1 Introduction

3D printing has become a widespread technique. Of the various printing technologies, fused filament fabrication (FFF) is certainly the one with the greatest expressiveness today. This results not only from the easy access to FFF equipment but also from its fastest pace of technical evolution, both at the equipment and the material levels.

With the possibility of printing with composite filaments, especially polymeric filaments with functional fillers, it is now possible to manufacture technical structures. If one of the main advantages of 3D printing was the almost complete geometric freedom of the components that can be produced, by enabling the printing with composite filaments one can produce composite parts with geometries impossible to obtain by conventional techniques. Considering the possibility to print multi-material parts with the aid of multi-extruder printers (hence parts composed of several distinct material filaments), then one opens an opportunity to produce hybrid and structured composite parts.

Several works described the development of successful hybrid multi-material FFF parts, but nothing is said about the selection of the pairs of materials used [1,2,3]. However, the performance of hybrid structures depends on the symbiotic interaction between their different materials, being especially critical at their interface. Although in conventional manufactured hybrid structures the interface is clear, on FFF printed parts it is more complex as there are multiple interfaces. Inherent to the printing technique and, therefore, unavoidable, there are multiple welds formed at: (i) the interface between adjacent deposited filaments of the same layer; and at (ii) the interface formed between different layers. From a mechanical perspective, interfaces are discontinuities and thus potential weak points on a given part. There are several studies on the effect of the printing conditions [4, 5], the orientation of the part [6], material inherent viscosity [7], and post-printing treatments [8], aiming at diminishing the detrimental effect of these interfaces.

The problem gains more relevance in multi-material prints since beyond those inherent interfaces between the same material, there are also zones where different materials interface and should weld together. In this case, a low chemical affinity between the two materials may hinder the required molecular diffusion at the interface. In previous research [9], it was shown that the lack of chemical affinity between the pair of materials used to print a multi-material part may compromise its mechanical performance. In addition, it was shown that designing a physical/mechanical interlocking at the two different materials’ interface may enhance part performance [10, 11], circumventing their eventual poor affinity.

Presently, there is a huge variety of filament materials for 3D printing that includes the majority of all the commercial thermoplastics (from commodity to engineering materials) and carbon-filled, wood-filled, and metallic-filled thermoplastic matrix composites, to mention just a few. In addition, while it becomes possible to find technical data sheets of these materials comprising the printing conditions as well as some data about the resulting part performance, there are no data about the affinity between different materials, which is essential for multi-material 3D printing. The most evident solution to support the selection of adequate pairs is to print hybrid tensile test specimens with the two materials to assess their mechanical performance. This was done in [9], but such a process is time- and material-consuming. Thus, an expedited process to test the affinity between different filament materials, representative of the FFF process, is needed.

In this work, an expedited method based on traditional plastics welding, targeting a fast selection of pairs of filament materials with adequate chemical affinity, is proposed and explored. The proposed technique attempted to replicate out of the 3D printer what occurs at the FFF extrusion site (the extrusion of a fused filament on top or along a previous extruded trace of a different material). If the proposed method is valid, one would get an indicator foreseeing the material’s affinity faster and with less filament-waste.

2 Filament hot plate welding

2.1 Welding

In the context of polymer processing, the word welding applies to: (i) the joining of solid components, which joining surfaces have to be previously melted [12]; (ii) the joining/re-joining of different molten flow fronts, resulting in the so-called weld-lines—see, for example, [13] for injection molding, and [14] for profile extrusion; or (iii) the sealing of a heated polymer preform, such as the parison in extrusion-based blow-molding technology, resulting in the so-called pinch-off zone of hollow containers [15]. In any case, a weld zone is always critical since it tends to be less resistant than the original material due to insufficient molecular diffusion and entanglement at the interface [12].

Welding of components is one of the most popular techniques used to join thermoplastic parts, competing with mechanical joining and adhesive bonding. There are several industrially relevant welding techniques, which are categorized by the heating method used to molten the surfaces to be joined: ultrasounds, laser, radiofrequency, hot plate, and hot gas, among others.

