Open Access
Add to BibSonomy
Abstract
Femtosecond laser-based two-photon polymerization is becoming increasingly popular in biofabrication. One of the key selling points of the technology is the possibility to use a variety of different materials to produce biology-oriented structures, for instance, liver cell regeneration. These include hybrid materials, lithographic resins, and hydrogels to name a few. However, while these materials are investigated separately, there is a severe lack of studies dedicated to directly comparing them in terms of structurability. Therefore, in this work, popular pre-polymers such as SZ2080, SU8, and GelMA are compared side by side in this manner. They are photosensitized using photoinitiators Irgacure 369, Irgacure 2959, and LAP. Structurability is tested using two different popular wavelengths - 800 nm and 515 nm. Acquired differences are subsequently partially explained by two-photon absorption measurement, giving insights into the efficiency of the photopolymerization process. Finally, biocompatibility is compared showing surprisingly small differences between all the tested materials.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Regenerative medicine is an expansive field inching ever closer and closer to widespread use [1]. To accommodate the needs of this field variety of manufacturing technologies were employed, including additive manufacturing [2]. Optical 3D printing stands out among other additive manufacturing techniques as a relatively simple and contactless fabrication method with the possibility to use a vast array of different materials [3]. This is further expanded if femtosecond (fs) lasers are employed, allowing to perform two-photon polymerization (TPP) thus expanding printing possibilities even further [4]. As a result, it was used in biomedical applications to fabricate structures that would allow investigating single-cell behavior [5], cultivate cells [6], or produce structures for direct implantation to living organisms [7].
While current progress is impressive, a lot of questions still need to be answered before this could proceed towards widespread approval and adoption in clinical practice. The primary one, in the light of regenerative medicine, is the relation between structurability and biocompatibility of materials used. Indeed, a multitude of different materials were tested so far, but in each case, biocompatibility tests $ex vivo$ were performed by proprietary protocols. The proliferation of TPP, especially at a commercial system level, also resulted in different lasers being used for fabrication. This makes direct comparison of biocompatibility and structurability extremely difficult. At the same time, some polymers, for instance, hybrid organic-inorganic ones [8], are preferred from a printing standpoint due to easy handling and good post-printing properties. In contrast, hydrogels or other bioinspired materials would mimic living organisms better, but structurability and post-processing properties might be lacking [9,10]. This creates a challenge in choosing appropriate materials, as there is no direct comparison between them from both structurability and biocompatibility standpoints. This is especially true for liver-related research. Indeed, there is a substantial amount of liver diseases that result in the need for transplantation [11]. As a result, numerous works aimed at Liver-on-Chip [12,13] or liver reconstruction experiments [14,15] were performed. Nevertheless, the field is still in its infancy. Primary requirements formulated so far are related to printing resolution, allowing adequate capillary and membrane features to be integrated during manufacturing, biocompatible materials, and printing throughput high enough to print structures in the mm-cm range. Luckily, most of these parameters should be achievable with TPP after adequate optimization of throughput and materials is performed.
As a result, this work aims at alleviating issues related to structurability and biocompatibility of various TPP compatible materials by directly comparing the biocompatibility of several distinct and popular photopolymers and photoinitiators (PI). These include popular in lithography SU8, which can be used in TPP fabrication [16]. Also, various sub-types of hybrid organic-inorganic photopolymers SZ2080 [17] are tested, as they provide easy structurability, good post-fabrication mechanical properties [18], and the possibility to perform photocrosslinking without PI [19]. The latter is very important in optical applications due to the increased laser-induced damage threshold (LIDT) [20], as well as reduced fluorescent background during bio-testing. Hydrogel Gelatin methacryloyl (GelMA) is also tested, as it is attracting more and more attention from the scientific community due to its potential in biomedical applications related to it’s generally superb biocompatibility [21,22] and biomechanical properties [23,24]. Also, some pilot works showed it to be compatible with high-resolution TPP 3D printing [25]. The two later materials are also mixed with several different PIs. The first one is popular in TPP 2-benzyl-2-dimethylamino-1-(4-morpholino phenyl)-butanone-1 (Irgacure 369) [26]. Another one is 2-hydroxy-4’-(2-hydroxyethoxy)-2-methylpropio phenone (Irgacure 2959), which was for a long time considered a go-to PI for bio-applications [27]. Finally, a novel bio-structuring oriented PI lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) is also tested, as it is hailed as one of the most promising PIs for regenerative medicine [28]. Comparison is done in terms of fabrication windows, two-photon absorption (TPA) cross-section, and biocompatibility. Such direct comparison is used to uncover peculiarities (like structure expansion/shrinkage during processing) of TPP material usage chosen for regenerative medicine-oriented structure fabrication. All this work is oriented toward further investigation in liver-oriented regenerative medicine research.
2. Materials and methods
2.1 Materials
A variety of materials were used in this work for testing (Fig. 1). Prepolymers were hybrid organic-inorganic resist SZ2080 from IESL-FORTH and GelMA from Sigma-Aldrich. To make them photoactive additional photoinitiators purchased from Sigma-Aldrich were used: Irgacure 369, Irgacure 2959, and LAP (Fig. 2). This resulted in such material combinations: SZ2080 with 1wt% of Irgacure 369, Irgacure 2959, or LAP and GelMA with 1wt% of Irgacure 2959 or LAP (these two past combinations were prepared by dissolving materials in deionized water at 60$^{\circ }$C under magnetic stirring). Moreover, SU-8 was tested as well and since it is standalone commercial material, was not modified in any manner. Samples were prepared for TPP experiments $via$ drop-casting materials on the cover glass substrates. Prior to the 3D laser structuring, SZ2080 based samples were heated at 65 $^{\circ }$C for 3h, SU-8 samples were ramped for 15 minutes to 65 $^{\circ }$C and kept at this temperature for 30 minutes, then ramped for 15 minutes to 90 $^{\circ }$C and kept at this temperature for 45 minutes. Since GelMA based combinations tend to gel while storing at 4 $^{\circ }$C, they were liquified by pre-heating them up to 60 $^{\circ }$C. After drop-casting liquified GelMA variations onto the cover glass substrates followed their gelation at 4 $^{\circ }$C. Development of laser-printed samples was carried out in 4-methyl-2-pentanone for material combinations based on SZ2080, in deionized water for material combinations based on GelMA and PGMEA for SU8.
