Frequently, only specialised 3D printing companies closely monitor the storage conditions of unsealed filaments. In most room conditions, filaments are subjected to an atmosphere of 20-50% relative humidity (RH). In such conditions, filaments tend to absorb water from the air, swell, and degrade. The aim of this research was to investigate the water sorption, desorption, and swelling of different polymer filaments used for 3D printing and model these processes.
To prevent filament exposure to humid air, filaments were contained in the producer vacuum package until the beginning of experiments. After unsealing the package, filaments were cut into cylindrical shape samples of length $l = 100$ mm and diameter of $2R = 2.8$ mm. Samples were put into desiccators with constant RH created by silica gel for 12% and saturated aqueous solutions of salts LiCl for 16%, KSCN for 47%, NaCl for 75%, and K2SO4 for 97%. Samples were taken out of desiccators every 24 h to weigh them for sorption and measure length for swelling calculations. Mettler Toledo XS205DU balance with precision $\pm 0.1$ mg was used for weighing samples. Mitutoyo Absolute digital micrometre with a precision of $\pm 0.001$ mm was used for sample length measurements. The sorption of filaments was modelled using Eq. (1) which allows calculating sample moisture content $Q$ at time $t$
$$Q = Q_{\infty} + \frac{8(Q_0 - Q_{\infty})}{\pi^2} \sum_{k=1}^{\infty} \sum_{m=1}^{\infty} \frac{\exp(-\lambda_{k,m}^2 Dt)}{m^2 \gamma_k^2} \left(1 - (-1)^m\right)^2, \ \lambda_{k,m}^2 = \left(\frac{\gamma_k}{R}\right)^2 + \left(\frac{\pi m}{l}\right)^2 \tag{1}$$where $D$ is the diffusion coefficient, $Q_{\infty}$ and $Q_0$ are the equilibrium and starting moisture contents, $\lambda_{m,k}$ is a geometry-based parameter with $\gamma_k$ being a table constant. Two unknown variables, $D$ and $Q_{\infty}$, were found using curve fitting to experimental data. The aim function minimised during the fitting procedure was calculated as average relative deviation. The sorption isotherm indicated the change in equilibrium moisture content at different RH and was approximated by Eq. (2).
$$Q_{co} = a \cdot RH^c + b \tag{2}$$where $a$, $b$, and $c$ are coefficients unique to every material. This approximation was based on the BET theory and Henry's law. Similarly, a relation between equilibrium moisture content and swelling was plotted and modelled as a linear function.
All materials have lost mass in dry environments and gained in humid. Most materials have an equilibrium moisture content of 0.3-0.6% at 97% RH. There are outliers: Koltron has the lowest of 0.1%, and Nylon Transparent has the highest of 8%, followed by ABS-ESD of 3.5%. Examination of diffusion coefficients dependence on RH showed that for PEI, PEKK, PC white, CPE White, Nylon Transparent, and PLA Tough coefficient remained constant; while for PLA LAVA and PET-G White it increased at higher humidity; but for ABS White, ABS ESD, Koltron, and PLA Conductive coefficient at higher humidity decreased. All materials swelled maximally by 0.1-0.45% in the atmosphere 97% RH. Nylon Transparent is an exception that shrunk in the humid environment by 2.43%.
Obtained results demonstrated the specific significance of implementing storage rules for some 3D printing filaments during the work.