Fabrication of microcolumn arrays for high-throughput oligonucleotide synthesis using 3D printing

Abstract

Advances in high-density oligonucleotide array synthesis have enabled the creation of increasingly large DNA libraries. However, current planar array platforms are designed to synthesize only femtomole-scale amounts of oligonucleotides per feature, due to their inherently small feature sizes and high spatial density. Conversely, column-based synthesis supports higher per-feature output but accommodates fewer sequences. In this study, we introduce a 3D-printed DNA synthesis substrate that combines the strengths of both approaches. Using an LCD-based 3D printer and photopolymer resins, we fabricated micrometer-scale columns filled with controlled pore glass (CPG), enabling high-density oligonucleotide synthesis with sub-nanomole-scale output per feature. We screened commercial resins for chemical resistance and optimized printing parameters to fabricate rectangular microholes with gap widths smaller than CPG beads, capable of securely retaining CPG beads. A 15-mer poly(dT) sequence was synthesized across 1000 microcolumns on the array using an inkjet-based phosphoramidite system. The substrate was fabricated in ~ 2.5 h at an estimated material cost of just $0.12 per unit.

Introduction

Oligonucleotide libraries are foundational tools across various fields of biology, including synthetic biology¹, genomics², and DNA-based data storage^3,4. To enable the parallel synthesis of thousands to millions of unique sequences, several array-based oligonucleotide synthesis methods have been developed, such as photolithographic arrays^5–7, inkjet printing^8–10, and electrochemical synthesis¹¹. However, these in situ techniques operate on planar surfaces, which inherently limit the amount of oligonucleotides that can be synthesized per feature. This design prioritizes high spatial density, often at the cost of synthesis scale, typically resulting in femtomole-scale output per feature^12–16. Consequently, downstream applications frequently require amplification to obtain sufficient material. Moreover, as feature size decreases, edge effects¹⁷ and spatial misalignments¹⁸ increase the risk of sequence errors, necessitating highly precise equipment for accurate synthesis.

In contrast, column-based synthesis—typically using CPG beads—enables oligonucleotide synthesis at nanomole- to micromole-scale per feature^1,19–22, due to its three-dimensional reaction environment. These columns are spatially separated, minimizing cross-contamination and allowing synthesis to be performed with relatively simple equipment. Nevertheless, conventional column-based platforms are limited to low-throughput formats such as 96- or 384-well plates^19–22 and consume large volumes of reagents, making them inefficient for high-throughput applications.

In our previous work, we explored photolithography-based fabrication of high-density CPG arrays on silicon wafers²³. While oligonucleotide synthesis was achievable, the closed-well design obstructed reagent flow through the CPG beads, leading to inefficient washing and increased reagent consumption. Moreover, the fabrication process demanded specialized equipment for silicon wafer processing, resulting in relatively high production costs.

To address these limitations, we developed an upgraded microcolumn structure fabricated using LCD-based 3D printing technology. This design enables efficient washing and minimizes reagent use, while also offering rapid and cost-effective production. The substrate comprises micrometer-scale columns arranged in a grid at several hundred micrometers apart, each filled with CPG beads and functioning as an individual microcolumn for sequence-specific oligonucleotide synthesis. A schematic overview of the system is presented on Fig. 1. Figure 1a illustrates the integration of array-based synthesis and column-based synthesis strategies. This substrate compensates for the limitation of each substrate and increased efficiency through combination of both synthesis strategies (Table 1). Figure 1b shows the structural design of the columns that retain CPG beads by combining a wide top, allowing bead insertion and a narrow bottom to prevent the CPG from escaping. Figure 1c demonstrates the bulk delivery of reagents through vacuum suction. The reagents wet the CPG through this flow, causing a reaction, then waste liquid is effectively removed. Figure 1d shows the precise delivery of amidites via inkjet printing. Amidites are deposited on the CPG according to the synthetic sequence. Since the array surface is smooth and hydrophobic, while the CPG inside the columns is hydrophilic, the amidite is absorbed into the CPG, reducing the risk of overflow between the columns. Figure 1e conceptually depicts the parallel synthesis of oligonucleotide sequences in each column.