The phenomena occurring in the deposited traces welding, in FFF, are similar to those that take place in the hot plate welding of plastics. Despite the similarities between these two processes, the thermomechanical conditions in which they occur are different and more favorable in traditional welding, namely higher joining temperatures and contact pressures. In hot plate welding, the two surfaces to be joined are previously heated by direct contact with a hot plate. After the time needed to molten the required length of material, the surfaces are retracted, the hot plate is removed, and the two hot surfaces are put in contact under a preset and controlled pressure. In 3D printing, an extruded trace is deposited over, or side by side, another material trace previously deposited. As a consequence, the temperature of the trace being deposited is always higher and controllable when compared to the already deposited trace that can be already at a much lower temperature, depending on the printing sequence and printing environment temperature [16,17,18]. Another drawback of the 3D printing welding process is that the pressure applied by the nozzle over the extruded filament, when existing, is instantaneous.

Despite the differences pointed out, an adequate material pair for hot plate welding is expected to be also a good candidate pair for multi-material printing. The method suggested in this work is based on this premise.

2.2 Methodology overview

A typical hot plate welding equipment encompasses the following main components/systems, as illustrated in Fig. 1:

  1. (i)

    a controlled temperature hot tool—a flat metallic plate coated with a low friction coefficient material (usually, a PTFE film) to avoid plastics parts to stick;

  2. (ii)

    a locking system that holds the parts together during the weld cycle, and guarantees their relative alignment during the joining stage;

  3. (iii)

    a pneumatic system that performs the required displacements (highlighted with grey arrows in Fig. 1);

  4. (iv)

    two different pairs of rigid mechanical stops that control the out flow of molten polymer during the heating stage (heating stops) and joining stage (joining stops);

  5. (v)

    a control system that handles the welding process (duration of heating, joining, and contact stages).

Fig. 1
figure 1

Schematic view of hot plate welding equipment and some of the process variables

Full size image

The cycle sequence and main variables of the hot plate welding process [12] are listed in Table 1.

Table 1 Main variables of the hot plate welding process (adapted from [12]), according to the cycle sequence
Full size table

The method proposed to select pairs of different materials with potential for the FFF multi-material technique consists in assessing their adequacy for hot plate welding, as illustrated in Fig. 2. In summary, the comparison between the mechanical performance of pristine filaments of each material and that of the corresponding hybrid welded filaments will determine the affinity between the materials and, therefore, their potential adequacy, as a pair, for multi-material 3D printing.

Fig. 2
figure 2

Methodology proposed for the selection of pairs of materials for the FFF multi-material technique

Full size image

This method is similar to the one used in [9], differing in the use of filaments (monolithic and hybrid welded) instead of printed tensile test specimens. This is the main advantage of the proposed method that saves time and the required amount of raw materials.

3 Materials and methods

3.1 Strategy

This paper aims at testing the hypothesis of using hot plate welding equipment as an expedited approach to assessing the affinity between different filament materials for multi-material 3D printing. To test this thesis, the plan of experiments was built in such a way that it will enable one to infer about:

  1. (i)

    the influence of the welding technique—the use of a process per se has a given influence on the part’s performance. Thus, getting a magnitude order of such influence is critical to assess the feasibility of the process as an expedited solution.

  2. (ii)

    the materials’ affinity—the chemical affinity between different materials is the critical element in this study. The large variety of available materials for FFF 3D printers hinders the process of understanding if a given material is compatible with another one when designing a multi-material part for 3D printing. Moreover, concluding about filaments’ compatibility based on their chemical structure is a complex subject, not accessible to the standard user. Thus, assessing the chemical affinity without the need for chemical details is required.

  3. (iii)

    how do the results correlate to the 3D-printed part performance—considering that the present paper proposes the use of an alternative technique to confirm materials filament affinity, it is important to ensure that the results obtained from this alternative process, plastics welding, are representative of the ones obtained from an FFF 3D multi-material printing.

To ensure this 3-level validation, the following strategy was defined:

  1. (i)

    a neat and monolithic filament will be compared with a filament welded together with itself (Fig. 3a). In this case, all filaments were taken from the same spools. This enables the assessment of the welding process’s influence.

  2. (ii)

    multi-material filaments will be produced by welding two different filament materials together (Fig. 3b). This introduces the chemical affinity question in the study.

  3. (iii)

    multi-material 3D-printed test specimens will be produced with the same pair of materials as multi-material filaments to evaluate how the FFF results correlate with the welding ones (Fig. 3c). These samples were printed in a dual-extruder printer.

Fig. 3
figure 3

Different tensile test specimens considered in the study: a–b filament specimens; c 3D-printed test specimens based on DIN 53504-S2a. A and B represent different materials

Full size image

The mechanical performance of the samples was characterized through tensile tests. Figure 3 illustrates the different test specimens used in the study.

