Urethane-acrylate-based photo-inks for digital light processing of flexible materials

Urethane–acrylate-based photo-inks containing various concentrations (0.1–1.5 wt.%) of two photo-initiators, namely ethyl phenyl(2,4,6-trimethylbenzoyl)phosphinate (TPOL) or diphenyl(2,4,6-trimethylbenzoyl)phosphine oxide (BPO), for digital light processing (DLP) were developed. According to photo-DSC kinetics investigations, no significant difference was detected between the photo-activity of formulations containing BPO or TPOL at various concentrations. BPO (1.0 wt.%) with a high molar extinction coefficient (500 L/mol·cm at 365 nm) resulted in higher controllability on the layer thickness (100 µm) during the 3D printing process. The surface cracks that appeared during the post-curing process could be avoided by splitting the exposure time (5 min) into short intervals (5 × 1 min) without affecting double bond conversion (DBC). Several flexible objects were successfully 3D printed in good quality and their thermomechanical properties and layer-by-layer morphology were investigated.


Introduction
3D printing technology, called also additive manufacturing (AM), is a process of joining materials layer-by-layer to produce complex objects without molding or machining [1]. The photo-polymerization-based 3D printing technologies such as stereolithography (SLA), digital light processing (DLP), liquid crystal precision (LCP), and digital light synthesis (DLS) or continuous liquid interface production (CLIP) enable the fabrication of customized/complex multifunctional 3D objects with controlled chemical and mechanical properties [2,3]. In the DLP technology, a light pattern is projected to a photo-curable ink (photo-ink) to solidify it in the X-/Y-directions with a defined depth and attach it to a building stage. In the next step, The building stage is lifted up/down in the Z-direction by a defined height allowing the liquid photo-ink to wet the surface of the solidified layer. The photo-curing depth (C d ) of the photoink is slightly larger than the defined thickness layer (Δz) to ensure proper chemical/physical adherence between the layers [4][5][6]. Therefore, C d determines the Δz limits and consequently the Z-resolution for the DLP process, which is in the range of 10-200 µm [5].
The 3D printing photo-inks are formulations of monomers/oligomers and photo-initiators, which are converted from liquid to solid under exposure to light with a specific wavelength. The irradiation wavelength and intensity of the 3D printer light source play an important role in formulating the photo-inks, especially in choosing photo-initiators. The type and concentration of photo-initiators are adjusted with the 3D printing parameters to ensure a high 3D printing quality.
Mono(acyl)phosphine and bis(acyl)phosphine oxides are commonly used as photo-initiators in photo-inks due to their superior characteristics. Upon light irradiation, an α-cleavage reaction occurs for mono(acyl)phosphine oxides under triplet-state to produce benzoyl and phosphinoyl radicals (Scheme 1a). For bis(acyl)phosphine oxides, the α-cleavage generates benzoyl and phosphinoyl oxide radicals (Scheme 1b). The addition of the phosphinoyl oxide radical to the monomers leads to a macro-mono(acyl) phosphine oxide, which can undergo a second α-cleavage reaction to form another benzoyl and macro-phosphinoyl radicals (Scheme 1c) [7,8].

Instruments
UV-Vis spectroscopy of monomers and photo-initiator solutions in acetonitrile (0.8 g/L) was done using a PerkinElmer instrument (Lambda 950, USA) operating in the range of 200-700 cm −1 . Acetonitrile was used as a blank solution.
A Netzsch machine (DSC 204 F1 Phoenix, Germany) equipped with a spot UV curing system (Lumen Dynamics, OmniCure ® S2000, Canada) was employed to evaluate the photo-curing of the formulations through differential scanning photo-calorimetry (photo-DSC). The experiments were done in an isothermal mode at 25 °C with a uniform UV intensity of 0.1 mW/cm 2 under an N 2 atmosphere (40 mL/min). The irradiation step was prolonged for 3 min to fully photo-cure the samples and reach a plateau region, which was employed as a baseline for the peak integrations. The reported values are an average of at least three measurements.
The thermal transitions of monomers were evaluated by differential scanning calorimetry (DSC, Netzsch, DSC 204 F1 Phoenix, Germany) in the range of -100 to 100 °C at a heating rate of 10 °C/min under an N 2 atmosphere (40 mL/ min). The glass transition temperatures (T g ) were extracted from the middle point of the baseline change in the first heating cycle. The crystallization (T c ) and melting (T m ) temperatures were extracted from the onset of the corresponding peaks.

