The Effects of Washing and Formaldehyde Sterilization on the Mechanical Performance of Poly(methyl Methacrylate) (PMMA) Parts Produced by Material Extrusion‐Based Additive Manufacturing or Material Jetting

Nowadays, personalized medical implants are frequently produced through additive manufacturing. As all medical devices have to undergo specific washing and sterilization before application, the effects of a predefined cleaning routine that is available to the clinical institutes, washing with chemical agent and formaldehyde fumigation, on the mechanical behavior of printed parts are examined. Mechanical properties of parts manufactured by fused filament fabrication (FFF) and ARBURG plastic freeforming (APF) using two poly(methyl methacrylate) (PMMA)‐based materials, 3Diakon and CYROLITE MD H12, respectively, are analyzed using flexural and impact tests. An influence of cleaning treatments on the mechanical properties of APF samples is not detected. FFF samples, however, show lower impact strength after washing, but not after sterilization. The fracture surfaces, porosity values, or chemical structure assessed by Fourier‐transform infrared (FTIR) spectroscopy could not explain this decrease. Influence of the cleaning treatments on the material itself is assessed using thin compression‐molded specimens. The influence on the stress–strain curves is negligible, apart from a slight but significant reduction in the yield stress. FTIR spectroscopy and scanning electron microscopy analyses of the fracture surfaces do not show detectable differences among differentially treated samples.


Introduction
Additive manufacturing (AM) is of particular interest in the production of medical implants as it allows faster design and manufacturing of personalized prostheses. For polymers, especially the material extrusion based, AM technology known as fused filament fabrication (FFF) has already been well established. [1][2][3] Nonetheless, all AM methods are continuously improving and new technologies are developed. The ARBURG plastic freeforming (APF) process is a relatively new AM method, where granules, instead of filaments, are molten and deposited as droplets. Therefore, a plasticization unit similar to that of an injection-molding machine provides the molten material and pressure for the deposition process. After plasticization, the polymeric material enters the discharge unit, consisting of a nozzle and a piezoelectric valve, which opens the nozzle up to 200 times per second. [4] As the nozzle opens and closes at such high frequencies, the extruded melt forms droplets in the freeforming method instead of the continuous string obtained in the FFF process. Hence, it is classified as a material jetting technology according to ISO/ASTM 52 900.
Nowadays, personalized medical implants are frequently produced through additive manufacturing. As all medical devices have to undergo specific washing and sterilization before application, the effects of a predefined cleaning routine that is available to the clinical institutes, washing with chemical agent and formaldehyde fumigation, on the mechanical behavior of printed parts are examined. Mechanical properties of parts manufactured by fused filament fabrication (FFF) and ARBURG plastic freeforming (APF) using two poly(methyl methacrylate) (PMMA)-based materials, 3Diakon and CYROLITE MD H12, respectively, are analyzed using flexural and impact tests. An influence of cleaning treatments on the mechanical properties of APF samples is not detected. FFF samples, however, show lower impact strength after washing, but not after sterilization. The fracture surfaces, porosity values, or chemical structure assessed by Fourier-transform infrared (FTIR) spectroscopy could not explain this decrease. Influence of the cleaning treatments on the material itself is assessed using thin compression-molded specimens. The influence on the stress-strain curves is negligible, apart from a slight but significant reduction in the yield stress. FTIR spectroscopy and scanning electron microscopy analyses of the fracture surfaces do not show detectable differences among differentially treated samples.
Several materials are suitable for processing with both manufacturing methods. Among these, poly(methyl methacrylate) (PMMA) stands out with its promising properties. It has already been used for many years for various medical applications such as optical lenses, bone cement for orthopedic and cranial implants, or prostheses in dentistry. [4,5] Therefore, it offers a good starting point for further investigations regarding the use of AM and the impact of nonthermal sterilization methods on its biocompatibility and biomechanical properties.
To enable the implantation of prosthetic devices produced by FFF and APF in humans, the material and manufacturing processes have to comply with the clinical safety measures. In addition to durability and biocompatibility, implant materials have to be sterilizable. The medical products must be free of any pathogens and contaminants, while reliably maintaining their mechanical properties and dimensional accuracy. [6] Although autoclaving is a suitable sterilization method for implants made of metals, ceramics, and high-performance thermoplastics, polymers with low melting temperatures (e.g., PMMA) are deformed at the high temperatures of the autoclaving process. Sterilization methods that could be utilized for specimens manufactured from polymers with low melting temperatures include chemical sterilization (ethylene oxide or formaldehyde), radiation sterilization (gamma, electron beam, or X-ray), plasma sterilization, and microwave sterilization. [7,8] In this work, sterilization by formaldehyde fumigation was studied, as it is a sterilization procedure commonly available in most clinical institutes.
Several studies analyzing the effects of different sterilization methods on the resulting properties of PMMA have already been conducted. [9][10][11][12][13][14] However, medical devices usually undergo a cleaning step preceding sterilization, [15] which could influence the material, as this process usually involves temperature, pressure, and an aqueous solution of washing agent. [16] It is known that polymers in general, and thus also PMMA, exhibit temperature-, [17] pressure- [18] and moisture-dependent [19,20] properties. Avila et al., [21] for instance, showed that heat treatment of 3D-printed PMMA at 97°C for 60 min led to an increase in strength of about 20 MPa. Given their porous structure, additive-manufactured specimens could be influenced by these procedures to an extent larger than the molded specimens, as their absorption-desorption behavior might differ from the solid, nonporous, molded specimen with a rather smooth surface.
Furthermore, the AM process introduces a certain level of porosity, which could potentially influence the mechanical performance depending on its extent. Hence, the porosity of FFF and APF samples was analyzed before testing and after the different treatment steps, to rule out the effect of different porosity levels while analyzing the treatment influence on the mechanical performance. This has been done by means of X-ray microcomputed tomography (μCT), as it is well established as a nondestructive method to evaluate defect sizes, distributions, and individual shapes. [22] Consequently, this study investigates the effects of washing and formaldehyde sterilization on the porosity and mechanical properties of PMMA-based materials manufactured with FFF and APF. Bending and impact tests showed that the washing and sterilization procedures did not exert a significant change in mechanical performance of the FFF-and APF-manufactured specimens. μCT analyses ruled out any confounding influence of porosity on the results.

