Discarded single-use plastic water bottles seem to be everywhere, and the polymer they are made from (PET, or polyethylene terephthalate) can be recycled via an extrusion process. This makes PET bottles attractive as a source of feedstock for fused granular fabrication (FGF) printers such as re:3D’s GigabotX. However, printing with it does present challenges. PET is very hygroscopic - that is, it readily absorbs water from the environment. High moisture content during the extrusion process can lead to a number of problems with printed articles, from poor surface quality to reduced part strength.
PET has an equilibrium moisture content of 2000-6000 PPM (0.20-0.60%) [1], but published references typically recommended a maximum moisture content of 100-500 PPM (or less) [2] [3] for PET pellets or flake used in extrusion processing. There are a number of technologies used for reducing the moisture content of plastic feedstock. re:3D uses both industrial recirculating desiccant dryers and small forced hot-air appliances for this purpose (Figure One). Recently, some drying analyses of rPET water bottle flake were conducted using both styles of dryers to determine how effectively they perform.
Figure One: Forced hot-air food dehydrator (left); Industrial desiccant dryer (right).
Desiccant dryers operate by flowing dry, heated air through the granular material. The hot air is recycled through a desiccant bed to remove any moisture released from the material and maintain a very low dew point of the processing air (typically -40°C.) Desiccant drying of the flake at temperatures of 135-160°C is recommended [1], but this is above the glass transition temperature of PET (69-85°C) [4], and drying at the recommended temperatures can sometimes require constant agitation or stirring to prevent clumping of the softened flakes.
In order to determine the effectiveness of drying the PET feedstock at lower temperatures, 3.6 kg of rPET water bottle flake were processed in a Dri-Air MPD 5-30D portable desiccant dryer at a temperature of 65°C (dew point of -45°C.) The moisture content of the flake was measured at intervals using a Radwag PMR 50/1 gravimetric analyzer. The Radwag analyzer determines the moisture based on the “Loss on Drying” method. An analysis temperature of 110°C using the analyzer’s “Auto 5” method for the end condition was selected.
Replicant samples (n=5-7, typical) of 4-7 grams of PET flake were collected from the dryer and analyzed throughout the drying period. The initial moisture content of the flake was approximately 1.1% (11,000 PPM.) The high moisture content is likely caused by not having completely dried the bottles prior to granulation (i.e., there were still water drops in many of the ‘empty’ bottles.) This value is consistent with moisture content values for PET bottle flake found online [5]. After more than 16 hours of drying, the average flake moisture measurement was still over 1000 PPM as shown in Figure Two. The drying temperature was then increased to 120°C. Samples collected after an additional seven hours of drying had an average moisture content of 800 PPM. There was no apparent clumping of the flake in the dryer hopper after this period of drying at the elevated temperature.
Figure Two: Measured moisture content of rPET flake as it is dried in a desiccant dryer.
At four hours or more of drying, the gravimetric moisture content measurements begin to show a time-series trend within each interval group (Figure Three). The first moisture measurement in each set is less than 400 PPM, and the moisture content increased with each successive measurement. For six measurement replicates per drying interval, each one averaging 5-8 minutes to complete, the measurement set can take a total of 35-50 minutes. The results show that even if the flake is dried to less than 100 PPM as recommended, simply leaving the dried flake exposed to room ambient conditions (21°C, 35-40 %RH) can quickly increase the flake moisture content to 1500 PPM, with a final equilibrium value of 2500 PPM or more.
Figure Three: Time-dependency of moisture content of rPET flake over successive replicate measurements; desiccant dryer samples.
This observed rapid absorption of moisture by PET is consistent with published data (Figure Four) [6] and calls to question the applicable relevance of achieving a moisture content below 500 PPM. SInce the GigabotX exposes the flake feedstock to ambient temperature and humidity in a hopper during the course of printing (which can take multiple hours or days for large objects), the feedstock will quickly assume some moisture condition well above its initial dried value. Given this consideration, desiccant drying of the rPET flake at 65°C beyond 4-6 hours may provide limited benefit, with the understanding that absorption of moisture from the environment can potentially lead to processing issues and material property degradation of printed parts.
