resistance shift (R) from the application of the solder paste for oven curing. In consistency with conductivity tests for the printed individual lines, IPL sintering of pads produces higher interface resistance after soldering, possibly through a combination of two effects—a lower surface adhesion of the solder paste (leaching) and lower and nonuniform conductivity across the printed pad.
We have conducted deposition characterization of the solder paste using an auger valve from EFD determining an optimized set of parameters leading to reliable and consistent printing results. However, with the current configuration, depositions at the extreme ends of 0.125 s and 2 min yielded deposits of the size in the range of 800 m and 4 mm.
4.4 Evaluation of Operating Amplifier Functionality for Tested Flexible Printed Circuit Board.
Final assembly and testing of the FPC were performed to evaluate functionality of all elements of the integrated device. Fully assembled test circuits are shown in Fig. 11. The FPC on the left is constructed with all the designed SMD components, and the FPC on the right is constructed with a jumper wire from the sensor signal output to the output pin of the FPC in place of the INA 821 SMD.

Final FPC assembly with all SMD components (left) andFPC assembled with a jumper in place of the amplifier (right)
Each strain gauge sensor was loaded with a 307-g mass for 5 s and then removed for 5 s. This was cycled ten times over 2 min for both sensors. The results of these experiments are shown in Fig. 12.

Load cycling of the strain gauge with and without amplifier collected by Hioki LR8431-20 Memory HiLogger
The signal without amplification shows a total range of 0.52 mV including noise. In contrast, the circuit with the integrated amplifier has distinct peaks and valleys in line with the applied load that insight a change of more than 2 V.
5 Discussion
5.1 Comparison of Sintering Methods.
With the current state of optimization, IPL sintering produces lower surface adhesion and more defects compared to oven sintering. Surface adhesion affects the silver's durability and resistance to leaching during soldering. A lower continuity rate greatly reduces the desired output success rate; only circuits with 100% continuity can be passed on to further processes. The observations proved the current state of IPL curing to be unfit at this time for first-layer sintering.
5.2 Evaluation of Printed Insulating Materials.
Plasma treatment was performed after sintering the first conducting layer. While there is potential for surface oxidation due to this process, the additional sintering process for the second layer of the conductor significantly reduces the resistance below initially measured values such that any increase in resistance due to oxidation is negated. Samples tested for resistance first after oven treatment and again after full multilayer completion, assembly, and finally left to oxidize for 131 days, maintained a lower resistance per millimeter by 24%.
A key advantage of the Total Boat resin was its ability to produce quality lines without the need for plasma treatment. Plasma treatment required removal from the system to manually process the sample through an oxygen plasma system. Once this was completed, the plasma-treated surface experienced a short period of heightened wetting effect making this a time-sensitive procedure. Once replaced within the NeXus system, recalibration is necessary to locate the sample within the robotic system of coordinates for printing and assembly. This process is currently performed by the operator by visually locating a fiducial marking with the optical inspection station. Once aligned, the printing process can commence. Alignment and calibration of the substrate and printing tools is performed to a precision of 5 m; however, because this is a user-performed task, an error of 10 could be expected for each calibration procedure. By eliminating the need for plasma treatment, this reduced the propagated error of repeated alignment of the substrate. However, the design includes a tolerance of 150 m for both x and y dimensions for printing both the insulating layer and final AJP layer over the insulator. Across 28 samples which made it to insulator printing, no defects were generated due to alignment issues.
Intense pulsed light sintering was also applied to the insulating bridge interconnects. In this application, the traces were doubled or even tripled, printing over the same path two to three times. This increases the volume of conductive material which positively impacts the reliability of electrical interconnects. In addition, since IPL is a thermal treatment that produces significantly higher temperatures at the surface of the material [24], many of the issues with the insulating structure that appear during oven sintering are eliminated. With this in mind, Total Boat® was chosen for its higher reliability and better surface wetting. The increased surface wetting removed the need for plasma treatment before printing the insulating layer, simplifying the process flow.
Four sensors were tested with oven sintering for their first layer and IPL for their second. All four achieved conductivity, averaging 0.233 /mm. Images of one of these samples are shown in Fig. 13. The improvement in insulator quality from utilizing IPL is visually apparent from the clarity of the Total Boat® resin. This is highlighted well by observation of the optical distortion, shown in Figs. 13(b) and 13(c), where the resin is acting like a lens to the features below it.

Top down (a), 45 deg (b), and horizontal view of Total Boat® UV Resin Bridging with IPL sintered conductor
5.3 Evaluation of Low-Temperature Solder and Circuit Functionality.
The results showing the difference in sensor response with and without the integrated amplifier clearly show that the designed circuit is functional and can significantly enhance the signal from the strain gauge with an acceptable signal-to-noise ratio. Furthermore, it can be concluded that the use of the low-cost solder paste with AJP pads on Kapton does not impair the operating amplifier's functionality, considering possible disruptive effects from the residual resistance or capacitance at the interfaces of the printed silver pad, solder, and SMD electrodes.
6 Conclusion
In this work, we have developed a process for manufacturing a multilayered FPC on the NeXus system by evaluating and developing the necessary interfacing technologies including:
pipeline from PCB design suit to machine-specific printing files
integration of novel, high precision, printed smart features (strain gauge) with efficient low resistance traces
sintering of novel conductive inks in complex geometries on Kapton substrates
compatibility of printed insulating materials to produce 3D geometries with AJP
soldering of SMD components to AJP conductive pads
While the process is not currently fully automated from start to finish, the demonstrator shows functionality in amplifying the signal of the sensor by the set gain of 1830 with a quality signal. Furthermore, the successful fabrication of the sensor, with the realization of all the processes in one location, demonstrates the rapid prototyping capability of the NeXus system.
Critical steps for future work to reach full automation include scripted automation of PnP and solder deposition, optimization of IPL sintering, and further software development to reduce complexity for the end user. Relative coordinates for all of the components can be retrieved from the CAD files and could be applied to a scripted automation process. Optimization of the IPL process would improve the processing time as well as reduce the amount of mandatory user interaction improving the reliability and accuracy of the system. Further software development would serve to reduce complexity for the end user and improve the process flow at the front end. A custom “slicer” to go from a CAD file standard in the industry, such as Gerber, to all the machine-ready scripts would greatly reduce the complexity for the user.
Funding Data
National Science Foundation EPSCOR RII-Track-1 (Grant No. 1849213; Funder ID: 10.13039/100000001).
Data Availability Statement
The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.