For bio-testing, 2 mm thick 20 mm diameter round glass samples were used and materials were spin-coated on them. The spin coater used was from "Specialty Coating Systems, Inc." (model P6700). The position of the glass substrate was fixed using vacuum and the process of spin-coating proceeded in three steps varying in time and speed of spinning: 100 rpm for 10 s (during this step 5 – 6 drops of the prepared mixture was put onto the spinning substrate), 500 rpm for 10s (during this step the material dripped was evenly distributed on the glass substrate) and 1500 rpm for 30 s (during this step the excess of material was eliminated and the unhardened coating was produced). To obtain hard coatings the process of thermal polymerization was conducted. The SZ2080 based coatings were kept at 200 $^{\circ }$C for 10 minutes, SU-8 coatings – at 250 $^{\circ }$C for 15 minutes, GelMA based coatings – at 100 $^{\circ }$C for 10 minutes.
2.2 Fabrication setup
3DLL fabrication was performed using a "Laser Nanofactory" (Femtika) setup. Key concepts and components of the system are discussed in [29]. While the standard system of this kind uses amplified Yb:KGW laser source and first two harmonics (1030 nm and 515 nm), additional setup variation with 800 nm 100 fs 100 MHz oscillator (Menlo Systems) was also applied. It has no significant differences from a standard system apart optical chain designed to support 800 nm wavelength. In all cases, 3DPoli control software was used for the control of the systems.
2.3 TPA measurment
Nonlinear absorption of photoinitiators and photoinitiator/prepolymer mixtures was measured using a home-built setup shown in Fig. 3. Femtosecond laser pulses were produced by a standard amplified Ti:Sapphire laser system (Coherent Libra-USP-HE) and used either directly (measurements at 800 nm), or after converting them to 515 nm in an optical parametric amplifier (Light Conversion Topas-800-fs). The pulses were suitably attenuated and spectrally cleaned using a set of high-reflectance mirrors M1 and M2 and loosely focused onto the sample using a telescope made from lenses L1 and L2 and iris diaphragm Iris1. The focal distance of the telescope was 1200 mm, resulting in a focal spot of 130 $\mu$m with the Rayleigh distance exceeding 40 mm, ensuring that the beam diameter was approximately constant throughout the entire length of the sample (1 cm).
Fig. 3. The setup used for nonlinear absorption measurements. M1, M2 – mirrors, L1, L2 – lenses, VNDF – variable neutral density filter, Iris1 – iris diaphragm, BS1, BS2 – UVFS wedge beamsplitters, FM1 – flipping mirror, PD1, PD2 - photodiodes.
The energy of the pulses could be varied over approximately two orders of magnitude using an attenuator consisting of reflective circular variable neutral density filter VNDF (Thorlabs NDC-50-2) mounted on a motorized rotation stage. After the attenuator, part of the beam was sampled by a beamsplitter BS1 reflecting a small fraction of the beam to a reference photodiode PD1. The remaining part of the beam passed the sample (solution of photoinitiator and/or prepolymer in UV fused silica cell with the optical path length of 10 mm) and was split again by the beamsplitter BS2 to measure the relative pulse energy after the sample and the absolute average power using a power meter (Ophir Juno/PD300). Beam profile was recorded using a CMOS camera (FLIR Chameleon-CM3) placed instead of the sample. The switchover between the sample and beam profiler was motorized, with both sample and camera mounted on a motorized translation stage moving perpendicular to the beam propagation direction. This enabled measuring the beam profile before each nonlinear absorption measurement. Pulse duration measurements were performed using a home-built-scanning autocorrelator with non-collinear second harmonic generation in 0.01 mm long BBO crystal. The beam could be directed between autocorrelation and nonlinear absorption measurement using a manual flipping mirror FM1.
The measurement procedure consisted of several steps. First, the beam profile and pulse autocorrelation function (ACF) was recorded. Then, a baseline scan was performed by varying the pulse energy and recording PD1, PD2, and power meter values without the sample in order to calibrate the incident power and evaluate the linearity of the PD1/PD2 signal, which was typically constant within 0.5% over the entire power range. Subsequently, the sample was inserted and the PD1 and PD2 signals were recorded as a function of the VNDF angle. For comparison and the evaluation of measurement sensitivity, the transmittance data was also recorded on a neat solvent. 50 repetitions of each measurement were performed for each data point used for calculations.
2.4 Biocompatibility testing procedure
Material discs were rinsed with sterile aqua dest, wiped with a disinfectant cloth, and shortly submerged in 70% Ethanol for initial disinfection. For further use, discs were placed into a sterile 12-well cell culture plate (Costar, Corning, NY, USA) and UV irradiated for 20 min. The sterilized discs were then used for biocompatibility testing by using the human liver HCC cell line HepG2(GS) and the human hepatoma cell line PLC/PRF/5. For maintenance, cells were kept under standard conditions (37 $^{\circ }$C, 5% CO$_2$ in humified atmosphere) in MEM media (Gibco, Thermo Fisher Scientific, Vienna, Austria) containing 10% fetal bovine serum (GE Healthcare Life Sciences, UT, USA) and 1% Penicillin/Streptomycin (Sigma Aldrich, Vienna, Austria). The medium was renewed every other day while re-seeding was done upon 80-90% confluence.
For biocompatibility testing experiments 1x10$^4$ cells/cm$^2$ (HepG2(GS)) or 1x10$^3$ cells/cm$^2$ (PLC/PRF/5) were seeded on each material disc in maintenance media. For comparison to standard conditions, cells were seeded in 12-well plates without material discs at the same conditions. The cells were kept under standard conditions for 7 or 14 days and cell proliferation was determined by means of the commercially available WST-1 reagent (Roche, Vienna, Austria). Optical density at 450 nm and 620 nm as a reference was determined after 30 min of incubation with WST-1 reagent by a Spectrostar microplate reader (BMG labtech, Ortenberg, Germany). The experiment was repeated two times. Proliferation data were calculated as a relative proliferation of respective standard conditions. Methodology used - for a statistical analysis: SPPS, version 26. For group comparison, Kruskal-Wallis and Bonferroni tests were used. In addition, a PostHoc Test was used for pairwise comparison.