Fig. 1.

Overview of microcolumn array. (a) A microcolumn array combines the advantages of column-based large-volume and high-density oligonucleotide synthesis. The different oligonucleotide sequences synthesized on CPG in each column in high capacity. (b) After the microcolumn array is printed via 3D printing, the CPG is inserted. Each column has a relatively wide square hole on top, and a narrow hole on bottom. (c) Bulk solutions are dispensed onto the upper surface of the array. The lower surface of the array draws the reagents using a vacuum. (d) In the coupling step, amidites according to the synthetic sequence are printed on each column via inkjet. (e) Parallel synthesis of different oligonucleotide sequences in each column.

Table 1.

Comparison of each synthesis method.

	Manufacturer	Features	Advantage	Disadvantage	References
Column	BioAutomation	Up to 192 in a single run	Low throughput Low error rate Up to mmol scale synthesis	Large scale synthesis High cost from large reagent usage	25,26
	Biolytic	Up to 768 bp (two 384 plates)			26,27
Planar	Agilent	Up to 244,000 oligos in parallel, Femtomole scale	Relatively reduced consumption and cost High throughput	Relatively short average oligo length Needs amplification for many downstream uses Sensitive to alignment/edge effects.	28
	Twist bioscience	Synthesis up to 300nt. Femtomole scale			9,29,30
	Affymetrix	Over 600,000 oligos per chip		Necessary of numerous masks. Relatively short average oligo length Needs amplification for many downstream uses Sensitive to alignment/edge effects.	31
	CustomArray	Potential capacity of reaching 200 billion.		Unintentional reaction and cross contamination from light diffraction. Relatively short average oligo length Needs amplification for many downstream uses Sensitive to alignment/edge effects.	32,33
Microcolumn arrays	This research	Significant amount of oligo (~ nmol scale) Significant yield increasement than planar platform High flexibility from 3D printing	Easy fabrication Low cost	Current stepwise coupling efficiency is slightly lower than conventional method.

To realize this concept, we evaluated the chemical resistance of commercial photopolymer resins against harsh reagents. Additionally, both the hardware and resin formulations were modified to achieve smooth surfaces and high array density. These optimizations enabled the reliable fabrication of a high-density synthesis substrate containing 1000 microcolumns. The substrate, featuring embedded microstructures, was designed for inkjet-based reagent delivery systems, as described in the following sections.

Results

Screening of photopolymer resins for chemical resistance to oligonucleotide synthesis reagents

Oligonucleotide synthesis requires repeated exposure to harsh organic solvents and acidic reagents, which can severely degrade polymer-based materials. To identify suitable materials for substrate fabrication, we screened eight commercially available photocurable resins for chemical resistance. The tested resins, along with their primary chemical components and manufacturers, are summarized in Table 2.

Table 2.

Components and manufacturers of eight commercial resins tested for chemical resistance.

Resin	Main component	Additives	Manufacturer
Strong X	Tricyclodecanedimethanol diacrylate	Bisphenol A ethoxylate dimethacrylate Pentaerythritol tetraacrylate	Liqcreate
Deep blue	Bisphenol A ethoxylate dimethacrylate	Diphenyl(2,4,6-trimethylbenzoyl)phosphine oxide
Composite X	Bisphenol A ethoxylate dimethacrylate	Diphenyl(2,4,6-trimethylbenzoyl)phosphine oxide Pentaerythritol tetraacrylate
Aqua-Clear plus	2-Propenoic acid, 2-hydroxyethyl ester, polymer with 1,1-methylenebis4-isocyanatocyclohexane and .alpha.,.alpha.,.alpha.-1,2,3-propanetriyltris.omega.-hydroxypolyoxy(methyl-1,2-ethanediyl)	4-Acryloylmorpholine	Phrozen
Aqua-Hyperfine	Oxybis(methyl-2,1-ethanediyl) diacrylate	2-Propenoic acid, monoester with 1,2-propanediol, polymer with (chloromethyl)oxirane, dihydro-2,5-furandione and 4,4-(1-methylethylidene)bisphenol Tris[2-(acryloyloxy)ethyl] isocyanurate 4-Acryloylmorpholine Hexane, 1,6-diisocyanato-, homopolymer, 2-hydroxyethyl acrylate-blocked
Speed	2-Propenoic acid, 2-hydroxyethyl ester, polymer with 1,1-methylenebis4-isocyanatocyclohexane and .alpha.,.alpha.,.alpha.-1,2,3-propanetriyltris.omega.-hydroxypolyoxy(methyl-1,2-ethanediyl)	4-Acryloylmorpholine
Aqua 8 K	Oxybis(methyl-2,1-ethanediyl) diacrylate	4-Acryloylmorpholine 2-Propenoic acid, monoester with 1,2-propanediol, polymer with (chloromethyl)oxirane, dihydro-2,5-furandione and 4,4-(1-methylethylidene)bisphenol
PLA pro	Acrylated aliphatic urethane	1,6-Hexanediol diacrylate	eSUN

