Development of a low‐cost hydrogel microextrusion printer based on a Kossel delta 3D printer platform

The core of bioprinting related research aims to reduce the gap between ex vivo cell cultures and in vivo cellular tissue models to further its application within the biomedical field. While additive manufacturing is touted as disruptive technology, bioprinter equipment costs exceed limited resource budgets of many research laboratories restricting the scope for further development for biomedical research and potential medical application. In line with this, a relatively low‐cost bioprinter (SidneV1, SV1) was successfully designed and manufactured using a low‐cost, commercially available FDM Delta 3D printer as a prototype base with a successfully custom designed and manufactured micro‐extrusion printhead. Printing accuracies assessed were 65% (for width measurements) and 64% (for height measurements). This study aimed to demonstrate a way to achieve low‐cost cell‐free hydrogel printing and establish the basis for future system modifications and refinements such that this technology becomes more accessible to under‐funded research groups in resource limited laboratories globally. Optimization of printing parameters, hydrogel and formulations and sterilization techniques will allow for the patterning and engineering of three‐dimensional macroscale scaffolds using this low‐cost printing technology.

sustain post-printed viability of cells within the 3D structures. 3,4General access to desktop & open-source additive manufacturing technologies has seen advancement in this field in recent years.Arguably, this has spurred the bioprinting revolution.
Extrusion based bioprinting has been largely focused on the manufacture of 3D tissue constructs or scaffold due to its many advantages over other methods, including, but not limited to; high cell viability, 5 flexible geometric shapes, 6 ability to incorporate multiple biomaterials and cell types, 7 both homogenous and heterogeneous structures can be created. 6,8In extrusion based bioprinting, hydrogels are extruded out of a nozzle tip to form a continuous line structure driven by either pneumatic pressure or mechanical pistons; the extruded product is referred to as filaments instead of droplets. 5,6,8A three-axis/Cartesian automatic extrusion system is typically used in this type of printing, equipped with a fluid dispensing nozzle. 91][12][13][14][15][16] Most systems make use of standard sterile plastic syringes which provide many benefits including; wide availability, low cost, pyrogen free, and compatibility with a range of needle sizes used as print nozzle heads. 17Although the most widely employed method, it is not without its limitations.Most syringe-based extruders are designed to incorporate the fluid reservoir into the extruder carriage which is solely responsible for the high mass typically associated with this approach.A heavy extruder carriage can cause various issues during the printing process by affecting speed and resolution which may cause compromises in geometries of the printed constructs. 12Reducing the volume of the fluid reservoir could provide a solution to the increased mass, however this will negatively impact the ability to print complex and larger constructs.Other larger volume systems have utilized the Bowden approach to minimize the weight of the extruder carriage.This approach makes use of Bowden tubing to connect the fluid reservoir (which is completely removed from the extruder carriage) to the nozzle.These systems are typically pneumatically driven which then brings in another separate set of limitations such as; poor or no retraction, the need for vacuum during printing and unstable extrusion pressures. 12The ability of the printer to retract is important during the printing process as this prevents material from dripping during non-extruding moves which in turn reduces printing fidelity of the construct. 12Pusch et al. 12 have addressed the two major concerns of extruder carriage weight and inability to retract using the large volume extruder (LVE) design.This design utilizes the Bowden tube approach with stepper motors and a leadscrew driven extruder.The use of stepper motors is suggested to achieve retraction through straightforward reversal of direction and to apply constant pressure using a standard syringe. 18

Design criteria
Overall, the bioprinting process has multiple facets to consider which we have previously described under one of the three pillars required for successful tissue engineering-hardware, wetware, and software considerations. 19The outcome should include low operational cost, ability to use a wide variety of materials, allow for fine deposition of materials with varying viscosities, maintain high cell viabilities, reduced maturation time of printed constructs, and finally, minimal handling of the constructs.Here, we discuss the criteria required for the hardware development process applied in this study.The overall design and development process outlined in this study is specifically for the printing of cell free scaffolds.

