Influence of printing material moisture on part adhesion in fused filament fabrication 3D printing

ABSTRACT Fused filament fabrication plays an increasingly important role in modern manufacturing. Despite this, issues like deformations caused by thermal shrinkage are still common. These process failures, called warping, can easily be avoided by ensuring sufficient adhesion of the printed part to the build surface during the manufacturing process. Nevertheless, only a few is known about the factors which have an impact on the adhesion between a part to be printed and the build surface. Although the content of moisture in the used polymer plays an important role in every established processing method, its influence on adhesion is still unknown for fused filament fabrication. This publication investigates the influence of moisture in the printing material for the examples of build surfaces made from Pertinax and borosilicate glass and printing materials such as polylactide acid, polyvinyl alcohol and polyamide. These printing materials were characterized by thermogravimetric analyses and differential scanning calorimetry. Following adhesion tests showed that the moisture content of the printing material can alter the adhesion between the printed part and the build surface up to 68%. It was also shown that there is an optimum moisture content at which maximum adhesion is reached.


Introduction
Fused filament fabrication (FFF) or material extrusion (MEX) is one of the most used additive manufacturing processes. [1]It offers many advantages like high freedom of design or the possibility to produce highly individualized products at low costs and with a wide variety of materials. [2]Many industries like aerospace, medical technology [3] and automotive use this technology. [2]espite the widespread usage of fused filament fabrication, a common issue yet remains: Thermal shrinkage caused by the cooling of the extruded material can result in deviations between the desired and the printed geometry. [4]This phenomenon is called warping.The most important way to avoid warping is to ensure sufficient adhesion between the part to be printed and the build surface. [4]Despite its relevance, only a few publications deal in detail with this topic.M. Kujawa [5] investigated the adhesion between acrylonitrile butadiene styrene (ABS) and borosilicate glass in dependence on the first layer parameters.Further work was done by M. Spoerk et al. [6,7] who investigated the influence of several process parameters like temperature, roughness or material of the chosen build surface.
In a previous study, [8] we examined the influence of printing speed, first layer's height, temperature settings and various build surface materials on the adhesion between the printed part and the build surface.
Yet unknown is the influence of the moisture adsorbed by the printed plastic onto the adhesion.It is well known that moisture influences the part properties not only in injection molding [9] but also in fused filament fabrication.E. Kim et al. [10] measured the moisture of ABS as printing material which is up to seven mass-percent.Due to the process-related cavities, parts made by fused filament fabrication have higher moisture adsorption than specimens manufactured via injection molding.This has an impact on the measured mechanical properties of the specimens.The measured tensile strength of moist specimens is lower than the one measured for dried specimens.Also, A. Banjo et al. [11] investigated the influence of moisture on 3D-printed parts.Due to the effects of moisture, the crystallinity of the printed polymers is changed which results in different mechanical properties of the manufactured part.Furthermore, the presence of water with its hydroxy groups can alter the strength of adhesive bonds [12] due to the formation of hydrogen bonds. [13]hese results raise the question if moisture can influence the adhesion between the part to be printed and the build surface.
For the terminology used in this publication reference is made to DIN EN ISO/ASTM 52,900 :2017 Additive manufacturing -General principles -Terminology. [14]The terms "build surface" and "build platform" are used synonymously because only the first printed layer is considered for all measurements done.This is in accordance with point 2.3.7 note 1 of the cited standard.

Experimental setup
In the following, the characterization of the printed material and the build surface is described.Also, the used method and device for measuring the adhesion between the printed material and the build surface are explained.