3. Results
3.1 Laser structurability of biopolymers
We began our work by testing fabrication windows of all chosen materials. We used the previously reported methodology, of forming 3D structures (size - 25 x 25 x 25 $\mu$m) with embedded woodpiles, which allows us to judge structurability in both micro- and nanoscale [19]. Distance between structures - 100 $\mu$m. Because medical manufacturing requires relatively big structures with some leniency in regards to achievable resolution, a 20x 0.8 NA objective was used for testing. We considered the fabrication window to be an area in translation velocity ($v$) and average laser power ($P$) plot where a 3D structure survived the development process. In this particular work embedded woodpiles did not play into this consideration but were judged as an additional feature showing how easy is to print using a given material.
Results of such experiments performed with 800 nm radiation are given in Fig. 4 SZ2080 with Irgacure 369 and LAP photoinitiators, as well as SU8 without post bake, performed the best, having most of the array surviving whole manufacturing procedure. The latter result is especially interesting, because SU8, due to its polymerization pathway, normally requires post-bake [30]. Nevertheless, due to energy deposition using fs lasers with a high repetition rate ($\sim$MHz range and more) the whole polymerization process can be realized using just exposure stage [16]. If post bake is applied, structures expand, especially in the Z direction, resulting in surviving, yet heavily deformed objects. SZ2080 with Irgacure 2959 and both GelMA combinations performed the worst in the 800 nm case. GelMA result can be attributed to the material being a soft hydrogel, which is susceptible to deformations during the development procedure. However, SZ2080 and Irgacure 2959 combination is rather interesting, as SZ2080 normally has quite good mechanical strength and a wide fabrication window. One of the experimental observations made during working with this material was seemingly spontaneous overexposure defects occurring during fabrication even with seemingly correct exposure parameters. They would then form weak points in structures, resulting in them breaking. The possibility of such defects occurring mainly depended on fabrication time - the longer the fabrication, the more likely they would occur. One possible explanation is the potential inhomogeneity of materials and nanoclusters of photoinitiator, more sensitive than the overall material. As a result, damage probability between 0 and 1 is extended between large $P$ interval. Such behavior was noticed before in SZ2080 when testing for LIDT [20]. Considering that Irgacure 2959 might have dissolved a bit harder in SZ2080, combined with a higher repetition rate of the laser, it might have been the cause of the result.
Fig. 4. Fabrication windows of tested materials when 800 nm radiation is used. SZ2080 with Irgacure 369 and LAP photoinitiators as well as SU8 without post-bake performed the best. GelMA, due to weak mechanical properties, had the worst performance, with basically no structures surviving when Irgacure 2959 is used. Single structure size 25 x 25 x 25 $\mu$m, distance between them - 100 $\mu$m.
Results with 515 nm fs pulses [Fig. 5] have both some similarities and differences from the 800 nm case. First, GelMA again performed poorly. However, this time difference between both photoinitiators was smaller. At the same time, SU8 with post bake again resulted in overgrown structures, showing severe over-polymerization. Forgoing post-bake helped, but the improvement was substantially lesser than with 800 nm. Finally, all SZ2080 variations performed better. One of the possible reasons – two orders of magnitude (1 Mhz vs 100 MHz) lower repetition rate with 515 nm. Then, the aforementioned nanoclusters are excited less by accumulative processes, resulting in no quasi-random defects during printing. Therefore, if SZ2080 structures are needed 515 nm radiation is preferable, while SU8 works better with an 800 nm laser. GelMA, in both cases, is extremely tricky to work with. Overall, these results prove that laser choice is very important when choosing photopolymers for fabrication.
Fig. 5. Fabrication windows of tested materials when 515 nm radiation is used. All SZ2080 variations performed relatively well. GelMA, due to weak mechanical properties, had substantially worse performance than SZ2080. SU8 structures survived but were severely deformed due to over-polymerization, which was only slightly reduced when no post bake was used. Single structure size 25 x 25 x 25 $\mu$m, distance between them - 100 $\mu$m.
Apart from the structurability in general, provided results also point to some interesting peculiarities regarding laser printing itself. For instance, SEM images of SU8 made at different $P$ levels show tendency for the polymerization reaction to expand significantly (Fig. 6). While general ouline of structure is visible at lower power, whole structure becomes continuous at higher $P$, expanding both to inner part of the model and outside of it. This can be tied to potentialy very efficient reaction of the polymerization as well as localized heating during process. Both 800 nm and 515 nm radiation induces similar effect. Therefore, while SU8 is viable for 3D printing with fs laser, fabrication window have to be chosen very carefully, as even slight overexposure makes obtained objects loose their 3D shape. This also means that high-resolution nanoprinting, at least in such experimental conditions, is substantially more limited in comparison to other tested materials.
Fig. 6. (a) Model and dimensions of 3D structure used in resolution array experiment. (b) to (d) expansion of SU8 structure beyond model as the $P$ is increased during fabrication. Provided images were acquired with 800 nm laser, but general tendency can be observed with 515 nm also.
Another noteworthy observation concerns the shrinkage of the produced structures. It is well known that it depends on the pre-polymer material [9,17] and presence of different developers [31]. However, during experimentation another variable influencing structure shrinkage was observed. Looking at SZ2080 with IRG369 and LAP PIs severe difference in structure distortion can be noted (Fig. 7). Indeed, just by changing the photoinitiator and keeping material the same difference in overall size of the structure is $\Delta$ $S$ = $S_{model}$ - $S_{measured}$ = 25 - 17.7 = 7.3 $\mu$m, where $S$ is the length of the side of the structure. Therefore, it means that shrinkage can go from few % in the case of IRG (basically unmeasurable using the SEM), to $\sim$29.2% when LAP is applied. Thus, PI choice influences not only structurability and biocompatibility, but also considerably changes the properties of produced structures.
Fig. 7. (a) - Single resolution array structure made out of SZ2080 with IRG369 PI. Dimensions of model used for fabrication are overlayed on top. No major deviations in the structure size can be observed. (b) - Same structure produced out of SZ2080 with LAP. Severe deformation due to the shrinkage is present, with side length of the structure being just 17.7 $\mu$m instead of 25 $\mu$m in model, showing shrinkage of around $\sim$29.2%.