Rationale for printer selection
A commercially sourced, RepRap based Delta 3D printer kit was selected for modification to include a hydrogel paste extruder in the place of the original thermoplastic extruder.The high printing precision and accuracy expected of delta printers led to the choice for eventual cell-free biocompatible biomaterial scaffold printing. 20The Anycubic Kossel linear plus delta printer used in this project was based off the popular Kossel RepRap delta printer designed by Johann C. Rocholl.All documents and development kits for the Kossel delta are available online (https://reprap.org/wiki/Kossel)with many components of the model made readily available from Thingiverse (https://www.thingiverse.com/).This allows for successful modification to be made to the printer at relatively low-cost.The firmware supplied with the delta printer is the open-source marlin firmware (https://marlinfw.org/).This firmware allows for modifications allowing for paste extrusion with the delta hardware which allows for a relatively quick transition from thermoplastic extrusion to paste extrusion.
Delta printers contain an effector plate where the extruder nozzle is positioned and moves around the build volume, but the extruder stepper motor is situated outside the printing space, attached to the frame.The extruder nozzle must therefore be lightweight and secured on the effector plate to prevent unwanted movement.Suspending the extruder nozzle above the bed arguably eliminates any contamination from particles caused by friction during print moves.The dimensions of the extruder carriage and the mass allowed on the axis that holds the extruder therefore dictate the maximum dimensions of the hydrogel paste extruder.
Furthermore, the kit used in this study was selected because of the linear rail-based motion control allowing for increased XYZ precision required for bioprinting.The ball bearing design of the linear rail system allows for a more precise motion control compared to the linear rod systems, typically found in desktop 3D printers.There is a considerable reduction in binding occurrences (ball bearings getting caught on the rail during a movement) which contributes significantly to a much smoother printing movement and creates a jerk-free print.The stepper motors (NEMA17 with 1.8 • step angle) used in conjunction with the linear rails also contribute significantly to the increased precision found in this system, where micro-stepping allows for more controlled moves.
In this study a low cost commercially available 3D printer was modified to accommodate for bioprinting with cell-free hydrogels.This provided for majority of the printer parts at a relatively low cost.All additional parts were either 3D printed using thermoplastics or purchased at a low cost.Although the linear rails are considerably more expensive than smooth rod systems, the increase in printing precision is imperative to successful bioprinting and so outweighs the potential overall increase in build cost.

Design description
An Anycubic Kossel linear plus delta 3D printer kit was selected for modification.Based on the criteria for bioprinting, both the hardware and software of the delta printer was successfully modified to accommodate for hydrogel paste microextrusion.The manufacture and assembly of the final delta bioprinter was separated into three main parts; (1) the hardware modification of the 3D printer frame, (2) the replacement of the thermoplastic extruder for a cell-free hydrogel paste extruder, and (3) the software modifications for hydrogel paste extrusion.

Additive manufacturing of parts
All required 3D printed parts were printed using an Anycubic Kossel delta linear plus 3D printer fitted with an clone E3D v5 hotend with 0.4 mm nozzle (sourced from DIYElectronics Pty(Ltd), South Africa) following the same general printing parameters as listed: layer height: 0.1 mm, fill density: 100%, print speed: 80 mm/s, printing temperature: 210 • C, 100% flow, no support structures were required unless absolutely necessary with no bed adhesion.Poly-L-lactic acid (PLA) was used as the printing material (1.75 mm diameter eSUN PLA + 3D filament, DIYElectronics, South Africa).No additional post-print processing was required.The 3D modeling software SketchupMake2017 (v.17.2.2555) was used to design all printed parts (unless otherwise specified) and.STL files were sliced using Cura (v.15.04.2) to convert files to gcode.

SV-1 frame (Hardware) modifications
Enclosure: The printer was enclosed using both printed and non-printed parts (Figure 1A).The side frames and mid-sections (later referred to as bioprinter frames, windows, and doors) were designed in Autodesk Fusion 360 and lasercut in 3 mm clear Perspex (M&D Creations, Makhanda, South Africa).The 3D printed parts consist of three separate corners (Figure 1A).

Controller board box:
The Anycubic Tri-gorilla controller board was moved from under the build plate (as per manufacturers design) to outside the system; shown in Figure 1B.Wiring was adjusted to accommodate for the new board position.A board box was designed, and 3D printed to accommodate for the new location of the board.
Printing bed platform: A new bed platform was designed, and 3D printed, to hold two 220 mm diameter glass plates of 2 mm thickness (Figure 1B).The bed platform with the glass in place was designed to cover most of the base of the printer with a few openings to allow for air flow to reduce any pressure build-up internally should the system be modified with a HEPA filtered blower.