Materials
FFF-manufactured samples were produced using the commercially available PMMA filament (diameter of 1.75 mm) 3Diakon (Mitsubishi Chemical Advanced Materials Inc., USA), due to its excellent impact performance, which is preferable for several implant materials, such as cranial reconstruction materials. APF-manufactured samples were produced using pellets of CYROLITE MD H12 (kindly provided by Roehm GmbH, Germany), an amorphous thermoplastic compound based on PMMA (methyl methacrylate/styrene/ethyl acrylate terpolymer with an added impact modifier [23] ). CYROLITE MD H12 meets the requirements of the USA Pharmacopeia Class VI and is ISO 10 993-1 certified and approved by the USA Food and Drug Administration (FDA) for clinical use. [24] Throughout the article, 3Diakon and CYROLITE MD H12, are referred to as PMMA-D and PMMA-C, respectively.

Experimental Setup
Both PMMA types were divided into three different experimental groups (n ¼ 25 samples in each): untreated controls, washing, and washing þ sterilization. The "washing" specimens were washed at 60°C following a specific hygiene protocol. The "washing þ sterilization" group was additionally sterilized with formaldehyde fumigation. The mechanical properties of each group were assessed by three different testing methods: three-point bending, Charpy impact test, and Charpy-notched impact test. A flow diagram of the experimental setup is shown in Figure 1.