(b) (c)
A Thermo Fisher Scientific spectrometer (Nicolet™ iS20) equipped with an attenuated total reflection (ATR) unit (PIKE Technologies, GladiATR™) was used to record the Fourier-transform infrared (FTIR) spectra of the photoinks and 3D printed samples in the range of 4000-400 cm −1 at a 4 cm −1 resolution with 32 scans.
The layer-by-layer structure of the 3D printed samples was studied by scanning electron microscopy (SEM, Zeiss, GeminiSEM 300, Germany) at a standard accelerating voltage of 5 kV. An Everhart-Thornley detector (SE2) and an InLens detector (IL) were used as secondary electron detectors. Samples were coated with gold before microscopy.
A Zwick/Roell machine (Z020, Germany) was employed for investigating the tensile properties of the 3D printed bone-shaped specimens (ASTM D638 Type IV) according to ISO 527-1:2019. The load cell was 10 KN, the clamping length was 34 mm, the pre-load was 0.05 N, and the crosshead speed was 19 mm/min. The reported values are an average of at least three specimens for each sample.
Thermogravimetric analysis (TGA) was done on a Mettler Toledo instrument (TGA 2). The heating rate was 10 °C/min from 20 to 600 °C under N 2 flow at a rate of 50 mL/min.

Methods
The photo-curing rate (R p , 1/s) and double bond conversion (DBC, %) as a function of time were calculated from photo-DSC data using Eqs. where ΔH theor (J/g) is the theoretical total heat released during the complete polymerization of the monomers within the sample, f is the mass fractions of each monomer within the sample, n is the number of double bonds in each monomer, ΔH 0 (J/mol) is the standard heat of polymerization for methacrylate (54.8 kJ/mol [9]) or acrylate (86.2 kJ/mol [9]), M w (g/mol) is the molecular weight of each monomer, dH p /dt (J/s·g or W/g) is the normalized heat flow per second, and ΔH p (J/g) is the heat released from the start of photo-curing up to a certain time obtained from the integration of DSC thermograms.
The photo-curing depth of formulations was determined on a DLP 3D printer. For this purpose, each formulation was dropped on a glass slide ( Fig. S1 in Supplementary Information, SI) and exposed to different UV (365 nm) intensities and exposure times. Then, the remaining liquid photo-ink was wiped off so that only a solid photo-cured layer was left on the glass slide. The thickness of the layer was determined by a digital external micrometer.

Statistical analysis
The results expressed as mean ± SD are representing at least three independent experiments. The differences between groups were analyzed using the one-way ANOVA and LSD test for multiple comparisons (p < 0.05: significant difference).

UV-Vis absorption
Optimizing the photo-initiator within photo-inks is essential for the 3D printing application. Many parameters influence the performance of a photo-initiator such as dissociation quantum yield, light absorption in terms of molar extinction coefficient and absorption wavelength, rate constant for the addition of the generated radicals to monomers, and side reactions like oxygen quenching [11]. The chemical structures of TPOL and BPO are depicted in Scheme 1. TPOL is a liquid mono(acyl)phosphine oxide with high solubility that can be mixed easily with most photo-inks, while BPO is solid bis(acyl)phosphine oxide with low solubility in photoinks, especially at high concentrations, e.g. 4 wt.% [12]. The dissociation quantum yield (Φ) for TPOL and BPO are 0.3 [13] and 0.6 [8], respectively. The photo-initiator molecules absorb the UV light, convert to their excited state, and generate free radicals. The UV-Vis spectra of TPOL and BPO were recorded in acetonitrile solution (0.8 g/L, Fig. 1a). TPOL showed absorption in the range of 200-420 nm with two maximums at 273 and 371 nm. BPO showed absorption in the range of 200-440 nm with two maximums at 295 and 369 nm. The second absorptions at 371 and 369 nm for TPOL and BPO, respectively, make them photo-active for the commercial DLP 3D printers mainly functioning at 365 nm. Meanwhile, both UrDMA and UrA monomers acetonitrile solution (0.8 g/L) were transparent in the region of 280-700 nm, which makes the whole irradiated light accessible for the photo-initiator molecules during the photo-curing process.
where ε is the molar extinction coefficient, c is the molar concentration of the solution, and l is the path length. The molar extinction coefficient as a function of wavelength was calculated for photo-initiators (Fig. 1b). BPO showed a larger molar extinction coefficient than TPOL, especially in the wavelength region of 320-440 cm −1 . A larger molar extinction coefficient for the photo-ink corresponds to a low light penetration depth and allows a more controllable 3D printing process and avoids over-cure as much as possible [15]. Thus, BPO is expected to provide higher photo-activity for the photo-inks compared with TPOL due to its enhanced dissociation quantum yield (0.6 [8]) and molar extinction coefficient (500 L/mol·cm at 365 nm). However, low solubility in the photo-inks is the main drawback of BPO as it is impossible to fully dissolve it at high concentrations, e.g. 4 wt.% [12].