3.2 Biopolymer characterization
To understand the underlying reason for the acquired structurability results, an in-depth analysis of materials was carried out. TPA was chosen for this investigation. TPA in itself is a relatively underexplored area of TPP and is seldom used in recent research related to new material and photoinitiator development. At the same time, it can be considered that TPA does not paint a full picture during direct laser writing, as other processes, such as avalanche ionization and thermal accumulation can play a role, especially if longer (hundreds of fs) laser pulses are used [32]. Albeit it is still being argued in the literature to which extent each process contributes to polymerization [33]. Regardless, it is universally agreed that TPA should play a substantially important role in the printing process. Thus, by measuring it is possible to get some insights into underlying physical phenomena.
To evaluate TPA first let’s introduce its calculation formalism. The intensity of light $I$ after passing the sample exhibiting (only) nonlinear absorption is:
where $I_0$ is the incident intensity, $\beta$ is the nonlinear absorption coefficient, and $z$ is the sample path length. Therefore, the inverse transmittance 1/$T$ of the sample can be written as The left side of Eq. (2) is a relative dimensionless number measured directly by photodiodes PD1 and PD2. The value of PD1 is multiplied by the baseline factor recorded without the sample to result in unity transmission without the sample. To evaluate nonlinear absorption of the sample $\beta$ $z$, one needs to know the incident intensity, which, in the case of our measurement is a function of time and transverse spatial coordinates, i.e. $I_0$ = $I_0(x,y,t)$, where $t$ is time. The $t$ dependence is evaluated by approximating the measured ACF with a Gaussian curve and assuming that the pulse shape is also Gaussian with a width that is times less than that of the ACF (Fig. 8). As Fig. 8(a) indicates, the shape of the ACF was very close to Gaussian.Fig. 8. (a) - Measured and fitted ACF curves. (b) - experimentally recorded beam profile for 515 nm beam. (c) - Integrals of spatial profile across x and y direction.
The estimated pulse shape is then normalized to have an integral equal to unity and multiplied with an experimentally recorded beam profile. The integral of which is also normalized to unity. This results in a spatiotemporal intensity distribution $I_0$ = $I_0(x,y,t)$, which is multiplied by the pulse energy and used as the value of intensity to calculate the probability of photon absorption accounting for the spatiotemporal shape of the pulse using Eq 2. The resulting change in transmitted pulse energy is then summed over the temporal envelope of the pulse and spatial profile of the beam [Fig. 8(b)] to be able to calculate the total pulse energy loss due to nonlinear absorption, which directly comparable to the experimental data. The value used for calculating the model data is varied to match the slopes of measured and modeled curves (Fig. 9). The error margins of two-photon absorption measurements (Fig. 9) were estimated by performing identical measurements on the known sample (solvent) without two-photon absorption at wavelengths being investigated under identical conditions. The error was estimated as a standard deviation of 50 repetitions of the measurement.
Fig. 9. (a) - Calculated (green) and measured (red) nonlinear absorption dependence on incident pulse energy. The sample was methanol solution of rhodamine 6G. For comparison, data measured in neat methanol is shown (black). (b) - the same measurement of Irgacure-369 photoinitiator performed using 650 nm light.
If the concentration of absorbers is known, the determined value of (which is numerically equal to for 1 cm path length) can be recalculated into two-photon absorption cross-section expressed in Göppert-Mayer units using the relationship
where $E$ is the photon energy and $N$ is the number of absorber molecules per cm$^2$. Nevertheless, care should be taken when using this formalism. If $N$ can be relatively well defined for photoinitiator molecules within an organic solution, defining it for multi-component materials like pre-polymer is non-trivial. Therefore, in this work $\sigma ^{(2)}$ was only calculated for photoinitiator solutions, leaving $\beta$ as a representative value for pre-polymers.To ensure that procedure is trustworthy, they were validated. The experiment on Rhodamine 6G dye in methanol (Fig. 9) yielded the value of $\sigma ^{(2)}$ = 84 GM at 800 nm, which is a close match to the value of 79 GM published in a study by Reguardati et al. [34]. The sensitivity of the setup allows reliably detecting values above 1.5 cm$^2$/W, and the absolute error of estimated values above this detection threshold is of the order of 20%, arising from the combination errors in power measurement, measurement noise, and uncertainties related to beam profile and ACF width. An experiment using 650 nm excitation and Irgacure-369 photoinitiator dissolved in isopropanol at 1% wt yielded the value of $\sigma ^{(2)}$ =12 GM, which is within the range of values (7$\div$22) estimated using two different methods reported by Schaffer et al. [26]. Therefore, measurement techniques can be considered reliable.
Acquired results are presented in Fig. 10. Unsurprisingly, there is very good agreement between materials that performed the worst in terms of structurability and have the lowest $\beta$ (2 $\times$ 10$^{-14}$ cm$^2$/W), namely GelMA +1 wt% LAP at 800 nm. At the same time, SU8 has a very high $\beta$ at 515 nm, which, at the same time, proved to be a lot worse structurability vise than SU8 at 800 nm. This can be explained by the overexaggerated reaction at 515 (due to high $beta$), which results in severe reaction broadening. Thus, one can argue, that too high of a value of $\beta$ is also detrimental. One potential way to combat it would be the usage of quenchers, which would reduce reaction efficiency to manageable levels [35], yet this needs to be tested in subsequent works. At the same TPA, results should be interpreted with caution. While SZ2080 with both Irgacure 2959 and LAP has very similar $\beta$ for 800 nm radiation, their fabrication windows are substantially different, favoring LAP. Nevertheless, general results show that $\beta$ in order of magnitude of 10$^{-12}$-10$^{-13}$ cm$^2$/W is more-less ideal range (at least from the standpoint of TPA), where structurability can be achieved relatively consistently, with the main issue becoming mechanical properties and shrinkage of the structure.
Fig. 10. (a) and (b) shows $\beta _z$ and $\sigma$ of used photoinitiators respecively. Rhodamine 6G is used as control. (c) - $\beta _z$ of pre-polymers used in the work.