*Components present at less than 10% and unspecified additives are not listed.

Test samples, 3D-printed using each resin, post-cured for 1 h and immersed in five representative reagents commonly used in oligonucleotide synthesis—dichloroacetic acid (DCA), acetonitrile (ACN), oxidizing solution, propylene carbonate (PC), and 30% aqueous ammonia—for 24 h. Insufficiently post-cured resins exhibited severe deformation after immersion, including swelling, warping, surface cracking, and in some cases, disintegration, particularly when exposed to aggressive reagents such as DCA, ACN, and oxidizer (Fig. 2a). These deformations depend on each resin, however most resins exhibit such damage upon exposure to reagents. For example, on Fig. 2a, the Aqua 8 K resin shows warping in ammonia solution, disintegration and surface cracking in DCA, and swelling in ACN. When the post-cure time was extended to 2 h, severe deformation was no longer observed after 24 h of reagent exposure (Fig. 2b). However, slight swelling and minor surface cracking still occurred for some resins. To quantitatively assess material stability, we measured the weight change of each sample before and after immersion (Fig. 2c). Among the candidates, Aqua 8 K showed the highest weight gain, with a 5.26% increase in ACN and visible surface damage. Since inkjet-based oligonucleotide synthesis requires precise reagent delivery across microcolumns spaced within hundreds of microns, even minimal deformation can compromise synthesis accuracy. Therefore, we prioritized materials with minimal weight change across all reagents. PLA Pro, Deep Blue, and Composite X showed less than 2% weight gain in all tested reagents.

Fig. 2.

Screening of resin resistance to oligonucleotide synthesis reagents. (a) Examples of Aqua 8 K deformation caused by synthesis reagents resulting from insufficient post-curing. (b) Resin with optimized curing time no longer exhibits severe deformation. (c) Percent weight change of each resin after chemical exposure.

In particular, Deep Blue demonstrated excellent chemical resistance, with less than 0.23% weight gain for four of the five reagents and only 1.62% for DCA. Notably, no visible deformation or surface damage was observed on Deep Blue samples after reagent exposure, reinforcing its chemical robustness. Although Composite X exhibited slightly lower weight gain under DCA exposure (1.50%), Deep Blue was selected as the optimal resin due to its superior visual and quantitative performance. Moreover, the actual DCA exposure time per synthesis cycle is short (< 2 min), minimizing its impact.

Optimizing 3D printing parameters

Following the selection of a chemically resistant resin, we proceeded with the optimization of 3D printing parameters to enhance microstructural resolution. CPG beads, which are widely used in DNA synthesis, typically range from 70 to 150 μm in diameter. To securely retain these beads while ensuring smooth reagent flow, it was necessary to fabricate holes with widths less than 70 μm with high fidelity.

Although modern LCD 3D printers offer theoretical resolutions below 50 μm, a combination of optical and mechanical factors often compromises accuracy when producing microscale features. In particular, hole diameters below 200 μm frequently suffer from clogging and poor reproducibility. The main limiting factors were identified as follows: (1) UV over-curing due to resin transparency and free radical diffusion, (2) suboptimal curing settings, and (3) light reflection and scattering at the build plate surface. To address these limitations, we optimized not only the resin formulation, but also the printing parameters and printer hardware configurations.