Processing
Test specimens were manufactured in the shape of a rectangular prism in 80 mm Â 4 mm Â 10 mm dimensions with a single contour line, a 100% infill, and a AE 45°rectilinear infill strategy. Five samples were fabricated per print batch via so-called sequential printing for FFF and in a layer-by-layer manner for APF ( Figure 2). A detailed summary of the used processing parameters for both technologies is given in Table 1.
FFF samples were manufactured at MedMEX (HAGE3D GmbH, Austria), which works with a dual direct extrusion head. The slicing was done with the software Simplify3D v3.0 (Simplify3D, USA). The printing speed of the first layer was decreased to 15 mm s À1 to get the best adhesion on the glass surface. The samples were removed from the printer after bed temperature cooled down to a temperature of 60-to-80°C and then stored in vacuumed Allpax GOF 2030 bags (Allpax Products LLC, USA) at ambient temperature.
In APF, an already proven material profile was used for processing PMMA-C with the freeformer 200-3X (ARBURG GmbH þ Co KG, Germany). [4] The print job was prepared in the ARBURG freeformer software v2.30 (ARBURG GmbH þ Co KG, Germany). Prior to manufacturing, the material was dried at 70°C for 5 h in the integrated circulating air dryer (Helios GmbH, Germany). After the drying procedure, the hopper was kept at 50°C with reduced moisture to keep the material dry. The drop height was 0.22 mm with the set discharge value of 67%, resulting in a layer height of 0.2 mm. The droplet overlap was set to 25%.
Dimensions within the mechanical testing standards could be obtained using APF, but not using FFF. Here, width values in the range of 9.49-9.82 mm were obtained, which were below the tolerance of 10 AE 0.2 mm.

Washing and Formaldehyde Sterilization
The sample washing and sterilization were performed at the AEMP III (Processing Unit for Medical Devices) at the University Hospital Graz in accordance with ÖNORM EN ISO 25 424. Washing was performed using a Miele cleaning-disinfection device (Miele & Cie. KG, Germany) at 60°C with a disinfectant program suitable for thermolabile materials. Neodisher Septo DN (Chemische Fabrik Dr. Weigert GmbH & Co KG, Germany) was used as the washing agent, which contained 10.5% (w/w) glutaraldehyde and showed bactericidal, fungicidal, mycobactericidal, and viricidal activity. Sterilization was performed with 2% formaldehyde using a Webeco FA95 temperature steam sterilizer (Webeco NV, Belgium). A summary of washing and sterilization procedures is given in Table 2.

Analysis of Porosity
Samples were scanned in an Inveon μCT scanner (Inveon μ-PET/ SPECT/CT, Siemens, Germany) with scanning parameters of 80 kV potential, 500 μA current, 750 ms exposure time, and an  www.advancedsciencenews.com www.aem-journal.com effective pixel size of 35.19 μm. The raw data was reconstructed using the Inveon CT Recon Software v2.04 software (Siemens, Germany). Image sections were exported from the reconstructed scan data and saved as bitmap digital image files using ImageJ. Each specimen was imaged in 2699 sectional images with sagittal and/or horizontal alignment. Segmentation, 3D modeling, and volumetric analyses of the printed specimens and the internal gaps/holes were done using an open-source software 3D Slicer v4.10.2. [25] A gap/hole in an image slice was defined as an island that had a signal intensity below the threshold value without connection to the outer surface through neighboring image sections. Porosity was calculated as the percentage of gap volumes with respect to the specimen volume. Local porosity at the midsection was calculated the same way using the image slices that cover one-third of the whole specimen centered at the midlength in the axial orientation.

Flexural Tests
The flexural tests were performed in three-point bending mode on the universal testing machine Zwick Z10 (Zwick Roell, Germany) equipped with a 10 kN load cell. The tests were carried out according to EN ISO 178 at 23°C and 50% r.h. The testing speed was 2 mm min À1 . The deformations were measured with the makroXtens extensometer. The test ended, if no fracture occurred, at a deflection of 10 mm. The support distance was 64 mm. Supports and loading edges were 5 mm in radius. Prior to testing, the samples were stored at standardized climate (23°C, 50% r.h.) for at least 48 h. The following flexural material properties were evaluated according to EN ISO 178: the flexural modulus (E f ), the maximum flexural stress (σ fM ), the flexural stress at break (σ fB ), and the flexural stress at the conventional deflection (σ fC ), wherein the conventional deflection (s C ) was equal to 1.5 times the thickness (h) of the test specimen. E f is defined as the slope of the flexural stress-flexural strain curve in the flexural strain interval between 0.05 and 0.25%. All stresses and strains are engineering values, which take the initial cross section of the specimen into account. Microscopic images of the fracture surface of representative specimens were taken under a light microscope (SZH, Olympus Optical Co., Japan).