3.3 Biocompatibility
One of the ultimate goals of TPP-based bioprinting is to have as large as possible material selection. Then, the material can be chosen by the necessity of the exact application, would it be specialized scaffolds [5], lab-on-chip (LOC) [36] or organ-on-chip (OOC) device [37]. While structurability is very important from the printing perspective, good biocompatibility is the most important metric for usage in any biological environment. Therefore, all previously discussed materials were tested for their biocompatibility.
Fig. 11. Proliferation of human liver cell lines HepG2(GS) (a) and PLC/P (b) on different materials. 7d: Cells were grown on tested materials and standard conditions for 7 days, followed by proliferation determination. 14d: Cells were grown on tested materials and standard conditions for 14 days, followed by proliferation determination. No extreme differences were noticed comparing different materials and cells.
We used flat, UV structured polymer films on glass substrates for biocompatibility experiments. Such protocol was chosen to avoid any structure-influenced cell response [38]. As seen from Fig. 11 all material performed in a very similar manner (within margins of errors). Also, two different cell types - HepG2(GS) and PLC/PRF/5 were tested to see if they would react differently to used materials. As seen from Fig. 11 all tested materials performed in a relatively similar fashion, falling in between biocompatibility range between $\sim$0.5 to $\sim$1 range in comparison to glass substrate control. There was no statistically significant difference between the materials. To specify, neither with HepG2 nor with PLC at any timepoint of testing. Thus, it means, that fundamentally within tested parameters all the materials are biocompatible and could, in theory, be used for biofabrication interchangeably. This gives a lot of freedom, as hybrid materials could be used for devices needing mechanical strength, and hydrogels – for scaffolds that need to mimic living tissue. On the other hand, as seen from structurability experiments, at least for GelMA possibility to produce true 3D structures is a lot more complicated than SZ2080 variations. Thus, if bio-mimicking properties of hydrogels are not needed, there is no reason not to use material with better structurability.
4. Discussion
One of the key prerequisites for sustained expansion and proliferation of regenerative medicine is fabrication technologies capable of producing structures suitable for envisioned applications. 3D printing, and additive manufacturing in general, are also hailed as potentially disruptive in numerous fields [3], especially transplantation surgery. It brings even more flexibility and design freedom for desired structures. Also, due to the nature of additive manufacturing, the pipeline from idea to structure is extremely minimized, facilitating the potential adaptability of the field towards point-of-care individualized approaches. However, most current fabrication techniques have processing resolution from tens of $\mu$m to mm. Therefore, TPP and fs laser manufacturing, in general, is the next logical step in the development of regenerative medicine. It can fabricate structures with feature sizes in the true sub-$\mu$m range at the same time allowing overall sizes in mm [39]. Furthermore, it can be paired with other microfabrication techniques – both based on fs-laser [36,40] and not [41]. That way the capabilities of the approach are even more expanded, bringing the full technological portfolio to the table.
Nevertheless, some key questions in further fs structuring development for regenerative medicine remain. The field brings some contradicting requirements for the manufacturing techniques. On the one hand, resolution/feature size capabilities have to be preserved to get the full flexibility in choosing the 3D geometry of the final structure. At the same time structures have to be big enough for realistic applications, i.e. in the size range of mm to cm. All of this was shown to be compatible with current TPP capabilities [39]. However, most of them were achieved using hardware developments, such as continuous writing [29]. Furthermore, hardware-based approaches for improving direct laser writing are far from saturated. On-the-fly tunable resolution $via$ beam diameter [42], multi-focus [43], or layer-by-layer printing [44] are just a few examples of suggested ways to increase the capabilities of TPP. Then, usage of acousto–optical deflectors [45] for even higher throughput or SLM for feature size/aspect ratio control [46] are also very attractive in the long run. Thus, from the hardware side, TPP has massive room for improvement to fully match any requirements regenerative medicine might have in its development path.
But what about the material side? A supposedly vast array of TPP compatible materials is one of the reasons why this is so revered in the field [4]. Indeed, materials such as proteins [47,48], elastomers [32,49], hydrogels [50,51], and other bio-inspired polymers [52,53] were tried through the years. And while academic-level research resulted in a multitude of papers, one of the key problems remains their structurability. Some of the materials, for instance, can sustain only very limited translation velocities [49]. As shown in this work, this can be tied to low TPA and, subsequently, slow radical generation reaction requiring consistent energy deposition. This puts a hard upper limit to the possible throughput. To remedy it, multi-focal polymerization might be applied, but then hardware solutions are merely used to compensate for material drawbacks. Then, some of the problems are even more fundamental. For instance, all TPP processable materials shrink during manufacturing procedures. While some materials are very promising for bio-applications by being both soft and biodegradable, their shrinkage might be as high as 14.5% [9]. Then, severe pre-compensation needs to be done on the software level before fabrication. While this was shown to work in some extreme cases, like severe shrinkage during ceramic production from laser processed hybrid polymers [54], it is not convenient. These problems are compounded by the desire to bring laser processing to medical professionals who are not printing experts. Thus, there is no surprise that special, easy-to-process materials, such as SZ2080 [17], were created for TPP. Furthermore, this materials was shown time and time again to be biocompatible in vitro [6] and in vivo [7]. Furthermore, results acquired in this work show that, in general, when compared under identical conditions, SZ2080 is as biocompatible as other bio-oriented materials. Keeping in mind that its fabrication window is consistently among the biggest from all tested materials, and it shrinks minimally, it begs the question, is it worth using complicated, hard-to-work biomaterials if the gain in biocompatibility is negligible. At the same time, no one can deny the importance of hydrogels and especially bio-absorbable materials for regenerative medicine. Yet a lot of work needs to be done on the material development side before these polymers would turn from limiters to enablers in TPP.