One critical detail of the multi-material specimens (Fig. 3c), is the printing sequence and the possible impact that such sequence might have on the part performance and, therefore, on the results. By default, the printing sequence defined when printing the specimen with materials A and B is to minimize the number of extruders’ changes. This results in a specific sequence scheme, namely: (1) on layer i, the system will for example print section A first and then section B; (2) when changing to layer i + 1, it will print B and then A; then repeating this sequence. This results in the following sequence [A–B]–[B–A]–[A–B]–[B–A] if one considers a 4-layer part. Such sequence enables the homogenization of the sequence impact, especially when the printing temperatures are different for materials A and B. A different printing sequence as always A first and then B would result in a temperature imbalance during the deposition of the different materials’ traces at the interface of the two materials.

3.2 Hot plate welding equipment

The equipment used to weld hybrid filaments is an in-house prototype used in [19, 20], which can be generically described by the schematic representation shown in Fig. 1. The only modification done to the original equipment was the adaptation of the specimens’ holders (locking system in Fig. 1), which were originally designed for tensile test bars. These new supports should lock the half filaments and guarantee their relative alignment during the joining stage. In fact, and due to their reduced cross-section area, any misalignment would have a hugely detrimental effect on the weld performance.

3.3 Materials

Having in mind the nature of multi-material printed parts, it was assumed as relevant to assess how different types of materials combine. For this sake, two groups of materials were considered: rigid and flexible. The rigid materials group included two engineering thermoplastics, namely Polycarbonate (PC, ref. PC-PRM-175–750 from ORBI-TECH) and Polyamide (PA, Nylon 230 from taulman3D), while the flexible filament group included three thermoplastic elastomers, NinjaFlex (from Ninjatek), FilaFlex (from Recreus) and TPC 45 (from MCPP-3DP). All material filaments had a 1.75 mm diameter, and the main characteristics of these filaments are listed in Table 2.

Table 2 Filament materials’ characteristics
Full size table

3.4 Samples’ preparation

For the filament hybrid samples, the filaments were cut directly from the spool and welded according to the following sequence:

  1. (i)

    10 filament samples, 100 mm long each, were cut from each material spool;

  2. (ii)

    the samples were placed side by side, in pairs, and their ends were trimmed with a blade to ensure a flat and similar contact area with the hot plate;

  3. (iii)

    the filaments were placed on the samples’ holders on the welding machine. These holders were then brought together to check the filament alignment;

  4. (iv)

    with the alignment guaranteed, the welding process was performed.

A first decision concerning the welding temperatures and heating times to be used with each filament material and with each pair of different filament materials was required. Since there are differences in the printing temperature range of the different materials used (see Table 2), for each pair, the highest printing temperature of each material was selected for the welding process. In this way, ensuring the proper temperature and heating time for the material requiring the highest temperature will assure that the same will happen with the other unless thermal degradation occurs. In addition to the chemical compatibility between the materials, it is required to guarantee that the amount of molten material at both sides of the welding zone is sufficient to enable molecular diffusion at the interface and obtain a good quality weld.

The mechanical heating stop used in the welding process was retracted 0.5 mm in relation to the sample surface (to guarantee a clean contact surface at welding after heating), and the joining stop was 0.5 mm retracted in relation to the previous, to enable a displacement/compression during the joining stage. Thus, the heating temperature time to be used should guarantee the melting of, at least, 1 mm length of filament at each side of the hot plate. The process parameters were defined through the transient conduction heat transfer equation

$$\frac{{T - T_{S} }}{{T_{i} - T_{S} }} = {\text{erf}}\frac{{L_{0} }}{{2\sqrt {\alpha T} }}$$
(1)

where T is the material melting temperature (for the semi-crystalline polymers) or the Vicat softening point (for the amorphous ones), Ts is the constant boundary condition temperature, i.e., the temperature of the hot plate or welding temperature (here assumed as the printing temperature of the material), Ti is the initial filament temperature (room temperature, 20 ºC), α is the material thermal diffusivity, L0 is the length of molten material, erf is the error function and t is the heating time.

Based on the typical properties of the materials used in this study (melting/softening temperature and thermal diffusivity), the molten length resulting from the welding temperature and heating time used (20 s) was found to be at least 1.5 mm, i.e., higher than the required 1 mm.