First, we modified the formulation of Deep Blue resin, which has high UV transmittance due to its translucent blue color, making it susceptible to over-curing. To suppress this, we incorporated a UV absorber—Sudan Orange G (SOG)—at a concentration of 0.15 mg/mL into the resin formulation. Since LCD-based 3D printers utilize 405 nm wavelength UV light, SOG was selected as the UV absorber. It has been reported to effectively attenuate light at 405 nm even at low concentrations²⁴, making it particularly suitable for minimizing excess polymerization while preserving the resin’s chemical resistance. The exact stability of SOG containing Deep Blue resin was evaluated by immersion into reagents. Because of its inherently high stability, the modified resin still exhibited strong chemical resistance with less than 2% weight gain across all reagents. Optimization of UV absorber allowed us to retain the desirable mechanical and chemical properties of Deep Blue resin while improving the resolution of fine structural features (Fig. 2c).

Along with the resin modification, other printing parameters were also optimized. UV intensity was decreased to 80% of the default setting, and layer thickness set to 50 μm. These parameters were adopted to strike a balance between fidelity and printing speed. The targeted hole size on bottom side was 66 μm which is equivalent to three pixels.

Quantitative measurements of 100 holes per condition revealed that hole width decreased with increasing exposure time. A 12-s exposure produced an average width of 68.9 μm, most closely matching the design target of 66 μm. A 13-s exposure resulted in narrower holes with an average width of 58.7 μm, whereas most holes cured for 14 s were completely closed (Fig. 3a). Representative microscopic images also supported this trend: the hole printed with a 13-s exposure maintained an open rectangular shape suitable for CPG retention, while the 14-s exposure led to complete closure of the feature (Fig. 3b). Although the 12-s exposure produced holes closest to the intended design width, 13-s exposure was chosen as the optimal time. It consistently produced open holes and was expected to retain CPG beads more effectively, despite resulting in slightly smaller holes.

Fig. 3.

Optimization of UV exposure time for microscale hole 3D printing. (a) Quantitative analysis of hole widths (mean ± SD, n = 100) as a function of curing time. An exposure time of 13 s resulted in an optimal hole size (58.7 μm) that securely retains CPG beads while maintaining structural fidelity. Error bars represent standard error of the mean (SEM, n = 100). (b) Representative microscopic images of holes printed with 13 s (left) and 14 s (right) of UV exposure showing open and closed morphologies, respectively.

After exposure optimization, we addressed the influence of the build plate surface on the fidelity of the initial printed layer. Most commercial printers use patterned metal plates that enhance adhesion via diffuse reflection; however, this also promotes undesired surface curing (Fig. 4a). When synthesis substrates were printed directly onto these plates, persistent over-curing led to frequent clogging of microholes in the initial layer, even at reduced exposure times (Fig. 4b). To mitigate light scattering from the build plate surface, the patterned metal plate was replaced with a smooth, black glass plate (Fig. 4c). This replacement significantly reduced light reflection and scattering. This was demonstrated by increasing exposure time required for proper adhesion of the initial printed layer on the black glass surface, compared to simply overlaying a transparent glass plate on a conventional build plate. Additionally, this modification resulted in a smooth substrate surface with improved hydrophobicity, promoting enhanced amidite absorption onto the hydrophilic CPG and facilitating reagent washing. Under these improved conditions, we successfully fabricated microholes with dimensions of 275 μm × 275 μm in the initial layer (Fig. 4d). Regarding the microhole diameter, we observed that in larger sizes where CPG beads were not held tightly, the beads were lost after multiple reagents dispenses and inert gas drying steps, resulting in empty columns. This CPG loss tended to increase as the hole diameter became larger; however, at the 275 μm diameter, no CPG loss was observed.

Fig. 4.

Impact of build plate surface on light scattering and first-layer fidelity. (a) Schematic illustration of light scattering from a patterned metal build plate, promoting overcuring and adhesion. (b) Clogged (left) and overcured (right) microstructures when printed on metal plates. (c) Schematic illustration showing reduced scattering when using a black glass build plate. (d) Successful fabrication of open microholes on a glass build plate using optimized curing parameters.