5. Conclusions
In this work, we presented a combined study directly comparing TPP structurability of different pre-polymers, such as SZ2080, SU8, and GelMA. We showed that depending on photoinitiator wavelength of the laser plays important role in the processability of pre-polymer. SZ2080 photosensitized with either 1% wt of Irgacure 369 or LAP showed the best results with both tested wavelengths (800 nm and 515 nm). At the same time, mechanically weak hydrogel GelMA consistently proved itself to be hard to work with the material, requiring special care to acquire true 3D structures. Additionally, it was shown that SU8 can be structured at high translation velocities (up to 1 cm/s) with both wavelengths. Interestingly, better results were acquired without post-bake, pointing to combined photochemical and thermal effects taking place during laser exposure. Further TPA measurements gave even more insights into acquired results. Very poor structurability was tied with low $\beta$ values - ar the order of magnitude of 10$^{-14}$ cm$^2$/W. At the same time, if $\beta$ exceeded 10$^{-11}$ cm$^2$/W laser formed structures tended to overgrow, showing excessive polymerization reaction. Overall, consistently best results were acquired at $\beta$ values in the range of 10$^{-12}$-10$^{-13}$ cm$^2$/W. It was also discovered that different PIs can heavily influence shape retention of the material. SZ2080 with IRG369 had minimal shrinkage, while SZ2080 with LAP was shown to be prone to shrinking up to $\sim$29.2%. Finally, biocompatibility testing resulted in no serve differences between all investigated materials, allowing us to consider them interchangeable in this regard. Therefore, from an application standpoint, a material with better structurability can be chosen if the application requires it without risking of losing any biological functionality. At the same time, if a very specialized environment needs to be recreated, for instance liver, additional research and understanding of material-cell interaction might be needed.
Funding
European Commission (01.2.2-MITA-K-702-10-0006).
Acknowledgments
E.E. performed all TPP experiments and their evaluation, prepared samples for biocompatibility testing, worked on data acquisition and analysis; M.V. and K.G. performed and interpreted TPA experiments; G.M. provided insights into biopolymer peculiarities; B.L. performed biocompatibility testing; R.G. performed SEM measurements. P.S. and P.S. supervised biocompatibility experiments; S. Š. helped with questions regarding material chemistry; A.K.G and K.S. supervised biocompatibility sample logistics and general communication; L.J. formulated a general plan for the study, supervised all experiments and parties involved, evaluated and processed data, and wrote the manuscript. Principal Investigators in the project: L.J. - laser material processing side, K.S. - medical side (shared seniority). All authors contributed to the final version of the manuscript.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. C. Mason and P. Dunnill, “A brief definition of regenerative medicine,” Regen. Med. 3(1), 1–5 (2008). [CrossRef]
2. M. Ramola, V. Yadav, and R. Jain, “On the adoption of additive manufacturing in healthcare: a literature review,” J. Manuf. Technol. Manag. 30(1), 48–69 (2019). [CrossRef]
3. L. Jonušauskas, S. Juodkazis, and M. Malinauskas, “Optical 3D printing: bridging the gaps in the mesoscale,” J. Opt. 20(5), 053001 (2018). [CrossRef]
4. G. Merkininkaitė, D. Gailevičius, S. Šakirzanovas, and L. Jonušauskas, “Polymers for regenerative medicine structures made via multiphoton 3D lithography,” Int. J. Polym. Sci. 2019, 1–23 (2019). [CrossRef]
5. B. Richter, V. Hahn, S. Bertels, T. K. Claus, M. Wegener, G. Delaittre, C. Barner-Kowollik, and M. Bastmeyer, “Guiding cell attachment in 3D microscaffolds selectively functionalized with two distinct adhesion proteins,” Adv. Mater. 29(5), 1604342 (2017). [CrossRef]
6. D. Ricci, M. M. Nava, T. Zandrini, G. Cerullo, M. T. Raimondi, and R. Osellame, “Scaling-up techniques for the nanofabrication of cell culture substrates via two-photon polymerization for industrial-scale expansion of stem cells,” Materials 10(1), 66 (2017). [CrossRef]
7. J. Mačiulaitis, M. Deveikytė, S. Rekštytė, M. Bratchikov, A. Darinskas, A. Šimbelytė, G. Daunoras, A. Laurinavičienė, A. Laurinavičius, R. Gudas, M. Malinauskas, and R. Mačiulaitis, “Preclinical study of SZ2080 material 3D microstructured scaffolds for cartilage tissue engineering made by femtosecond direct laser writing lithography,” Biofabrication 7(1), 015015 (2015). [CrossRef]
8. M. Farsari, M. Vamvakaki, and B. N. Chichkov, “Multiphoton polymerization of hybrid materials,” J. Opt. 12(12), 124001 (2010). [CrossRef]
9. S. Pashneh-Tala, R. Owen, H. Bahmaee, S. Rekštytė, M. Malinauskas, and F. Claeyssens, “Synthesis, characterization and 3D micro-structuring via 2-photon polymerization of poly(glycerol sebacate)-methacrylate - an elastomeric degradable polymer,” Front. Phys. 6, 41 (2018). [CrossRef]
10. K. Parkatzidis, M. Chatzinikolaidou, M. Kaliva, A. Bakopoulou, M. Farsari, and M. Vamvakaki, “Multiphoton 3D printing of biopolymer-based hydrogels,” ACS Biomater. Sci. Eng. 5(11), 6161–6170 (2019). [CrossRef]