Charpy Impact Tests
Instrumented Charpy impact tests were performed on the pendulum impact tester HIT25/50P (Zwick Roell, Germany) equipped with a 2 J pendulum at 23°C and 50% r.h. The tests were carried out according to EN ISO 179-2 on unnotched and notched specimens via edge-wise blows. The notch was introduced in the geometry of shape A according to ISO 179-1, resulting in a characteristic V-shape with 2 mm depth and 0.25 mm tip radius. The impact speed was 2.9 m s À1 according to the standard. Prior to testing, the samples were stored at standardized climate (23°C, 50% r.h.) for at least 48 h. The following parameters were evaluated according to standard (EN ISO 179-2): the Charpy unnotched (a cU ) and notched (a cN ) impact strength. Microscopic images of the fracture surface of representative specimens were taken under a light microscope (SZH, Olympus Optical Co., Japan). were randomly assigned for differential cleaning treatment (untreated, washing, washingþsterilization, n ¼ 5 per group).

Tensile Tests on CM Samples
The tensile tests were performed on a universal testing machine Zwick Z10 (Zwick Roell, Germany) equipped with a 10 kN load cell and mechanical clamps, in accordance with DIN EN ISO 527-1 with a testing speed of 1 mm min À1 for the evaluation of the Young's modulus (E) and 50 mm min À1 for the recording of the remaining curve. The clamping length was set to 50 mm. The deformations were measured with the makroXtens extensometer until yield and by the crosshead travel afterward. All stresses and strains are engineering values, which consider the initial cross section of the specimen. The Young's modulus and tensile strength (σ M ) or yield stress (σ y ) are calculated and compared. E is defined as the slope of the stress-strain curve in the strain interval between 0.05 and 0.25% according to DIN EN ISO 527-1. Furthermore, the tensile strength is the global stress maximum and the yield stress gives the stress value at the yield point, which is characterized by a global stress maximum followed by stress reduction due to narrowing of the cross section. [26]

Fourier Transform Infrared Spectroscopy
Fourier transform infrared (FTIR) spectroscopy by attenuated total reflection was performed with a Bruker IFS 66 v s À1 FTIR spectrometer (Bruker Corporation, USA) in the range of 600-4000 cm À1 . The penetration depth into the sample is up to %10 μm, depending on the wavelength of light. [27] Spectra were acquired and examined with 16 scans and 4 cm À1 resolution after spectral correction with ambient atmosphere. A spectrum of the washing agent was also recorded for comparison. The spectra were examined for alterations in the chemical structure or the presence of residual media after cleaning treatments.

Scanning Electron Microscopy
Scanning electron microscopy (SEM) was performed for selected unnotched Charpy samples using a TESCAN Vega II (TESCAN Brno, s.r.o., Czech Republic) at 5 kV using secondary electrons. The parts of the specimens to be analyzed were fixed on SEM sample holders and were gold-sputtered with the SCD 005 Cool Sputter Coater (BAL-TEC AG, Liechtenstein) for 160 s at 20 mA.

Statistical Analyses
Statistical analyses were conducted using SPSS ( A difference with a p-value ≤0.05 in any case was deemed statistically significant.