Photo-curing
UrDMA as a viscous di-functional monomer (η = 9500 mPa·s) was used as a crosslinker in photo-inks to improve the dimensional stability of the 3D printed objects by generating a thermoset network structure [9,16]. DSC analysis showed that UrDMA is a completely amorphous monomer with a glass transition temperature (T g ) of -33 °C and no melting point (Fig. S2 in SI). This is due to the existence of two enantiomers for UrDMA4, which prevents the packing and crystallization of UrDMA molecules (Scheme 2). Mono-functional UrA (η = 35 mPa·s) was employed as a reactive diluent to decrease the viscosity of photo-inks while resulting in flexibility for the 3D printed objects [9,17]. According to DSC results, UrA is a semi-crystalline monomer with T g of -81 °C and a melting point (T m ) of 5 °C (Fig. S3 in SI).
As mentioned, due to higher dissociation quantum yield and molar extinction coefficient values, BPO could provide higher photo-activity for the photo-inks compared with TPOL. Meanwhile, BPO as a bis(acyl)phosphine could be more effective due to the generation of four radicals compared with TPOL as a mono(acyl)phosphine oxide leading to two radicals.
Photo-DSC has been widely utilized to evaluate the photo-activity of 3D printing photo-inks [9]. The photopolymerization kinetics of formulations based on UrDMA/ UrA (40/60, wt./wt.) containing TPOL or BPO (0.5, 1.0, or 1.5 wt.% of total monomers' mass) was studied in the isothermal mode at 25 °C (Figs. S4 and S5 in SI). For all formulations, the curve of R p versus photo-curing time showed an initial increase and a late decrease, known as the auto-acceleration and auto-deceleration phenomena, respectively, which is due to the gradual increase of viscosity with conversion and subsequent decrease of molecular mobility for monomers and macroradicals within the photo-curing mixture over time [9]. At low viscosity, R p is constant and dependent on the double bonds' activity (chemistry-controlled). At middle viscosity (up to conversions of 11-13%), when the coupling of macroradicals for the termination is diffusion-limited but the monomers are still mobile, R p increases (auto-acceleration). At high viscosity, when the photo-curing mixture is transforming into a rubbery/glassy network (gel point) and the diffusion of monomers is significantly restricted, R p decreases (autodeceleration) [9].
The influence of the photo-initiator type and concentration on the photo-polymerization kinetic parameters such as the maximum photo-curing rate (R p,max ), the time to reach R p,max known as gel point time (t GP ), the conversion observed at R p,max known as gel point conversion (DBC GP ), and total double bound conversion (DBC total ) ( Fig. 2 and Table S1 in SI). Both TPOL-and BPO-based formulations displayed fast photo-curing and high conversion. Increasing the photoinitiator concentration from 0.1 to 1.0 wt.% resulted in a partial improvement in R p and further increasing to 1.5 wt.% led to a partial reduction in R p (Fig. 2a), while the other kinetic parameters (t GP , DBC GP , and DBC total ) were not significantly changed (p > 0.05). On reason for lower R p values at high photo-initiator concentration (1.5 wt.% compared to 1.0 wt.%) can be due to the major photon absorption at the top of the sample and limited photon penetration into the depth of the sample. TPOL-based formulations displayed partially higher R p , earlier gel point (lower t GP ), and higher conversion (DBC GP and DBC total ) compared with the corresponding BPO-based formulations. It is in opposition to the expectation obtained from the lower dissociation quantum yield (0.3 [13]) and molar extinction coefficient (145 L/ mol·cm at 365 nm) for TPOL. One reason can be the lower molecular weight of TPOL (316.38 g/mol) compared to BPO (418.17 g/mol) resulting in higher molar percentages at the same weight percentages. For example, formulation TPOL 1.0% contains 0.87 mol.% of photo-initiator, while formulation BPO 1.0% has 0.66 mol.% of photo-initiator.