F I G U R E 1
Exploded view of modifications made to delta printer frame including, repositioning of the electronic board, and new printing platform

Extruder modifications
The syringe extruder was based off the RepRapPro paste extruder designed by Adrian Bowyer (available under a GPL; all files available from the RepRapPro Ltd Github repository at https://github.com/reprappro/Paste-extruderand at http://reprapltd.com/reprappro/documentation/building-the-paste-extruder/index.html).The syringe-based extruder designed in this project is made up of printed parts, non-printed parts, and electronic components."MOUNTING BRACKET" was custom designed and 3D printed.The extruder was assembled as per instructions given on the RepRapPro website.Finally, a Luer-lock effector plate adaptor was designed to attach to the original effector plate provided with the delta printer and incorporates a male Luer-lock connection fitting and a Bowden tube connection fitting (Figure 3).
All design files are available at the source file repository at https://doi.org/10.17605/OSF.IO/7TQKB.A full description of the individual parts is available here in the supplemental information (Supplementary Document 1).Furthermore, a full Bill of Materials (BoM) is provided (Supplementary Document 2).A step-by-step build guide is provided as additional documentation (Supplementary Document 3).

Preparation of alginate hydrogel
A 6% (w/v) alginate solution was prepared by dissolving alginic acid sodium salt (Sigma Aldrich) in sterile double distilled water under continuous stirring on a hot plate for 1 hour.The alginate was prepared and used immediately before each experiment.A 0.3 M CaCl 2 (Sigma-Aldrich) solution was prepared and used for cross-linking of the alginate post-printing.All constructs used for testing the printer were designed in SketchupMake2017 (v.17.2.2555) and sliced at 0.1 mm layer heights in Cura (v.15.04.2).Post print images were captured through macrophotography on a smartphone, compared to a scale ruler and analyzed in ImageJ (v 1.4.6.r).All 5 ml sterile syringes and needles were purchased from LASEC Pty (Ltd) or RS Components (South Africa).

Petroleum jelly printing
Commercially available petroleum jelly (purity baby petroleum jelly) was used as a model gel extrusion paste to simulate printing of high viscosity materials.Petroleum jelly was loaded in 5 ml syringes and printed at low speed (9 mm/s) and high flow (250%).Constructs were designed in TinkerCad and sliced in Cura (v.15.04.2) at 0.2 mm layer height and 100% infill.Images were captured on a smartphone, macroscale constructs were measured with a Vernier caliper.

Assembly of SV-1
Following printing and laser cutting, the SV-1 prototype was assembled as described in Supplementary Document 3.
The fully assembled enclosure (Figure 4) allowed for a controlled environment for printing with the frame mounted RepRapPro (RRP) paste extruder (Figure 5).The syringe unit was designed to carry a 5 ml syringe but the open-source design allows for expansion of this to larger volume syringes.The NEMA 17 stepper motor driven RRP extruder utilized in this study uses a Bowden tube approach along with a threaded rod driven extruder system to allow for controlled deposition of hydrogels and pastes.As with the LVE approach, attaching the RRP paste extruder to the printer frame places all the weight on the printer frame and not the extruder carriage (Figure 5).This adds minimal payload to printer movements and should in turn maximize the print speeds and reduce any vibration of the nozzle during printing which may lead to nozzle dripping (Figure 5C).
Using the measurements of height and width (mm), of both Figure 6A,B, the printing accuracies (%) were calculated for the assembled hydrogel extruder (shown in Figure 6C).Printing accuracies (%) for height and width were calculated at 64.13% and 64.58%, respectively.The settings selected were suitable for experimental trials based on cell-free 6% (w/v) concentrated hydrogel inks, hence these parameters were kept for the remaining experiments except for flow rate and needle gauge optimization studies.

3.2.1
Optimization studies: Does needle gauge and flow rate affect printability?

Effect of needle gauge on printing accuracy
The printing accuracy (%) was assessed using the following equation Printing accuracy (%) = Experimental value Theoretical value × 100 The fully assembled SV-1.The modified Anycubic Kossel linear plus printer enclosed in perspex and the effector plate was modified to allow for the addition of the RepRapPro paste extruder A 6% (w/v) alginate solution was used as extrusion material to print the designed constructs with subsequent cross-linking in CaCl 2 solution for 2 min.Images were taken of the cross-linked constructs and measurements determined with ImageJ software (v1.4.6.r).Prints were repeated three times for each needle gauge tested (15G, 18G, 23G, and 25G).The flow percentage used in the experiment was 100%.This experiment aimed to determine the effect of nozzle diameter (using different needle gauges) on the printed construct shape quality, under certain constant parameters (flow rate = 100%, print speed = 10 mm/s).Figure 7 shows the images of the printed constructs (post-printing and cross-linking), printed using various needle gauges; 15G, 18G, and 23G.The inner diameters for each needle gauge are as follows; 15G = 1.372, 18G = 0.838, 23G = 0.337 mm.A final needle gauge was included in the experiment; 25G (inner diameter = 0.260 mm), however, the printed constructs produced were not uniform and therefore printing accuracies based on filament width were unable to be calculated; Figure 7B(ii).The printing accuracies (%) produced for each needle gauge experiment include; 136.53%, 86.01%, and 129.93% for 15G, 18G, and 23G, respectively.Although no statistical significance was found between groups, the 18G needle produced a printing accuracy closest to 100% (Figure 7A).Kahl et al. 21found that the needle diameters are essential in optimizing printing resolutions such that; the larger the needle diameter, the smaller the resolution and therefore the lower the printing accuracy.The choice in needle diameter, length and shape (i.e., conically shaped or cylindrical) has additional consequences in terms of cell viabilities.Findings suggest shorter, conically shaped needles allow for higher cell viabilities, post-printing, compared to cylindrically shaped needles. 11,22,23These findings have major implications in cell-laden bioprinting approaches but little implications in cell-free approaches, as the cells are seeded onto the bioprinted constructs post-printing.