Results
No significant differences between the different types of treatments were observed in the flexural stress-flexural strain curves for both PMMA types as well as manufacturing methods ( Figure 3). Furthermore, FFF specimens neither showed a maximum nor failure before the conventional deflection s C (1.5 h % 6 mm, test ended at 10 mm) is reached ( Figure 3a). As the test is limited by s C , the flexural strength (σ fM ) and the flexural stress at break (σ fB ) were not evaluated and only the flexural modulus (E f ) and the flexural stress at s C (σ fC ) were analyzed. APF specimens fractured before reaching the deflection limit ( Figure 3b). Therefore, the maximum flexural stress (σ fM ) and the flexural stress at break (σ fB ), which were equal were evaluated in addition to the flexural modulus (E f ). Specimens from all treatment groups had comparable flexural moduli in the bending tests with both FFF-printed PMMA-D samples (F(2) ¼ 0.127, p ¼ 0.882) and APF-printed PMMA-C samples (F(2) ¼ 1.256, p ¼ 0.320) (Figure 3c). Similarly, the flexural stress (at the conventional deflection or at break) did not show statistically significant differences among the treatment groups either in FFF samples (F(2) ¼ 1.393, p ¼ 0.286) (Figure 3d) or in APF samples (Welch's F(2, 6.020) ¼ 1.208, p ¼ 0.362) (Figure 3e). It is important to note that the different flexural behavior was a result of the intrinsic mechanical properties of the different types of PMMA. PMMA-D is more compliant than PMMA-C, as suggested by the manufacturer's specifications. [24,28] Moreover, the fracture surfaces of the APF samples showed a characteristic brittle fracture and no significant differences between the different treatments were visible (Figure 3f ).
In the FFF-manufactured PMMA-D group, statistically significant differences were observed in Charpy impact strength among differentially treated samples for both unnotched (χ 2 (2) ¼ 11.180; p ¼ 0.004) and notched specimens (F(2) ¼ 4.165; p ¼ 0.027) (Figure 4a). Pairwise comparisons showed that the impact strength was significantly lower in the washed group in comparison with the untreated controls (p ¼ 0.003). Interestingly washed þ sterilized specimens were comparable to the controls (p ¼ 0.143). In the notched specimens, however, the impact strength values of the washed samples were comparable to the controls (p ¼ 0.120), while washedþsterilized samples showed slightly but significantly higher impact strength in comparison with the control group (p ¼ 0.027).
www.advancedsciencenews.com www.aem-journal.com In the APF-manufactured PMMA-C group, an influence of cleaning treatment on the impact strength was not observed in the Charpy impact tests with unnotched specimens (χ 2 (2) ¼ 1.257; p ¼ 0.533). Note that higher impact strength values were observed in the washing þ sterilization group with notched specimens, but the differences were statistically not significant (χ 2 (2)¼5.886; p ¼ 0.053) (Figure 4b).
The fracture surfaces did not show any major differences between the treatment groups for either type of material ( Figure 4c). All samples showed characteristic brittle fracture surfaces. In FFF samples, the layer next to the build platform was wider and denser most properly due to the heated print bed. This could also be attributable to the calibration of the nozzle distance to the surface of the platform. Calibration done at ambient temperature does not take the heat-induced expansion in the print bed, which causes the first layer to be deposited by the nozzle from a smaller distance resulting in a wider first layer. [29] The individual layers, in addition, could easily be distinguished in FFF, but not in APF samples.
The lower Charpy impact strength observed in unnotched, FFF-printed PMMA-D samples after washing (Figure 4a) could not be explained by a difference in the fracture surface assessed with light microscopy (Figure 4c). Therefore, it was checked whether the observed differences could be associated with variabilities in the porosity level of the specimens. A Spearman's rank correlation coefficient test with all groups pooled (with or without a cleaning treatment) revealed that neither flexural stress at the conventional deflection nor Charpy impact strength showed a statistically significant correlation to the global porosity or to the local porosity levels at the midsection ( Figure 5). A significant correlation, moreover, was not observed when the treatment groups were separately analyzed (data not shown).
Given that there was a skewed data distribution, the focus was placed on comparison of the FFF-manufactured PMMA-D   www.advancedsciencenews.com www.aem-journal.com values evaluated by the Charpy impact test were quite comparable (Figure 6b), suggesting that the observed porosity levels did not have any detectable influence on the mechanical performance. Next, the porosity levels of the whole samples (global porosity) of randomly selected samples with or without a cleaning procedure were compared (Figure 7). The global porosity in the FFF-printed PMMA-D samples was comparable in all three groups (χ 2 (2) ¼ 2.880; p ¼ 0.237). In APF-printed samples, however, there was a statistically significant difference in the global porosity levels (χ 2 (2) ¼ 8.060; p ¼ 0.018). Pairwise comparison showed significantly different values between the washed and washed þ sterilized samples (p ¼ 0.027). Nevertheless, these differences did not show a detectable effect on the mechanical performance of the specimens (Figure 3 and 4b).