The FFF specimens were printed in an xpim pom printer, equipped with 4 direct drive extruders, model E3D Titan Aero, with the conditions listed in Table 3.

Table 3 3D printing parameters
Full size table

3.5 Samples’ characterization

Tensile tests were performed using an INSTRON 5969 universal mechanical testing machine. The tensile testing speed was set at 50 mm/s, and the distance between jaws was 110 mm and 20 mm for the filament and the 3D FFF test specimens, respectively. For each condition, 5 samples were tested.

4 Results and discussion

In Fig. 4, one can see the maximum force that monolithic and welded filaments of the same material can withstand. The comparison is made to assess the weldability of a material with itself. PC and PA materials show a significant loss of maximum strength value when welded with themselves, while the flexible materials almost kept their performance.

Fig. 4
figure 4

Maximum force for monolithic and (same material) welded filaments

Full size image

In addition, PC and PA filaments are the ones that show the highest variability scores among the different materials. Such variability might be a consequence of the highest temperatures (270 ºC and 250 ºC, respectively) used for welding. The hot plate consists of a 150 mm diameter metal disk, which at this range of temperatures has more difficulty in maintaining the set temperature.

The rupture of the rigid materials occurred early in the tensile test, i.e., at a lower deformation and lower corresponding force.

One possible explanation for the visible distinct values corresponding to rigid and flexible materials may lie in the fact that flexible materials show ductile behavior, being able to hold large deformations when subjected to axial stress. In general, there is a decrease in the values obtained in the welded samples when compared to their respective filaments, as expected. Supporting this decrease is the fact that the welding zone is a potential zone of greatest weakness due to the eventual molecular structure discontinuity (originated by poor molecular diffusion), the existence of voids, and/or impurities.

Conversely, to most materials, welded FilaFlex has a slightly higher value than its monolithic filament, possibly due to the increased resistant cross-section area generated by the dewlap formed during the joining process. Lastly, the substantial loss of strength of rigid materials (PC and PA) compared to flexible materials is noticeable.

Shown in Fig. 5 is a comparison of Young’s modulus of these same samples. Since Young’s modulus is an intrinsic characteristic of each material, it would be expected, among its peers, to obtain values within the same range. In the particular case of flexible materials, the values are within the same range, except for the welded TPC 45, which shows a value apparently greater than its monolithic counterpart. As for rigid materials, the welded samples show a range of values almost twice their original filaments, which may come from a larger area in the weld zone.

Fig. 5
figure 5

Young’s modulus of monolithic and (same material) welded filaments

Full size image

To select the best hybrid materials pair, the maximum force obtained with the single materials welded with themselves (AA and BB) samples and the corresponding hybrid pair (A welded to B) is shown in Fig. 6. During the tensile tests of the three hybrid pairs, the strain over time was visible essentially in the flexible half of the specimen, and the rupture occurred at the welding site. In case of good affinity between the materials, the maximum force values obtained with hybrid pairs would be expected to be close to the lowest value of their constituent materials.

Fig. 6
figure 6

Maximum force of (same material) welded filaments and pairs of different materials welded filaments

Full size image

This expectation is based on the fact that at the materials interface the weaker material will limit the pair performance. Analyzing the collected values, it is possible to confirm this hypothesis for the three hybrid pairs tested.

In Fig. 7, the welding between the different hybrid pairs is also evaluated through their Young’s modulus. In all cases, for the hybrid pair, this property value lies between the values obtained for each of its constituents but is closer to the lowest value. The apparent better performance of the hybrid pair when compared to the weakest material is most probably related to the fact that the deformation of the sample is not uniform. In fact, and as already referred, the imposed deformation is occurring only on the flexible material, and thus the force considered in the modulus evaluation is higher (since it corresponds to a real higher deformation of the flexible material in the hybrid samples). Therefore, the values collected are considered good indicators of effective welding between the materials pairs, since if they were lower than the lowest value and/or close to zero they would indicate a poor affinity or even incompatibility.

Fig. 7
figure 7

Young’s modulus of (same material) welded filaments and pairs of different materials welded filaments

Full size image

In terms of the comparison between the welded and printed samples, Fig. 8 shows the tensile strength corresponding to each hybrid sample obtained by the two processes: filament welding and 3D printing.