Printing material
Three different printing materials were used for the following investigations: polyvinyl alcohol (PVA) from FormFutura, Netherlands, polylactide acid (PLA) from Filamentworld, Germany (article number PLA175XBLK1) as well as polyamide F3 PA PurePro from Fiberthree, Germany (following called PA).The samples for the following measurements were collected by rasping the filament off into an unused glass vial with a clean and unused rasp.These rasps were only used for differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) described following.
At first, the glass transition temperatures T g were determined via DSC measurements for moist and dried specimens.The used measurement method is based on the method suggested by A. Sambale et al. [15] A device type DSC3 from Mettler Toledo, Germany, using an FRS 5+ sensor with the corresponding software Star SW version 16.00 and a heating rate of 20 K/min under a dry nitrogen atmosphere was used.The specimens were weighted in at room temperature, rapidly cooled down to the starting temperature of −50°C and then heated to 130°C.The first part of the specimens was dried for 24 h at 60°C and the second was stored in a non-air-conditioned laboratory at room temperature for 24 h at a relative humidity of 40%.A third part made from PA and PLA was stored in a tap water bath for 48 h at 22°C.The measured glass transition points are listed in Table 1.
The amount of absorbed moisture was determined via TGA analysis.For this, one specimen from each printing material was dried at 60°C for 4 h (called dried specimen).A further specimen for each material was stored for 24 h in a non-air-conditioned room (called moist specimen).The temperature and humidity in this room were recorded and ranged between 19°C and 23°C respective 25% and 30%.Further specimens made from PA and PLA were stored for 48 h in tap water at 22°C (called wet specimens).
The thermogravimetric analysis was done with a device type TGA 1 from Mettler Toledo, Germany, under an air atmosphere with a heating rate of 10 K/min starting from room temperature up to a temperature of 200°C.The analysis of the measurement data was done using the software STAR SW version 16.00.All results are shown in Table 2.
To ensure that the printing material is in a stable state, a moisture uptake curve was determined.For this, spools of each printing material were first dried at 60°C for 24 h and then stored at room temperature and without air conditioning (approximately 22°C and 50% humidity).Meanwhile, the weight is measured every 5 s with a precision of ±1 g.The determined moisture uptake curves are shown in Figure 1.As shown in Figure 1, the weight of the PLA filament stays nearly constant right from the beginning of the measurement and changes only within the error of the scale.It can therefore be assumed that the moisture absorption from PLA is neglectable which matches the result of the thermogravimetric analysis (see Table 2).This is also confirmed by a previous study done by A. Duigou et al.. [8] In contrast to PLA, the weight of the PA filament increases within the first 17 h by about 0.35% and remains constant afterwards (see Figure 1).This is also in line with the results from the thermogravimetric analyses which showed a difference of 0.5% between the moisture content of the dried specimen and the one stored under ambient conditions (see Table 2).A previous study done by P. Adriaensens et al. [16] measured a time span of about 15 h till the moisture content of PA saturated which fits the results shown in Figure 1.
PVA shows in comparison to PA and PLA the largest relative increase of weight and the shortest time to saturation (see Figure 1).Within about 2.5 h the filament weight increased about 1% and remains constant afterwards.This increase of moisture matches the results of the thermogravimetric analysis (see Table 2).
These results show that storing the printing material for 24 h under ambient conditions leads to a stable and saturated moist condition.

Build surface material
Build surfaces made from borosilicate glass (Filafarm, Germany, article number FF-SG-100100) and Pertinax (Filafarm, Germany, article number FF-PERTIP-200200) were used.Because the surface roughness influences the adhesion between the part to be printed and the build surface, [6] the surface roughness R z and the mean surface roughness R a were determined (see Table 3).For this, a confocal microscope type PLu Neox from Sensofar, Spain, was used.

Measurement method and device
The adhesion between the printed part and the build surface adhesion is measured via a 90° adhesion peel test based on DIN EN 28,510-1. [17]This standard was originally developed for testing adhesives and is adapted to determine the adhesion of parts manufactured by fused filament fabrication.
It was chosen because of its similarity to the failures occurring in warping.In the peeling test, the specimen adheres at first over the complete contact zone to the surface.Then, the specimen is loaded with an upwards-directed force by a tensile testing machine.Rising forces lead to a detachment of the specimen beginning from the edge and progressing through the contact zone between the specimen and the build surface.Warping also leads to a progressing detachment starting from one edge, which is similar to the detachment progress observed for a 90° adhesion peel test.Because only the force acting in the moment of detachment is used for evaluation, the adhesion force equals the measured force at this point in time.For this reason, hereafter the measured force is called adhesion force.
To ensure testing conditions are as similar as possible to the conditions prevailing in the manufacturing process, the tensile testing machine and FFF system are combined in one device (see Figure 2).For example, process temperatures can be kept during the complete measurement process.A detailed description of the used measurement device including validation of the method is described in Laumann et al. [18] To be able to measure the adhesion forces of 3D printed parts, an eyelet must be added to the specimen described in DIN EN 28,510-1 (see Figure 3).In this eyelet, a hook is inserted to imprint forces onto the specimen.To prevent moisture adsorption of the dried material during the printing process, a filament storage box was developed and added to the measurement device (see Figure 4).
This sealed box has space for one filament spool with a maximum diameter of 220 mm and allows storage under controlled temperature.To ensure dry conditions, the temperature inside can be chosen between room temperature and up to about 65°C.Also, a box (Figure 4 number 9) with 20 g silica gel with moisture indicator (Steiner Chemie, Germany, type 340-534996) is mounted at the bottom part.Air humidity cannot be controlled but is measured and, in contrast to commercially available filament storage boxes, logged together with the inside temperature.This is done by a DHT 22 type moisture sensor and Figure 2. Device for measuring the adhesion force between specimen and build surface.The upper part contains the tensile testing device and the lower part the FFF system. 1) filament storage box 2) traverse of the tensile testing machine 3) load cell 4) 600 mm long wire between hook and load cell 5) extrusion head 6) specimen with eyelet and inserted hook 7) build surface. [18] Semitec 104NT type temperature sensor which are read by an Arduino Mega 2560.A self-developed Python script writes the values received by the Arduino to a CSV-File every 20 s.The filament dried in the universal oven was stored between 64°C and 66°C at a relative air humidity of 19% during printing.[18] all dimensions are given in millimeter.The dried printing material is guided through a PTFE tube to the extrusion unit and from there to the extrusion head.Only at the extruder unit, the printing material is exposed to the environment (Figure 4 number 13).The exposure time is about 25 s which is negligible as our measurements (see Figure 1) and previous studies [10] show.
In the following diagrams, showing the adhesion force, each data point indicates the average out of the maximum measured adhesion forces of five single measurements.The error bars display the standard error calculated out of the five single maximum measured forces.All process parameters used for manufacturing the specimens are stated in Table 4.The first layer's height has a significant influence on the adhesion force.In a previous study, [18,19] we found this to be the optimum for PLA.As long as other printed materials do not exceed the possible extrusion force of the printer, sufficient material will be extruded and the first layer is filled with the same amount of material and hence the conditions should be comparable. [19,20]The printing speed is defined as the speed set in the slicing software and is set constant for the complete printing process.A comparison between the real printing speed and the one set in the software was done in a previous study. [18]