3D printing
For 3D printing of high-quality objects, good chemical/ physical adherence between the interfaces of the cured layers is essential, which requires a C d higher than Δz (C d > Δz). Photo-curing a layer of photo-ink up to the gel point does not provide sufficient stiffness to allow the continuous layerby-layer joining process [5]. As a result, DBC at the layer interface should be slightly higher than DBC GP . Meanwhile, very high C d can cause overexposure (C d ≫ Δz) and unintended photo-curing of space-filling features, e.g. porous inclusions [5]. Furthermore, if C d = Δz, the layer-by-layer joining process can fail due to poor adherence between the sequentially photo-cured layers [5].
C d can be regulated by changing the photo-initiator concentration, irradiation intensity, and exposure time. At low photo-initiator concentration, a minor fraction of photons are being absorbed, thus C d is high but DBC is low. On contrary, high photo-initiator concentration results in major photon absorption, hence C d is low but DBC is high. Each formulation was dropped on a glass slide (Fig. S1 in SI) and photo-cured on a DLP 3D printer under different UV (365 nm) intensities and exposure times (Fig. 3). For each formulation, employing higher irradiation intensity and longer exposure time increased C d . When the photoinitiator concentration, irradiation intensity, and exposure time were not adequate to reach the gel point, the formulations did not photo-cured and C d was zero (BPO 0.1% and TPOL 1.0%). The formulation with 1 wt.% of TPOL presented higher C d compared to the formulation with 1 wt.% of BPO, especially at higher irradiation intensities and longer exposure times, which can be attributed to the lower molar extinction coefficient of TPOL (145 L/ mol·cm at 365 nm) compared to that of BPO (500 L/ mol·cm at 365 nm). As expected, the formulations based on BPO with a large molar extinction coefficient (500 L/ mol·cm at 365 nm) allow a more controllable photo-curing process. C d was improved by increasing the BPO concentration from 0.1 to 0.5 wt.% and then decreased at 1.0 and 1.5 wt.%. BPO molecules can absorb the UV light and limit the light penetration depth into the formulations, thus the photo-curing only occurs near the surface of the formulations (BPO 1.0% and BPO 1.5%) [18]. The formulation with 1 wt.% of BPO was selected as the optimum formulation for 3D printing the objects. For 3D printing with a layer thickness of Δz = 100 µm, the optimal UV (365 nm) intensity and exposure time were determined as 3.72 mW/cm 2 and 3.0 s, respectively. The shrinkage of the formulation in the course of the 3D printing process was determined as 0.81% for the X-direction and 0.91% for the Y-direction. Several complex CAD models were designed and successfully 3D printed (Fig. 4). The 3D printed objects were flexible and in good shape.

Post-curing
Post-curing the 3D printed objects can raise the DBC and improve their mechanical strength, dimensional stability, and biocompatibility. Both thermal and UV treatments have been used commonly as the post-curing processes. UV treatment is through polymerizing the uncured double bounds under a broad-band UV irradiation. The 3D printed objects were washed twice with fresh isopropanol and post-cured under a high-intensity UV lamp for different exposure times. In the course of post-curing for 5 min, the 3D printed objects got stiffer and their light yellow color disappeared due to the complete decomposition of photo-initiator molecules, while several cracks appeared on their surface (Fig. 5a). During the post-curing process, the double bonds further polymerized resulting in additional shrinkage and consequently mechanical internal stress within the 3D printed objects, which could not be released immediately and caused fractures, especially on their surfaces [19][20][21]. At the same time, the temperature of the 3D printed objects raised sharply under the highintensity UV lamp, which led to the rapid evaporation of the absorbed isopropanol during the washing step. To avoid surface cracks, the total exposure time of 5 min was split into 5 × 1 min, while the UV lamp was turned off for 2 min between each irradiation sequence. This routing led to the continuous stiffening of the 3D printed objects and the gradual disappearance of their light yellow color, while no surface cracks were observed (Fig. 5a).