Fig. 8
figure 8

Tensile strength of different hybrid samples obtained by welding and 3D printing processes

Full size image

As can be seen, the hybrid filament samples obtained by the welding process have higher tensile strength than their printed counterparts. These differences are certainly associated with the previously mentioned dissimilarities in the thermomechanical conditions occurring in each process. In the welding process, the two-component materials are heated and melted simultaneously, and pressure is applied during the joining stage, facilitating molecular diffusion at their interface. On the other hand, and in the printing process, when the active extrusion head deposits one material on the printing bed or over the previous layer, the previously deposited material is already at a lower temperature. Furthermore, there is no pressure applied to help promote the joining. In addition, the filaments, which are produced by extrusion, are more homogeneous and more compact than the printed samples, which may have voids resulting from printing speed, temperature inhomogeneity, and cooling, thus promoting imperfections in the printed structure, especially critical at the interface between the different materials [21].

In Fig. 9, the tensile stress of the same samples (NinjaFlex), broken down by form factors (filament and printed specimens), is shown. In both cases, there is a loss of maximum stress from the first column to the second one, which is expected and supported by the fact that the specimens represented by the second column are the result of a non-continuous process, involving the joining of two surfaces, which results in a less homogeneous and weak interface zone.

Fig. 9
figure 9

Tensile strength of the four types of NinjaFlex test specimens obtained for monolithic and welded filaments, and single and two extruders printed parts

Full size image

In these mechanical tests, the high flexibility of the NinjaFlex material was visible, since it reached the maximum deformation (limited by the maximum displacement of the jaws in the testing machine) without rupturing, in the case of the 3D-printed sample.

In turn, the tensile strength of the same PA specimens is shown Fig. 10, also divided by the two form factors (filament and printed specimens). Similar to the NinjaFlex case, a performance loss from the first columns of each process to their respective second columns is visible. In particular, PA filament behaves as a rigid material, having a rupture stress value of around 188 MPa. During the mechanical tests, and after some displacement, the five tested specimens ruptured near the vicinity of the jaws of the testing equipment. For both two-component part specimens (i.e., welded filaments and two extruders 3D-printed parts), the five specimens broke at the interface zone, after relatively low deformations. Finally, it should be noted that in the specimens printed by a single extruder, as the specimens underwent deformation, there was a progressive rupture.

Fig. 10
figure 10

Tensile strength of the four types of PA test specimens obtained for monolithic and welded filaments and single and two extruders printed parts

Full size image

To complement the results shown, it should be mentioned that in a preliminary study performed during the development of this methodology, high impact polystyrene (HIPS) filament could not weld with any of the flexible materials used in the present study. In addition, the corresponding hybrid prints broke during handling, after printing, preventing tensile tests from taking place. Thus, this is the case of non-compatible material pairs also identified by the developed methodology.

5 Conclusion

Given the technological advance of the multi-material FFF 3D printing technique, the present paper had as its main intention to explore an expedited method to infer the affinity between different material filaments.

Faced with the enormous challenge of printing with two different materials, one question beforehand is if the materials selected will perform well together. The trial-and-error study to evaluate such performance is time demanding and material-consuming, thus the hypothesis of considering filament welding as an indicator for material compatibility would represent a quick and simple test. To explore this hypothesis, the study encompassed: (i) the analysis of performance loss when filaments of the same material are welded together; (ii) the analysis of the performance of hybrid filaments, i.e., filaments constituted by two different material filaments welded together; and (iii) the correlation between the filaments performance under welding vs. 3D FFF printed parts. The study considered two different groups of materials, rigid (PC and PA) and flexible (NinjaFlex, FilaFlex, and TPC 45).

The results showed that welding by itself induces a loss of performance in the welded filament when compared to the monolithic one. This would be expected, considering that the welding site is a weak zone, representing a discontinuity. Still, filaments that had a good performance after being welded together also were able to be printed together in a multi-material printed part, and despite the magnitude of the results being different, the performance profile was the same. Thus, one can confirm that filament welding can serve as a quick and easy test to infer materials’ affinity in 3D-printed parts.

While this work used a relatively sophisticated hot plate welding equipment to guarantee reproducible conditions (alignment, heating time and temperature, and joining force), a common user can use simpler methods such as iron welding (commonly used on electronics) to perform similar filament affinity testing. Still, it is important to lay out that welding is highly dependent on the relative filament alignment, as misalignment can promote failure by reduction of effective resistant area and not necessarily by poor materials affinity.