Effect of material flow rate on hydrogel extrusion and filament thickness
The spreading ratio was determined at different flow percentages (100%, 125%, 150%, 200%) using the following equation:

Spreading ratio =
Filament diameter Nozzle diameter .The flow percentages listed are extrusion multipliers.These values instruct the printers to extrude more material during the prints.Post-printing images were taken immediately after printing.Calcium chloride was not added to the printed constructs for cross-linking.A nozzle diameter of 0.838 mm (18G needle) was used for the syringe extruder based on the results obtained from the needle gauge printability experiment.Three constructs were printed for each flow percentage tested.An average of the width, measured at various positions along the filament, of the printed filaments were divided by the needle diameter which remained constant throughout the experiment; 0.838 mm.The constructs were printed using 6% (w/v) alginate hydrogel with no post-printing cross-linking methods employed.Figure 8 shows a positive correlation between flow percentage and spreading ratios obtained.The spreading ratios obtained for 100% and 200% flow percentage groups showed a statistically significant difference; p = 0.0363; p < 0.05, whereas no other statistical significance was found between groups.
Similarly, Kahl et al. 21shows the higher the extrusion rate, at a constant nozzle diameter, results in lower shape fidelity of the printed construct.Images in Figure 8 show the inconsistency in filament structure produced when printing at a 100% flow percentage, even though a low spreading ratio was noted.Due to the complex nature of bioprinting and the necessity to print complex geometries at extremely high resolutions, the final printed construct must be highly precise and accurate.A high spreading ratio is, therefore, an undesirable characteristic and further optimizing must take place in order to accommodate for this. 24

Printing of materials of high viscosity
Biological structural complexity requires that the printers patterning biomaterials for tissue engineering need to be capable of manufacturing geometrically complex structures.Using petroleum jelly as a model material with high viscosity (e.g., nanofibrillated cellulose) we show the versatility of the SV-1 for printing more viscous pastes.As can be seen in Figure 9 (also Supplementary Video 1) the dimensional accuracy of simple shapes are translated and maintained when printing at low speed (9 mm/s) and increased flow (250%) at 0.2 mm layer heights.Spreading of the petroleum jelly appears to affect the XY dimensional accuracy as can be seen from the inner and outer diameter measurements determined post printing (Figure 9A-C).Interestingly, the layering is more evident in the reflection of the standard petri dish that was used as a printing surface (Figure 9C white arrow).The circularity of the printed construct is translated well.In Figure 9D,E we see how font kerning is translated from the CAD file to actual printed construct.Again, we see the spreading of the petroleum jelly likely affecting the printing of sharp defined edges, but the overall structures are maintained indicating a potential for further optimization for biomaterial printing.