Surface morphology
SEM images for the cross-section of the 3D printed objects exposed striped features indicating the individual layers formed during the continuous layer-by-layer 3D printing process (Fig. 6). The distance between the strips was approximately 100 μm, equal to the predetermined Δz value for each layer. Moreover, the interface between the layers was observed as intensive and contrasting vertical stripes.

Mechanical properties
Changing the ratio of multi-functional crosslinker to monofunctional reactive diluents in photo-inks and altering the average double bound functionality, can affect the mechanical properties of the 3D printed objects [9]. Here, the ratio of UrDMA/UrA in the formulations was changed from 40/60 to 30/70 and 20/80, and the mechanical properties of the corresponding 3D printed bone-shaped samples were studied by determining tensile modulus (E), maximum tensile (σ max ), and maximum elongation (ε max ), at ambient temperature derived from stress-strain curves (Fig. 7). All samples exposed the typical stress-strain curves of the thermoset polymer, a linear elastic region followed by a failure point without any yield point (Fig. 7a) [26][27][28]. As expected, increasing the UrA content from 60 to 80% resulted in a decrease in E and σ max values (Fig. 7b-c) and an increase in ε max values (Fig. 7d) [9]. Reducing the average double bound functionality in photo-inks leads to a fall in crosslinking density for the corresponding polymeric network. Similar observations were previously reported about photo-cured mono-and di-functional urethane-acrylates [9,29].

Thermal stability
The thermal stability of the 3D printed objects based on UrDMA/UrA mixtures (40/60, 30/70, and 20/80) was evaluated by TGA (Fig. 8). All samples underwent a two-step thermal degradation profile with peak temperatures of 340-350 and 410-430 °C (Fig. 8a-b). The first weight loss was due to the thermal degradation of urethane bonds generating primary amine and olefin or secondary amine and carbon dioxide [26,30]. The second weight loss was related to the thermal degradation of the aliphatic hydrocarbon moieties [26,[30][31][32]. The onset of thermal degradation (T 95% , the temperature at which 5% weight loss took place) was higher for the 3D printed objects based on the UrDMA/UrA mixture of 40/60 (266 °C) due to their higher crosslinking density (Fig. 8c). The char yield (rest mass at 600 °C) was higher for the 3D printed objects based on the UrDMA/UrA mixture of 20/80 (Fig. 8d). This phenomenon was due to the formation of nitrogenous char generated through the decomposition of urethane moieties. The content of urethane moieties for the UrDMA/ UrA mixture (40/60, 30/70, and 20/80) was 4.49, 4.53, and 4.57 mmol/g, respectively. Thus, the 3D printed objects based on the UrDMA/UrA mixture of 20/80 with higher urethane content displayed higher char yield (3.2% at 600 °C).

Conclusions
Although kinetics investigations based on photo-DSC results did not show any significant difference between the photo-activity of formulations containing BPO or TPOL at various concentrations, BPO with a high molar extinction coefficient and thus a lower light penetration depth for the formulations resulted in higher controllability on the layer thickness during the 3D printing process. For the desired layer thickness of 100 µm, employing BPO with a concentration of 1.0 wt.% was optimal. The appearance of surface cracks during the post-curing process could be dissolved by splitting the exposure time (5 min) into short (5 × 1 min) intervals. By changing the ratio of UrDMA/ UrA, a range of mechanical properties, e.g. stiffness, flexibility, and hardness, is achievable. Increasing the UrA content decreased E and σ max for the 3D printed objects, where the UrDMA/UrA mixture of 30/70 exhibited the highest ε max value.

Acknowledgements
The authors acknowledge the financial support by the Federal Ministry of Education and Research of Germany in the framework of "ProMatLeben -Polymere" (project number 13XP5087E, PolyKARD).
Funding Open Access funding enabled and organized by Projekt DEAL.

Data availability
The raw/processed data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of interest
There are no conflicts to declare.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.