Results and discussion
As shown in Tables 1 and 2, PVA and PA can have a significant moisture content and show different glass transition temperatures for dry and moist material.As previously described, [5][6][7]19] there is an optimum build surface temperature slightly above the glass transition temperature for borosilicate glass as well as Pertinax as build surface. Henc, a measurement series was done to investigate if there is a shift in this optimum build surface temperature according to the shift of the glass transition temperature caused by the moisture.Therefore, specimens were printed with three different moisture contents onto build surfaces made from borosilicate glass and Pertinax.These three different moisture contents of the printing material are following called dry, moist and wet.Dry printing material is printed out of the filament box as described above.Moist printing material is first stored for at least 24 h at room temperature without air conditioning (about 19°C to 23°C with a humidity between 25% and 30%).Wet printing material was stored for at least for 24 h in a tap water bath at 22°C.The processing of the wet printing material also took place from this water bath.Due to the solubility of PVA in water, no specimens made out of wet PVA were printed The results of the measurement series are shown in Figure 5. Here, the measured adhesion force is plotted versus the chosen build surface temperature for wet, moist and dried printing material.The highest adhesion force is measured for moist PVA as printing material with a maximum of 208.9 N as printed onto a build surface made out of borosilicate glass.After drying the PVA, the maximum measured adhesion force decreases about 60% to a value of 82.8 N. Also, a shift of the optimum build surface temperature is observed.Following, the optimum build surface temperature is defined as the temperature at which the maximum adhesion force is measured.In case of PVA printed onto borosilicate glass (see Figure 5b), the optimum build surface temperature is found at 60°C for moist material.Drying the printing material leads to a higher optimum build surface temperature of 80°C.
For PA printed onto Pertinax (Figure 5c) and borosilicate glass (Figure 5d) the highest maximum adhesion force is measured for moist printing material with about 166 N for both build surface materials.Drying reduces the maximum measured adhesion force by about 30 N to 50 N depending on the build surface material.Also, a higher moisture content decreases the adhesion force depending highly on the build surface material.For Pertinax as build surface material the adhesion force is reduced by about 60 N.In contrast, a decrease of about 110 N is observed for borosilicate glass as build surface material.
Both materials have a high moisture content (see Table 2) in comparison to PLA.For PLA a moisture content was determined which is an order of magnitude smaller than measured for PVA and PA.A possible explanation are additives. [21]Due to this small moisture content, for PLA printed on Pertinax and borosilicate glass a nearly constant maximum measured adhesion force is measured for dry, moist and wet printing material (see Figure 5(e,f)).Furthermore, the shift of the optimum build surface temperature is only small in comparison with the observations made for PVA and PA as printing material.For PLA printed onto borosilicate glass, the optimum build surface temperature changes only about 10°C while for Pertinax as build surface material no shift was measured.
For the observed decrease of the maximum measured adhesion force after drying, the following explanation is suggested: Absorbed water increases the mobility of the polymer chains [16,22] because it increases the space between them. [23]As previous studies show, the adhesion of polymers depends on the distribution of the polymer chain mobility. [24,25]On the one hand, the polymer chains must be mobile enough to adapt to the surface and to form a large contact area for a strong adhesion.On the other hand, the polymer chains must be firmly connected with other chains to ensure a good adhesion of the  1).All specimens were printed with the process parameters stated in table 4. Due to the solubility of PVA in water no measurements were conducted for wet PVA.
complete specimen.This means that there is an optimum chain mobility, where the adhesion reaches a maximum.Therefore, altering the chain mobility by absorbing water can lead to higher or lower adhesion. [24,25]nother point to be mentioned is the influence of absorbed water on the ability to form intermolecular bonds: Absorbed water tends to cluster at the polar groups of the polymers which leads to a higher capability of forming hydrogen bonds. [23]As the results of Park et al. indicate, [24] the adhesion between the build surface and printing material could be caused by hydrogen bonds.According to this, a higher capability of forming hydrogen bonds can explain the higher adhesion forces measured for moist printing materials like PVA and PA.
If too many bondings per area are formed, the distribution of mechanical stresses is affected which results in a superelevation of the mechanical tension. [26,27]This results in a decreased adhesion force as observed for the wet printing material.It is suggested that the high amount of absorbed water leads to a very large density of hydrogen bonds per area between the printed material and the build surface.
The shift of the optimum build surface temperature can also be explained by a changing mobility of the polymer chains.Absorbed water increases the chain mobility which leads to a decreased glass transition temperature. [22,23]19] A comparison between the glass transition temperatures determined via DSC (see Table 1) and the optimum build surface temperatures observed in the measurement series shown in Figure 5 confirms this: For PVA the glass transition temperature determined via DSC increases about 19°C after drying and the optimum build surface temperature shifts about 20°C towards higher values (see Figure 5(a,b)).This is observed for borosilicate glass as well as for Pertinax as build surface materials.Also, for PA as printing material the differences between the glass transition temperature and the optimum build surface temperature match (see Figure 5(c,d)).For PLA, only for the wet printing material, a change in the glass transition temperature was measured in the DSC analysis and consistent with this, only with the wet printing material a shift of the optimum build surface temperature was observed (see Figure 5(e,f)).