Next, focus was placed on the analyses of PMMA-D to investigate the decreased impact strength observed in FFF-manufactured PMMA-D specimens (Figure 4a), which was explained neither by a difference in porosity nor by a difference in fracture surface. For this purpose, tensile tests on thin compression-molded samples were conducted to check whether the cleaning treatments differentially influence the PMMA-D material itself. No differences in the stress-strain curves among the samples from different treatment groups were observed (Figure 8a). Moreover, the curves show pronounced yielding where the forces decrease after reaching a maximum while the deformation still increases which was accompanied with stress whitening (CM specimens were transparent before testing and opaque along the parallel length afterward). Stress whitening occurs as a color change at macroscale, which is caused by the formation of microvoids between polymer chains during deformation. Further increasing the deformation results in an opening of the voids and therefore microcracks and crazes, causing a dispersion of visible light. [30] The alignment of the polymer chains in the direction of the load leads to a strengthening of the material and thus to a further increase in the force, which is known as cold drawing (Figure 8a). Statistically significant  www.advancedsciencenews.com www.aem-journal.com decreases in the yield stress (σ y ) were observed among different groups (χ 2 (2) ¼ 7.340; p ¼ 0.025) (Figure 8b). Pairwise comparisons revealed significantly lower yield stress in washed þ sterilized samples (p ¼ 0.04) compared with the untreated controls. The differences between the washed and untreated controls, however, were not significant (p ¼ 0.085). The fracture surfaces after tensile testing, on the other hand, were indifferent under a light microscope (Figure 8c). FTIR spectra for FFF as well as CM samples after the different treatments are shown in Figure 9. No major change in the chemical structure of the material or media uptake was observed, as the bands seem to be unaffected. The evaluated spectra are in accordance to literature showing all characteristic bands [31] : 1) the α-methyl, ester-methyl, and methylene C-H stretching (3100-2800 cm À1 ) and bending (1500-1350 cm À1 ) modes; 2) the C═O stretching mode (1728 cm À1 ); 3) the ester group stretching vibrations or coupled C-O and antisymmetric C-C-O stretch as well as skeletal vibrations coupled to C-H deformations in the range of 1350-1100 cm À1 ; 4) the methylene rocking mode at 843 cm À1 ; and 5) the vibrations of the ester group, possibly the C-O-C symmetric stretching mode at 827 and 809 cm À1 .
Seeing that attenuated total reflection measurements only characterize the first few micrometers of a material, and no traces of any different media were found in the spectra, it appears unlikely that the used agents penetrate the material at all.
In a final step, SEM images of representative fracture surfaces of unnotched PMMA-D FFF Charpy samples were compared for Figure 8. a) Stress-strain curves and b) yield stress (σ y ) obtained for compression-molded PMMA-D in its untreated state, after washing and after washingþsterilization. In c) microscopic images of the representative fracture surfaces for each group after tensile testing are displayed. The statistical significance of the differences was assessed using Kruskal-Wallis H test followed by pairwise analysis with Bonferroni correction. n ¼ 5 per group. Stars show extreme outliers. Figure 9. FTIR spectra for PMMA-D manufactured by FFF and CM in its untreated state, after washing, and after washing þ sterilization. n ¼ 3 per group. One representative curve for each group is shown. In a) the spectra are vertically shifted against each other to better identify the individual spectra. In b) the spectra are shown as measured with the characteristic bands.
www.advancedsciencenews.com www.aem-journal.com each treatment step ( Figure 10). The images were taken at a predefined border region of the sample, as mainly the outer regions should be affected by the treatments. It is observed that the fracture surfaces of the untreated and washed samples look alike (Figure 10a,b), while the fracture surface appears smoother after sterilization (Figure 10c). This finding does not match the Charpy impact strength results in Figure 4a, where only washed samples have significantly lower values compared with the other treatment steps. Nevertheless, the slightly different fracture surfaces should not be overestimated, as during the SEM procedure the sample is only recorded very locally. Similar brittle fracture surfaces were found for FFF PMMA in previous works. [5,32] Moreover, it is not yet clear whether the differences found are related to the treatment or the print quality. Therefore, the future implementation of a study on the reproducibility of the printing process is recommended.