Through these three key optimizations—resin formulation adjustments to reduce light penetration and confine curing to target regions, determination of optimal curing time, and modification of the build plate surface structure to reduce light scattering—we reliably and reproducibly fabricated synthesis substrates. Each substrate measured 25 mm × 33 mm, and contained 1000 precisely formed microcolumns. The microcolumns were arranged with a center-to-center spacing of 550 μm (25 pixels) and individual microhole had dimensions of 275 μm × 275 μm (11 pixels). At this spacing, up to 324 microcolumns can be accommodated within a 1 cm² area. The optimized fabrication process required a printing time of 29.5 min followed by a 2-h post-curing step. Up to 12 substrates could be printed simultaneously. The material cost per substrate was approximately $0.12.

DNA synthesis for functional evaluation of the substrate

To evaluate the functional performance of the fabricated DNA synthesis substrate, oligonucleotide synthesis was performed using our previously developed inkjet-based synthesizer²³. Upon loading CPG beads into the microcolumns, each column retained 1–3 intact particles, and no leakage was observed through the ~ 60 μm wide bottom openings (Fig. 5a).

Fig. 5.

DNA synthesis for functional evaluation of the substrate. (a) Microscopic images of the substrate columns loaded with CPG beads (top: front view; bottom: bottom view). Each column retained 1–3 intact CPG beads without escaping through ~ 60 μm width holes. (b) Urea-PAGE analysis of the synthesized poly(dT) 15-mer. Lane 1: DNA ladder1; Lane 2: synthesized sample; Lane 3: control 15-mer; Lane 4: DNA ladder2. The full-length product (15-mer) was the dominant band, with a stepwise coupling efficiency of ~ 96.23% estimated from band intensity analysis.

The 15-mer poly(dT) oligonucleotide was synthesized on the substrate and the obtained oligos were analyzed with Urea-PAGE. The presence of the expected 15-mer product was confirmed, along with shorter truncated sequences (Fig. 5b). The result confirmed that the major product corresponded to the intended 15-mer poly(dT). This was synthesized as a 14-mer extension from the dT already present on the CPG bead, with several shorter failure sequences also detected. Quantification using ImageJ showed that the full-length product accounted for 60.7% of total signal, corresponding to an estimated stepwise coupling efficiency of 96.23%. The total yield, measured via spectrophotometry, was 705,513 ng. Assuming uniform synthesis across the array, this corresponds to an estimated 155.2 pmol per column. Despite the presence of minor byproducts, these results demonstrate that the synthesized substrate enables high-density oligonucleotide production at nanomole levels per feature.

Discussion

In this study, we implemented a microscale version of the conventional column-based oligonucleotide synthesis method by filling a 3D-printed microcolumn array with CPG beads. Each microcolumn features a square upper chamber and a small bottom slit that enables reagent flow while securely retaining the beads. This configuration merges the high-density layout of array-based synthesis methods with the high per-feature loading scale of column-based platforms, effectively integrating the advantages of both approaches.

In addition, the hydrophobic and smooth surface of the substrate, in contrast to the hydrophilic CPG-filled columns, promotes the preferential absorption of amidite droplets into the microcolumns—even in the presence of minor alignment errors. This feature minimizes edge effects and cross-contamination between adjacent features, thereby reducing the need for high-precision droplet positioning that is typically required in inkjet-based synthesis systems. Detritylation, oxidation, and wash reagents were dispensed onto the peripheral regions of the microcolumn array—areas devoid of columns—allowing the reagent solution cover the entire substrate surface. This indirect delivery approach prevented loss of the CPG beads. The reagents were subsequently drawn through the columns via vacuum suction, ensuring uniform exposure to the beads.

Our selection of LCD-based 3D printing was based on a rigorous comparison with other additive manufacturing technologies, driven by our primary requirement for micro-scale resolution (< 70 μm). Common methods like Material Extrusion (FDM) were disqualified due to fundamental nozzle size limitations and poor reproducibility for such features, an issue we confirmed in preliminary experiments. Powder Bed Fusion (SLS) is similarly unsuitable for forming precise micro-holes due to its thermal melting process. While Material Jetting can theoretically achieve this resolution, its high equipment cost (hundreds of thousands of dollars) and severe material limitations (requiring extremely low-viscosity resins) made it impractical. Therefore, vat photopolymerization (specifically LCD/DLP) offered the only viable combination of micrometer-scale resolution, high throughput via parallel layer curing, low cost, and a broad selection of materials.