11. R. Adam, V. Karam, V. Cailliez, J. G. O. Grady, D. Mirza, D. Cherqui, J. Klempnauer, M. Salizzoni, J. Pratschke, N. Jamieson, E. Hidalgo, A. Paul, R. L. Andujar, J. Lerut, L. Fisher, K. Boudjema, C. Fondevila, O. Soubrane, P. Bachellier, A. D. Pinna, G. Berlakovich, W. Bennet, M. Pinzani, P. Schemmer, K. Zieniewicz, C. J. Romero, P. D. Simone, B.-G. Ericzon, S. Schneeberger, S. J. Wigmore, J. F. Prous, M. Colledan, R. J. Porte, S. Yilmaz, D. Azoulay, J. Pirenne, P.-D. Line, P. Trunecka, F. Navarro, A. V. Lopez, L. D. Carlis, S. R. Pena, E. Kochs, and C. Duvoux, “2018 annual report of the european liver transplant registry (ELTR) - 50-year evolution of liver transplantation,” Transpl. Int. 31(12), 1293–1317 (2018). [CrossRef]
12. D. Y. No, K.-H. Lee, J. Lee, and S.-H. Lee, “3D liver models on a microplatform: well-defined culture, engineering of liver tissue and liver-on-a-chip,” Lab Chip 15(19), 3822–3837 (2015). [CrossRef]
13. A. Ehrlich, D. Duche, G. Ouedraogo, and Y. Nahmias, “Challenges and opportunities in the design of liver-on-chip microdevices,” Annu. Rev. Biomed. Eng. 21(1), 219–239 (2019). [CrossRef]
14. P. L. Lewis and R. N. Shah, “3D printing for liver tissue engineering: Current approaches and future challenges,” Curr. Transplant. Rep. 3(1), 100–108 (2016). [CrossRef]
15. J.-Z. Wang, N.-Y. Xiong, L.-Z. Zhao, J.-T. Hu, D.-C. Kong, and J.-Y. Yuan, “Review fantastic medical implications of 3D-printing in liver surgeries, liver regeneration, liver transplantation and drug hepatotoxicity testing: A review,” Int. J. Surg. Open 56, 1–6 (2018). [CrossRef]
16. K. K. Seet, S. Juodkazis, V. Jarutis, and H. Misawa, “Feature-size reduction of photopolymerized structures by femtosecond optical curing of SU-8,” Appl. Phys. Lett. 89(2), 024106 (2006). [CrossRef]
17. A. Ovsianikov, J. Viertl, B. Chichkov, M. Oubaha, B. MacCraith, I. Sakellari, A. Giakoumaki, D. Gray, M. Vamvakaki, M. Farsari, and C. Fotakis, “Ultra-low shrinkage hybrid photosensitive material for two-photon polymerization microfabrication,” ACS Nano 2(11), 2257–2262 (2008). [CrossRef]
18. L. Pertoldi, V. Zega, C. Comi, and R. Osellame, “Dynamic mechanical characterization of two-photon-polymerized SZ2080 photoresist,” J. Appl. Phys. 128(17), 175102 (2020). [CrossRef]
19. L. Jonušauskas, D. Gailevičius, L. Mikoliūnaitė, D. Sakalauskas, S. Šakirzanovas, S. Juodkazis, and M. Malinauskas, “Optically clear and resilient free-form µ-optics 3D-printed via ultrafast laser lithography,” Materials 10(1), 12 (2017). [CrossRef]
20. A. Butkutė, L. Čekanavičius, G. Rimšelis, D. Gailevičius, V. Mizeikis, A. Melninkaitis, T. Baldacchini, L. Jonušauskas, and M. Malinauskas, “Optical damage thresholds of microstructures made by laser three-dimensional nanolithography,” Opt. Lett. 45(1), 13–16 (2020). [CrossRef]
21. J. Hoorick, L. Tytgat, A. Dobos, H. Ottevaere, J. Erps, H. Thienpont, A. Ovsianikov, P. Dubruel, and S. Vlierberghe, “(Photo-)crosslinkable gelatin derivatives for biofabrication applications,” Acta Biomater. 97, 46–73 (2019). [CrossRef]
22. Z. Wu, S. Xie, Y. Kang, X. Shan, Q. Li, and Z. Cai, “Biocompatibility evaluation of a 3D-bioprinted alginate-GelMA-bacteria nanocellulose (BNC) scaffold laden with oriented-growth RSC96 cells,” Mater. Sci. Eng., C 129, 112393 (2021). [CrossRef]
23. K. Yue, G. T. de Santiago, M. M. Alvarez, A. Tamayol, N. Annabi, and A. Khademhosseini, “Synthesis, properties, and biomedical applications of gelatin methacryloyl (GelMA) hydrogels,” Biomaterials 73, 254–271 (2015). [CrossRef]
24. S. Krishnamoorthy, B. Noorani, and C. Xu, “Effects of encapsulated cells on the physical-mechanical properties and microstructure of gelatin methacrylate hydrogels,” Int. J. Mol. Sci. 20(20), 5061 (2019). [CrossRef]
25. M. Sun, X. Sun, Z. Wang, S. Guo, G. Yu, and H. Yang, “Synthesis and properties of gelatin methacryloyl (GelMA) hydrogels and their recent applications in load-bearing tissue,” Polymers 10(11), 1290 (2018). [CrossRef]
26. K. J. Schafer, J. M. Hales, M. Balu, K. D. Belfield, E. W. Stryland, and D. J. Hagan, “Two-photon absorption cross-sections of common photoinitiators,” J. Photochem. Photobiol., A 162(2-3), 497–502 (2004). [CrossRef]
27. A. Ovsianikov, A. Deiwick, S. Vlierberghe, M. Pflaum, M. Wilhelmi, P. Dubruel, and B. Chichkov, “Laser fabrication of 3D gelatin scaffolds for the generation of bioartificial tissues,” Materials 4(1), 288–299 (2011). [CrossRef]
28. G. Melilli, I. Carmagnola, C. Tonda-Turo, F. Pirri, G. Ciardelli, M. Sangermano, M. Hakkarainen, and A. Chiappone, “DLP 3D printing meets lignocellulosic biopolymers: Carboxymethyl cellulose inks for 3D biocompatible hydrogels,” Polymers 12(8), 1655 (2020). [CrossRef]
29. L. Jonušauskas, T. Baravykas, D. Andrijec, T. Gadišauskas, and V. Purlys, “Stitchless support-free 3D printing of free-form micromechanical structures with feature size on-demand,” Sci. Rep. 9(1), 17533 (2019). [CrossRef]
30. A. Campo and C. Greiner, “SU-8: a photoresist for high-aspect-ratio and 3D submicron lithography,” J. Micromech. Microeng. 17(6), R81–R95 (2007). [CrossRef]