CONCLUSIONS
The overall aim to create a low-cost hydrogel microextrusion based printer from a commercially available thermoplastic 3D delta printer was achieved.The final bioprinter designed and manufactured consists of an already established syringe-based extruder, incorporated with minor alterations to fit the delta style 3D printer, and an easily built commercially available delta printer with added enclosure and easily modifiable firmware.Converting the thermoplastic delta 3D printer into a bioprinter cost a total of $ 1007.52,making it affordable to most research laboratories (see complete BoM Supplementary Document 2).This system allows for the printing of geometrically complex structures using a range of material viscosities and can be easily modified (i.e., firmware parameters) for other paste extrusion materials and future cell-laden hydrogel bioprinting.
Securing the extruder carriage to the frame of the printer had the desired effect of removing the excessive weight from the effector plate which then allowed for smoother printing movements and higher printed construct quality.Another important design criterion stipulated was that of producing highly complex shapes with precise geometry.This criterion was partially met but remains an area that requires much improvement.Lowering the extruder carriage closer to the nozzle would aid in reducing the distance between the syringe feedstock and the nozzle which in turn allows for more control over the extrusion of the bioinks.More control over the extrusion means less possibility of under or overextrusion; gaps forming within the construct versus leakage of bioink during travel moves, respectively.
There is, however, much room for improvement where this study only aims to provide the much needed and promising foundation work for what could be a fully functional low-cost bioprinter with multiple applications.The modifications made to the hardware during the conversion from 3D printer to bioprinter were successful and allow for increased environmental control for the sensitive nature of bioprinting, as set out as a priority criterion.It is suggested that the addition of HEPA-filtered laminar air flow blower to the system would allow for further environmental control for sterility.Smaller sterility-related improvements may include the addition of germicidal UV lights to the system and re-designing of the internal hardware components (i.e., extruder carriage) into fully enclosed and easily cleaned singular components.
Overall, the results achieved with the final bioprinter prototype model were satisfactory and provide the much-needed foundation work for future optimization studies.Incorporating the use of cell-laden bioinks and assessing the different modes of cell-based bioprinting will be employed in future work, along with the assessment of various bioink formulations and cross-linking methods.While delta style bioprinters are commercially available (e.g., Pensees Vitarix and VitarixW) the systems are registered design closed-source premium systems.The leading goal of the SV1 was toward the development of a full open-source system.Relative to commercial systems the substantial cost reduction presented may prove to be impactful to research laboratories in lower resourced economies as well as in developed countries.
Figure 2 below shows the full syringe-based extruder, including six parts printed directly from the RepRapPro Extruder design.The part labeled F I G U R E 2 Rendering of assembled and mounted syringe based RepRapPro paste extruder Briefly, Marlin Firmware (v.1.1.0)was modified (Supplementary Document 4) using the Arduino Integrated Development Environment (IDE v 1.8.1.0)and firmware was transferred to the anycubic controller board.The modified firmware is available at the project source file repository.Modifications that were made temperature control and movement settings.A full list of modifications is available in Supplementary Document 4.Pronterface (printrun-201503101) host interface was used for controlling the printer X-, Y-and Z-movements as well as the extrusion motor.Default Pronterface settings were used at 250,000 baud rate.As mentioned above the firmware was modified for controlled hydrogel microextrusion.F I G U R E 3 Rendering of the Luer-lock effector plate adapter.(A) Full assembly with effector plate and needle nozzle, (B) front view, (C) top view, (D) bottom view, (E) side view

F I G U R E 5
The printed and assembled syringe-based extruder including the effector plate adaptor.(A) Fully 3D printed and assembled syringe extruder highlighting the RepRapPro syringe extruder (B) and the Luer-lock effector plate adaptor (C)

F I G U R E 6 6 F I G U R E 7
Assessment of 6% (w/v) printability by the SV-1 (A) CAD file of construct, (B) post-printing image of a representative printed construct, (C) post-printing accuracy (%) calculated for the printed alginate constructs (using height and width).Scale bars represent 5 mm.Data points shown are means; error bars represent SEM for n = Assessment of printability using different nozzle diameters.(A) Post-printing accuracy (%) calculated for alginate constructs printed using various nozzle diameters (15G, 18G, and 23G), 25G not included in graph.(B) Post-printing photographs of alginate constructs printed using (ii) 25G, (iii) 23G, (iv) 18G, and (v) 15G needle gauge.(i) CAD file for printed construct.Scale bars represent 5 mm.Data points are means; error bars represent SEM for n = 3.No statistical significance was found between groups (p = 0.0522; p > 0.05) F I G U R E 8 A comparison of hydrogel printability.(A) Post printing spreading ratio's (filament diameter/needle diameter) determined for various flow rates using 6% (w/v) alginate.(B) Post printing photographs of printed constructs for each flow rate (i) 100%, (ii) 125%, (iii) 150%, and (iv) 200%.Data points are means; error bars represent SEM for n = 3. Asterix represents statistical significance (p = 0.0363; p < 0.05), no asterisk shows no statistical significance (p > 0.05).Scale bars represent 5 mm

F
I G U R E 9 SV-1 printability of multilayer objects.(A) Simple cylinder designed in TinkerCad (AutoDesk) printed at 0.2 mm layer height in petroleum jelly (B, top) and (C, perspective view).White arrow indicates layering.(D) RU letter text designed in TinkerCad and (E) printed in petroleum jelly (0.2 mm layer height).Key for in figure table: H, height; OD, outer diameter; ID, inner diameter.All samples were printed in triplicate (n = 3)