The successful synthesis of a 15-mer oligonucleotide validates the functional reliability of the microcolumn array. The per-column yield, as estimated from the total synthesis output (detailed in the Results section), represents a 3- to 6-order-of-magnitude increase in synthesis scale compared to conventional planar array platforms. This substantial enhancement underscores the potential of the system for downstream applications that require higher DNA input quantities.

However, several limitations remain to be addressed. The stepwise synthesis efficiency in this study was lower than the conventional oligonucleotide synthesis efficiency (typically > 99%). We synthesized oligonucleotide using a custom-built synthesizer, as our novel substrate format required it. While we hypothesize that relatively low synthesis efficiency is attributable to our custom-built synthesizer not being as optimized as commercial instruments, the limitations of the microcolumns themselves cannot be disregarded. Specifically, it is possible that synthesis and wash reagents did not flow smoothly or that residual reagents were not completely removed, thereby impairing synthesis efficiency. It is also possible that such issues were more pronounced in certain columns (e.g., those located on the periphery), which could lower the apparent overall synthesis efficiency. Furthermore, while the Results section reported an average yield based on the assumption of uniform synthesis across the entire array, a more rigorous validation of per-column yield would require direct verification of the synthesis output from individual columns.

The most definitive method to verify this is to analyze the spatial uniformity of DNA synthesis using NGS after synthesizing unique sequences in each column. By analyzing NGS sequencing data from sequences that include both a barcode (to track the physical location of each column on the array) and a Unique Molecular Identifier (UMI) (to mitigate PCR bias), we can quantitatively map the synthesis error rate and yield variation according to the physical (x, y coordinate) location of each column on the array.

Separately, it is possible that the resin-fabricated substrate chemically interfered with the DNA synthesis. Although our chemical resistance tests (via reagent immersion) and observations during the synthesis process revealed no significant damage, commercial resins are complex mixtures of various chemical components. Therefore, the potential chemical influence of unlisted ingredients cannot be fully excluded. To address this, comparative studies could be performed by synthesizing DNA using various materials in addition to the Deep Blue resin we selected. The most robust solution would likely be to use Perfluoropolyether (PFPE), which is known to be one of the most chemically resistant materials compatible with 3D printing. However, 3D printing microstructures with PFPE is significantly more challenging and costly compared to using commercial resins, which presents its own set of challenges. As we are continuously developing our synthesizer and substrate fabrication, while comprehensive validation of synthesis uniformity via NGS is a critical next step that we have planned, the primary contribution of this study lies in establishing the proof-of-concept for this novel fabrication paradigm. Our study is particularly significant because it presents not only a substrate structural design that integrates the advantages of existing synthesis methods, but also a comprehensive fabrication approach that makes this structure inexpensive and easily reproducible via 3D printing. Given the low material cost (approx. $0.12 per substrate)—which justifies its use as a cost-effective, single-use disposable item (similar to the CPG itself)—short fabrication time (approx. 2.5 h), and the low cost of the fabrication equipment (an affordable 3D printer costing only a few hundred dollars), this approach is accessible not only to dedicated oligonucleotide synthesis facilities but also to academic laboratories and small-scale research groups.