Conclusion and outlook
The results presented above show for the first time that the moisture content of the printing material affects the adhesion in fused filament fabrication and how the process parameters should be chosen.For hygroscopic materials like PA and PVA, the optimum build surface temperature at which a maximum adhesion force is measured, shifts towards higher temperatures after drying.Therefore, drying should always going along with an adjustment of the process parameters.Furthermore, the maximum measured adhesion force is reduced by up to 60%.Increased moisture contents shift the optimum build surface temperature to lower temperatures.The overly moist printing material shows a strongly decreased adhesion which is lowered up to about 68%.A possible explanation for the observed effects could be the changed mobility of the polymer chains due to the absorption of water.Previous studies [22][23][24][25] showed that an increased chain mobility goes along with changes in the strength of the adhesion of polymers and decreased glass transition temperatures.Also, absorbed water tends to cluster at polar groups of polymers which increases the ability of forming hydrogen bonds.According to previous studies [26,27] too much bondings per area can decrease the strength of an adhesive contact due to an unfavorable distribution of mechanical tension.
Moisture also affects the viscosity of polymers. [28]The viscosity is a quantity which must be taken into account by choosing various process parameters. [29]herefore, drying the printing material could affect how the printing speed, the nozzle temperature or the first layer's height should be chosen to achieve a maximum adhesion force.Further investigations should be done to answer these questions.

Figure 1 .
Figure 1.Water uptake curves of the used printing materials taken at room temperature and a humidity of about 50%.

Figure 5 .
Figure 5. Adhesion forces measured for dried, moist and wet printing material and build surfaces made from borosilicate glass and Pertinax.The dashed lines mark the glass transition temperature determined via DSC (see table1).All specimens were printed with the process parameters stated in table 4. Due to the solubility of PVA in water no measurements were conducted for wet PVA.

Table 1 .
Glass transition temperatures of the used printing materials determined using DSC.Due to the solubility of PVA in water no measurements were conducted for wet PVA.

Table 2 .
Moisture content of the printing material determined using TGA.Due to the solubility in water, no measurements were conducted for wet PVA.

Table 3 .
Characterization of the build surfaces regarding the surface roughness.

Table 4 .
Process parameters used for the manufacturing of the specimens.