31. S. Rekštytė, D. Paipulas, M. Malinauskas, and V. Mizeikis, “Microactuation and sensing using reversible deformations of laser-written polymeric structures,” Nanotechnology 28(12), 124001 (2017). [CrossRef]
32. R. Buividas, S. Rekštytė, M. Malinauskas, and S. Juodkazis, “Nano-groove and 3D fabrication by controlled avalanche using femtosecond laser pulses,” Opt. Mater. Express 3(10), 1674–1686 (2013). [CrossRef]
33. M. Malinauskas, A. Žukauskas, G. Bičkauskaitė, R. Gadonas, and S. Juodkazis, “Mechanisms of three-dimensional structuring of photo-polymers by tightly focussed femtosecond laser pulses,” Opt. Express 18(10), 10209–10221 (2010). [CrossRef]
34. S. Reguardati, J. Pahapill, A. Mikhailov, Y. Stepanenko, and A. Rebane, “High-accuracy reference standards for two-photon absorption in the 680-1050 nm wavelength range,” Opt. Express 24(8), 9053–9066 (2016). [CrossRef]
35. I. Sakellari, E. Kabouraki, D. Gray, V. Purlys, C. Fotakis, A. Pikulin, N. Bityurin, M. Vamvakaki, and M. Farsari, “Diffusion-assisted high-resolution direct femtosecond laser writing,” ACS Nano 6(3), 2302–2311 (2012). [CrossRef]
36. D. Andrijec, D. Andriukaitis, R. Vargalis, T. Baravykas, T. Drevinskas, O. Kornyšova, A. Butkuė, V. Kaškonienė, M. Stankevičius, H. Gricius, A. Jagelavičius, A. Maruška, and L. Jonušauskas, “Hybrid additive-subtractive femtosecond 3D manufacturing of nanofilter-based microfluidic separator,” Appl. Phys. A 127(10), 781 (2021). [CrossRef]
37. L. Bakhchova, L. Jonušauskas, D. Andrijec, M. Kurachkina, T. Baravykas, A. Eremin, and U. Steinmann, “Femtosecond laser-based integration of nano-membranes into organ-on-a-chip systems,” Materials 13(14), 3076 (2020). [CrossRef]
38. H. Jeon, C. G. Simon, and G. Kim, “A mini-review: Cell response to microscale, nanoscale, and hierarchical patterning of surface structure,” J. Biomed. Mater. Res. 102(7), 1580–1594 (2014). [CrossRef]
39. L. Jonušauskas, D. Gailevičius, S. Rekštytė, T. Baldacchini, S. Juodkazis, and M. Malinauskas, “Mesoscale laser 3D printing,” Opt. Express 27(11), 15205–15221 (2019). [CrossRef]
40. D. Wu, L.-G. Niu, S.-Z. Wu, J. Xu, K. Midorikawa, and K. Sugioka, “Ship-in-a-bottle femtosecond laser integration of optofluidic microlens arrays with center-pass units enabling coupling-free parallel cell counting with a 100% success rate,” Lab Chip 15(6), 1515–1523 (2015). [CrossRef]
41. L. Amato, Y. Gu, N. Bellini, S. M. Eaton, G. Cerullo, and R. Osellame, “Integrated three-dimensional filter separates nanoscale from microscale elements in a microfluidic chip,” Lab Chip 12(6), 1135–1142 (2012). [CrossRef]
42. T. Tičkūnas, D. Paipulas, and V. Purlys, “Dynamic voxel size tuning for direct laser writing,” Opt. Mater. Express 10(6), 1432–1439 (2020). [CrossRef]
43. K. Obata, J. Koch, U. Hinze, and B. N. Chichkov, “Multi-focus two-photon polymerization technique based on individually controlled phase modulation,” Opt. Express 18(16), 17193–17200 (2010). [CrossRef]
44. G. Vizsnyiczai, L. Kelemen, and P. Ormos, “Holographic multi-focus 3D two-photon polymerization with real-time calculated holograms,” Opt. Express 22(20), 24217–24223 (2014). [CrossRef]
45. G. R. B. E. Romer and P. Bechtold, “Electro-optic and acousto-optic laser beam scanners,” Phys. Procedia 56, 29–39 (2014). [CrossRef]
46. G. Kontenis, D. Gailevičius, L. Jonušauskas, and V. Purlys, “Dynamic aberration correction via spatial light modulator (SLM) for femtosecond direct laser writing: towards spherical voxels,” Opt. Express 28(19), 27850–27864 (2020). [CrossRef]
47. K. Maximova, X. Wang, A. Balčytis, L. Fan, J. Li, and S. Juodkazis, “Silk patterns made by direct femtosecond laser writing,” Biomicrofluidics 10(5), 054101 (2016). [CrossRef]
48. Y.-L. Sun, Q. Li, S.-M. Sun, J.-C. Huang, B.-Y. Zheng, Q.-D. Chen, Z.-Z. Shao, and H.-B. Sun, “Aqueous multiphoton lithography with multifunctional silk-centred bio-resists,” Nat. Commun. 6(1), 8612 (2015). [CrossRef]
49. C. A. Coenjarts and C. K. Ober, “Two-photon three-dimensional microfabrication of poly(dimethylsiloxane) elastomers,” Chem. Mater. 16(26), 5556–5558 (2004). [CrossRef]
50. J.-F. Xing, M.-L. Zheng, and X.-M. Duan, “Two-photon polymerization microfabrication of hydrogels: an advanced 3D printing technology for tissue engineering and drug delivery,” Chem. Soc. Rev. 44(15), 5031–5039 (2015). [CrossRef]
51. J. Torgersen, X.-H. Qin, Z. Li, A. Ovsianikov, R. Liska, and J. Stampfl, “Hydrogels for two-photon polymerization: A toolbox for mimicking the extracellular matrix,” Adv. Funct. Mater. 23(36), 4542–4554 (2013). [CrossRef]
52. D. S. Correa, P. Tayalia, G. Cosendey, D. S. dos Santos, R. F. Aroca, E. Mazur, and C. R. Mendonca, “Two-photon polymerization for fabricating structures containing the biopolymer chitosan,” J. Nanosci. Nanotechnol. 9(10), 5845–5849 (2009). [CrossRef]
53. E. Skliutas, M. Lebedevaitė, S. Kašetaitė, S. Rekštytė, S. Lileikis, J. Ostrauskaitė, and M. Malinauskas, “A bio-based resin for a multi-scale optical 3D printing,” Sci. Rep. 10(1), 9758 (2020). [CrossRef]
54. D. Gailevičius, V. Padolskytė, L. Mikoliūnaitė, S. Šakirzanovas, S. Juodkazis, and M. Malinauskas, “Additive-manufacturing of 3D glass-ceramics down to nanoscale resolution,” Nanoscale Horiz. 4(3), 647–651 (2019). [CrossRef]