Moreover, while this report demonstrates the fabrication of 1000 microcolumns, the inherent scalability of the 3D-printed array fabrication process, along with its compatibility with inkjet-based synthesis, enables straightforward expansion to tens or even hundreds of thousands of features with only minor modifications. Although this substrate was developed in the context of oligonucleotide synthesis, its microcolumn architecture is broadly applicable as a modular solid-phase reaction environment. Each column effectively serves as an isolated microreactor, offering opportunities for expansion into synthetic chemistry and biochemical applications. Because RNA and peptide synthesis are also traditionally performed on bead-based solid supports, our platform can be straightforwardly adapted by simply loading the microcolumns with the appropriate functionalized beads (e.g., polymer resins for peptide synthesis or RNA-functionalized CPG for RNA synthesis) instead of the DNA-CPG used here. This transferability is further enhanced by our resin optimization strategy. Rather than relying on a single proprietary resin, our ‘stable base resin + SOG additive’ method is a practical and flexible strategy. It allows researchers to select any resin with the specific chemical resistance needed for a new application (e.g., peptide or RNA synthesis) and simply ‘tune’ the resolution by adding SOG. As 3D printing materials continue to evolve—with the development of photopolymers that offer enhanced chemical and thermal resistance—this substrate is poised to benefit from expanded functionality and broader application potential.

Methods

Screening of photopolymer resins

To evaluate the chemical resistance and compatibility of various photopolymer resins with the oligonucleotide synthesis process, a screening test was performed using several commercially available resins, as listed in Table 2. Deep Blue, Strong X, Composite X were purchased from Liqcreate. Aqua-Clear plus, Aqua-Hyperfine, Speed, and Aqua 8 K from Phrozen and eResin-PLA pro from eSUN were also obtained. The substrate was printed using each resin and washed with isopropanol to remove uncured resin residues. After complete drying, each resin was post-cured at 60 °C for 1 h to ensure its stability. Several resins exhibited slight deformation, so the post-curing time was adjusted to 2 h to enhance mechanical strength and chemical stability. Prior to immersion, the weight of each resin was measured using an analytical balance. Reagents of each synthesis step were prepared in 50mL conical tubes. The resins were immersed in DCA, oxidizer, ACN, PC, 30% aqueous ammonia. The samples were incubated in a reagent rack at room temperature for 24 h. Following incubation, the substrates were dried for a day and reweighed to evaluate any changes in weight.

Resin preparation

SOG was purchased from Aladdin scientific, and it was stored at 4 °C. In order to minimize light scattering during the printing and control the diffusion of free radical, SOG was added to the Deep Blue resin. To determine the optimal formulation, the ratio of resin to SOG was evaluated. It is critical to determine an appropriate balance, as an excessive amount of SOG may negatively impact the curing efficiency or reduce the resin’s chemical resistance, while an insufficient amount may fail to sufficiently suppress light penetration and improve accuracy. Based on preliminary trials, 15 mg of SOG was added into 100 ml of Deep Blue and thoroughly mixed to ensure homogeneity, a critical factor in LCD printing. Once SOG was added, the color of the Deep Blue changes to green. Following this modification, printing parameters, UV light exposure time, first layer exposure time, UV intensity and printing layer thickness, were carefully considered to optimize print performance. These optimizations enabled improved print resolution and structural fidelity.

Optimizing 3D printing conditions

Two optimized factors, the build plate and resin curing time, need to be carefully studied. The build plate, to which the printed resin adheres, was made of patterned metal to ensure fixation of 3D printed structure. While the patterned surface improves adhesion, it also leads to increased light reflection and diffusion. Black glass build plates were introduced to reduce light reflection and diffusion. The substrate with smooth surface also obtained by replacing glass plate. Various resin curing conditions were evaluated by printing test structures under different exposure times. Since the Phrozen Sonic Mini 8 K printer has a pixel pitch of 22 μm, hole structure was designed and printed to target width of 66 μm. Among the tested durations (12–14 s), a 13-s exposure time was selected with its suitable structure to CPG. The quality of the printed structures was assessed using a DinoCapture digital microscope, measuring feature dimensions. This process confirmed that the 13-s exposure condition yielded consistent and well-defined structures.

3D-printed substrate

A Sonic mini 8 K LCD printer from Phrozen was used and optimized for substrate fabrication. Under the optimized environment, the substrate was printed with arranged resin. The substrate was designed using Fusion 360 and sliced with Chitubox basic software. Chitubox basic can adjust various printing conditions according to the selected 3D printer and resin. To print the substate, UV intensity was set to 80%, exposure time was determined to 13-s and layer thickness was 50 μm. To ensure consistent printing quality, the homogeneity of the resin was maintained. The substrate printing takes about 30 min under the optimized environment. After printing, substrate was detached from the black glass plate and washed into the isopropanol bottle. The fully washed substrate was dried and post-cured at 60 °C for 2 h (Form cure Gen1, Formlabs) to ensure physical and chemical stability. Prior to use, substrates were stored in a desiccator. Just before synthesis, CPG beads were filled into substrate.