Discussion
Although PMMA-based materials have long been used for medical products such as implants, it seems that no study has yet been conducted analyzing the effects of a combination of washing and formaldehyde sterilization on the mechanical properties. Münker et al. [10] analyzed the effect of different sterilization methods (ethylene oxide, hydrogen peroxide gas plasma treatments, autoclave sterilization, and γ-irradiation) on the mechanical properties of PMMA-based materials. Whilst autoclave sterilization is not a choice for thermolabile materials like PMMA, the other three methods seem to be suitable candidates, with γ-irradiation resulting in increased flexural strength. Yavuz et al. [9] investigated the influence of sterilization via supercritical carbon dioxide, ultraviolet, heat, ethylene oxide, and hydrogen peroxide on the chemical structure and surface morphology of PMMA microchips. The chemical techniques slightly affected the surface roughness and channel profile. This effect was even more dominant for ultraviolet sterilization. On the other hand, opaque structures were observed after heat sterilization. Sharifi et al. [12] showed that electron beam sterilization of PMMA with the right energy dose only slightly affects the chemical, mechanical, and optical properties as well as biocompatibility. As stated by McKeen, [23] the mechanical properties, such as elongation at break and notched Izod impact, of CYROLITE compounds do no significantly decrease after gamma radiation at exposures of up to 7.5 Mrad or electron beam exposures up to 7.5 kGy. Moreover, by ethylene oxide sterilization, no significant change in key properties or yellowing takes place. Steam sterilization and dry heat sterilization, on the other hand, are not recommended.
None of the previous studies included the preceding hygiene process, which works with elevated temperatures, pressures, and the addition of a washing agent. Given that PMMA has hygroscopic characteristics, [19,20] both treatment steps could influence the material itself, but also the structure created by the AM process. Therefore, this study deals with the characterization of the effects of a predefined washing and sterilization routine on the porosity and mechanical properties of additively manufactured parts. Two different PMMA-based materials (PMMA-D: 3Diakon and PMMA-C: CYROLITE MD H12), each optimized for a different AM method (FFF and APF), were analyzed. Washing and formaldehyde sterilization did not influence the flexural modulus at the bending tests in FFF-or APF-printed specimens. In the Charpy impact tests with unnotched specimens, however, remarkably lower impact strengths were observed in the washed group, although the differences were significant only in the FFF-printed PMMA-D samples. As a confounding influence of differential porosity levels was strictly ruled out, the effect was attributable to the cleaning procedure. Samples that were sterilized after washing, however, did not show such a difference, suggesting that the sterilization procedure neutralized the effect introduced by washing. Although, material aging was shown to be induced by prolonged (12-24 months) immersion in water, [20] it may well be accelerated by water exposure at higher temperatures (60°C in washing process) followed by water desorption at 80°C and 60°C. However, whether this absorption/desorption stress is responsible for reduced impact strength in washed samples is questionable on the grounds that such an effect was not observed when the washing protocol was followed by a sterilization procedure. One major difference between the washing and sterilization procedure is that drying phases after sterilization were conducted under vacuum (alternating between 90 and 210 mbar compared with %1 bar after washing), which presumably results in a better desorption, particularly considering the porous structure of the printed samples. The fact that such an effect was not detected in impact tests with notched samples indicated that the influence, if any, should rather be effective at a limited depth. Nevertheless, Figure 10. SEM images of PMMA-D in FFF in its a) untreated state, b) after washing, and c) after washing þ sterilization. A predefined border region at the side of the build platform was analyzed.