Figure Four: Moisture content over time for PET and PEN (polyethylene naphthalate) exposed to ambient conditions. (Credit: Macdonald, 2012)
re:3D also uses small kitchen food dehydrators to dry material. These units force heated air through trays of feedstock, but the hot air is simply exhausted out of the unit, and the intake air is not dried. Because the drying air does not have a low dew point, this method is not considered as effective as desiccant drying in removing moisture from many materials. These food dehydrators are, however, inexpensive and readily available to many GigabotX users who might not have access to an industrial desiccant dryer. They are also convenient for drying small volumes of material at a time..
A drying analysis of rPET water bottle flake was performed using a Westinghouse 245 Watt food dehydrator (Model WFD101W) with the temperature set to 70°C, the maximum setting on the appliance. Freshly granulated rPET flake was loaded into the bottom two trays of the unit, and the exit air temperature and relative humidity was recorded by a digital data logger placed in the top, empty tray. As before, flake samples were pulled at intervals and analyzed for moisture content using the Radwag moisture analyzer. The results are shown in Figure Five.
The initial moisture content measurements had significant variance (likely due to observed instability in the Radwag analyzer) but averaged approximately 1.7% (17,000 PPM.) The food dehydrator proved surprisingly capable of rapidly lowering the moisture content of the small amount of rPET flake in the unit, and drying the flake beyond 6 hours did not yield appreciable additional reductions.
Once again, a time-dependency is noticeable in the replicate measurements from each time interval. Care was taken to record the timing of the replicate measurements relative to when each interval sample was removed from the dehydrator. As shown in Figure Six, while the first moisture content measurement of each sample may have been below 200 PPM, successive measurements reliably increased over time as the sample sat exposed to ambient conditions (recorded as 22°C, 50-60 %RH), with the flake approaching 2000 PPM moisture content over the course of an hour.
Figure Five: Measured moisture content of rPET flake as it is dried in a food dehydrator at 70°C.
Figure Six: Time-dependency of moisture content of rPET flake over successive replicate measurements; food dehydrator samples.
These results demonstrate that a consumer food dehydrator can be effective in drying small (< 2kg) samples of rPET water bottle flake in comparison to a commercial desiccant dryer. Neither system was capable of drying the flake below 100 PPM at 65-70°C, but PET’s rapid absorption of moisture from the environment might make this an unreasonable target when conditioning feedstock for GigabotX printers.
If the “dried” feedstock can be expected to assume a moisture content of 1500 - 2500 PPM during the course of a long print, one of the next steps is to measure the effect of elevated moisture on the quality and properties of parts printed with the GigabotX. re:3D has the capacity to test the mechanical properties of printed specimens, but there will need to be some clever thinking about controlling for changes in feedstock moisture as it sits exposed over the course of successive prints. Some smart folks are already working on this and hope to share results in the coming months.
If you have thoughts, comments or questions about printing with recycled PET flake, please share them below in the comments or reach out to us directly.
References
[1] (n.d.). Overview of Materials for Polyethylene Terephthalate (PET), Unreinforced. MatWeb Material Property Data. https://www.matweb.com/
[2] American Plastics Council (1997). Best Practices in PET Recycling: Drying Methods and Requirements. Pollution Prevention Infohouse. https://p2infohouse.org/ref/14/13537.pdf
[3] Ronkay, F., Molnar, B., Szabo, E., Marosi, G., & Bocz, K. (2022). Water Boosts Reactive Toughening of PET. Polymer Degradation and Stability. https://doi.org/10.1016/j.polymdegradstab.2022.110052
[4] C-Therm (2016). Glass Transition Temperature of PET. Thermal Analysis Labs. https://ctherm.com/resources/newsroom/thermal-analysis-labs/determining-glass-transition-temperature-of-pet/
[5] Kreyenborg GmbH & Co. (2015). Drying and Crystallization of PET to 50 ppm. Applications: Plastic. https://www.kreyenborg.com/en/application/drying-and-crystallization-of-pet-to-50-ppm/
[6] Mcdonald, W. A., Looney, M. K., MacKerron, D. A., Eveson, R., Adam, R., Hashimoto, K., & Rakos, K. (2012). Latest Advances in Substrates for Flexible Electronics. Journal of the Society for Information Display. https://doi.org/10.1889/1.2825093