Reagents preparation

DMT-dT phosphoramidite was purchased from Sigma-Aldrich and dissolved to a final concentration of 0.25 M. The activator, 4,5-dicyanoimidazole (DCI), was also purchased from Sigma-Aldrich and prepared at a concentration of 0.7 M. PC, used as the solvent, was purchased from Sigma-Aldrich and dried over molecular sieves for 24 h before use to reduce residual moisture content. To prevent nozzle clogging, all reagents were filtered through a 0.45 PTFE filter before being loaded into the inkjet printhead.

DCA dissolved in DCM (3:97 v/v), ACN, and the oxidizer (0.2 M iodine dissolved in water/THF/pyridine) were purchased from DUKSAN (Korea). All chemicals were stored and handled under standard anhydrous conditions to maintain reactivity throughout the synthesis process.

DNA synthesis on the 3D-printed substrate

DNA synthesis was performed using our previously developed inkjet-based synthesizer. CPG (Product No. T401000) was purchased from Sigma-Aldrich. CPG beads were loaded into the microholes by repeatedly sprinkling them onto the substrate and gently rubbing them with a gloved finger. Excess CPG beads remaining on the surface were removed during the pre-synthesis washing steps. Prior to synthesis, the reaction chamber was thoroughly purged with argon gas at a flow rate of 20 L/min for 30 min to completely remove residual moisture.

The synthesis cycle consisted of three main steps: detritylation, coupling, and oxidation, with intermediate washing steps using ACN between each stage. To ensure reagent delivery to all microcolumns, 2 mL of each reagent was dispensed in each step to completely cover the top side of the substrate, followed by vacuum suction through the bottom side. This dispense-and-suction process was repeated 10 times for washing steps and 5 times for all other steps. After dispensing, the incubation times were 60 s for detritylation, 60 s for coupling, 30 s for oxidation, and 10 s for washing. Before each coupling step, the substrate surface was dried by purging with argon gas at 5 psi.

This cycle was repeated 14 times to synthesize a 14-mer poly(dT) sequence onto the pre-loaded dT, yielding the final 15-mer poly(dT) oligonucleotide. After synthesis, the top surface of the substrate was covered with 1 mL of concentrated ammonium hydroxide at 55 °C and incubated in a sealed container for 2 h to cleave the oligonucleotides from the solid support. The cleaved product was precipitated using n-butanol and resuspended in deionized water for subsequent analysis. The concentration of the oligonucleotides was measured using a NanoDrop spectrophotometer.

Urea-PAGE of synthesized oligonucleotides

40% acrylamide-bis solution (19:1), 10×TBE, urea, TEMED, and 10% ammonium persulfate solution (APS) were purchased from Biosesang (KR) and were used to prepare a 24% urea-PAGE gel for DNA analysis. TEMED and APS were added simultaneously to a mixture of the acrylamide solution, 10×TBE, and urea. After the solution was inverted 2–3 times, it was poured into the gel cassette to solidify. A warm 1×TBE solution was used as the running buffer to ensure clear band separation. Before loading the samples, each well was flushed to remove residual urea. Gel loading buffer II (Invitrogen) and the samples were mixed in a 1:1 ratio to a total volume of 10–20 µL. After electrophoresis, the gel was incubated in a 1×TBE solution containing SYBR Gold and then washed.

Author contributions

J.H.K., H.E.K., and D.H.B. designed the study. J.H.K. optimized LCD 3D printer and printing settings. The oligo synthesis to evaluate substrate is also performed by J.H.K., and H.E.K. contributed to the examination of the resin and the analysis of the synthesized oligonucleotides. D.H.B. supervised all aspects of the study. All authors read and edited the manuscript.

Data availability

All data generated or analyzed during this study are included in this published article or will be provided by the corresponding authors upon request.

Declarations

Competing interests

The authors declare no competing interests.