www.advancedsciencenews.com www.aem-journal.com given that the impact tests were conducted after >48 h of storage at standardized climate (23°C, 50% r.h.), associating the reduction in impact strength to the residual moisture at the time of testing would be highly unlikely. Comparability of flexural moduli among the groups and identical peak profiles of FTIR spectra further supported this presumption. Regardless of the evaluation method (global or local), porosity values of less than 1% were measured for all samples. These values are very low in terms of the FFF process, considering that values of up to about 6% have been reported in the literature. [21,33] In general, the porosity of a printed part is strongly influenced by the material used and the processing parameters, such as the nozzle temperature, build platform temperature, and printing speed. [34,35] Varying printing qualities were observed in our samples among different batches, but also within one batch. The batch-wise difference could be the result of a minimal lower nozzle temperature, which leads to worsened diffusion between adjacent layers, larger voids, and therefore lower mechanical properties. [35] Differences within the batch are mostly attributable to the uneven temperature distribution on the print bed. [36] In addition, the interdiffusion depth decreases and thus the pore size increases the further away the layer is from the print bed. [34] Several researchers analyzed the effect of different pore sizes and porosity values on the mechanical properties of printed parts. In general, they found increasing mechanical properties with decreasing porosity values. [37][38][39][40] However, the analyzed porosity values in these studies deviated from each other by more than 5%, whereas the porosity values of all samples in this study are less than 1% (on average %0.18% for PMMA-D in FFF and 0.07% for PMMA-C in APF). In a previous study, conducted with another type of PMMA, a porosity value of 0.09% could be obtained for FFF-printed samples. [32] The porosity values from this and the previous study are both very low and the results indicated that the observed porosity levels were below the critical point and did not influence the mechanical properties.
Tensile tests on thin compression-molded samples, where the material is distributed more homogenously and the thickness of the specimens allows for better identification of surface influences through the treatments, indicated that the influence of cleaning treatments on the obtained stress-strain curves was negligible, apart from a slight but significant reduction in the yield stress. FTIR spectroscopy ruled out any detectable change in the chemical structure as well as presence of residual media in both printed and compression-molded material after washing and formaldehyde sterilization procedures. These results complement previous studies, in which sterilization by ethylene oxide, UV, heat, CO 2 , or hydrogen peroxide treatments also did not result in any major change in the chemical structure. [9] Comparison of the fracture surfaces in predefined border areas of the samples allows the detection of the presence of diffused media, as media-induced changes in fracture behavior are often accompanied by changes in the fracture surface. [41] However, scanning electron microscope analyses in our study did not show detectable differences among differentially treated samples, which can explain differences observed in mechanical tests. Nonetheless, variances in interlayer strength or diffusion depth, which can be caused by slight temperature fluctuations, [35] cannot be evaluated with the applied methods. To maintain a proper environmental temperature and thus maximize the temperature homogeneity in the printed parts, printers with closed chambers are to be preferred. At this point, it remains unclear whether the reason for changes in the Charpy impact strength lies in the treatment step or varying printing quality. Therefore, it is recommended to perform a reproducibility study of the printing process in the future.

Conclusion
Each sterilization method has advantages and disadvantages. As there is no method specified for a certain combination of material and process, the influence of washing and sterilization must be thoroughly analyzed for each new material-process combination before use.
In this study, the influence of formaldehyde sterilization and the preceding washing procedure did not show a significant influence on the bending properties of two different PMMAbased materials, each optimized for a different AM method (FFF and APF). However, significantly lower Charpy impact strengths were observed after washing in the FFF-printed samples. Any confounding influence of variabilities in specimen porosity was excluded. Observed porosity levels (less than 1%) were not found to have any correlation to the mechanical performance. Therefore, the effect was attributable to the cleaning procedure, but was neutralized after sterilization.
No reason was found not to use the applied sterilization routine for the analyzed PMMA-based materials. However, further tests should be conducted, such as the performance of cytotoxicity tests and repeatability/reproducibility tests of the printing process, before using this routine prior to real application.