Lithography-based 3D printing of hydrogels

Abstract

Additive manufacturing is an engineering tool that enables the creation of complex structures for biomedical use, such as for 3D scaffolds for tissue engineering and regenerative medicine and in vitro disease models for drug testing. Lithography-based techniques (e.g., digital light processing DLP, volumetric additive manufacturing VAM) have particularly advanced in recent years for the 3D processing of photoreactive resins into structured hydrogels. The aim of this review is to introduce the various light-based lithographic 3D printing methods that are being used to process hydrogels, provide a guide to lithography-based printing from bioresin selection to the optimization of print parameters, highlight examples of in vitro and/or in vivo biomedical applications of hydrogels where lithography-based approaches have been leveraged, and discuss recent advanced efforts to process hydrogels into heterogenous structures with multi-scale organization. Finally, a perspective on the challenges and opportunities ahead in this field is provided.

Short Summary

Lithography-based 3D printing has recently emerged to process of hydrogels for biomedical applications. This review highlights key examples, provides guidance on formulating resins and step-by-step manual for printing hydrogels, and discusses challenges and opportunities that lie ahead towards translation.

Introduction

Hydrogels – water swollen polymer networks – recapitulate many biochemical and biophysical properties of native tissues¹. This capability makes them promising for biomedical applications including tissue engineering², mechanobiology³, drug delivery⁴, and bioinspired medical devices⁵. Despite these promising attributes, hydrogels are often fabricated in the form of isotropic/uniform constructs that fail to mimic the complex architectures of the extracellular microenvironment and macroscopic tissue structures. Additive manufacturing (AM) allows for the processing of hydrogels into physiologically-relevant shapes, including with controlled placement of cells into desired 3D configurations. As a subset of AM, lithography-based 3D printing or vat photopolymerization techniques (e.g., stereolithography SLA, digital light processing DLP, volumetric additive manufacturing VAM) have emerged as promising technologies that rely on spatial crosslinking of liquid photoreactive resins. Patterns are defined via computer aided design (CAD) to form structures using either top-down or bottom-up light projection modalities. Lithography-based 3D printing methods generally exhibit finer resolution and faster printing speeds than classical AM methods (e.g., extrusion printing, inkjet printing), as the fabrication relies on layer-by-layer printing or exposure of large resin volumes to light instead of linear filament or droplet deposition⁶.

This review first provides an overview of the various vat photopolymerization methods that have been developed and compares and contrasts them to help the reader identify the approach that best fits their desired application. Emphasis is then placed on emerging vat photopolymerization methods (i.e., DLP, VAM), including a step-by-step guide from resin selection to print optimization for DLP-based 3D printing. Throughout the review, the use of the term ‘resin’ refers to solutions of monomer/macromer, crosslinker, and photoadditives (e.g., photoinitiator, photoabsorber) that are used to form hydrogels. We further highlight recent example applications from tissue engineering to in vitro tissue models where lithography-based 3D printing has been leveraged and elaborate on emergent methods that are expanding the complexity of printed systems. Finally, we discuss challenges and opportunities, including a roadmap to apply new vat photopolymerization approaches to hydrogel processing, including towards commercialization and clinical translation. Importantly, this review seeks to highlight recent findings in lithography-based processing methods for 3D printing of hydrogels with a goal to provide a forward-looking perspective to beginners and advanced users in the field. We direct our readers to more specialized reviews for comphrensive, in-depth insights on vat photopolymerization ^6–8.

Vat Photopolymerization Methods

Stereolithography (SLA) and two-photon polymerization (2PP) utilize laser scanning (e.g., continuous, pulsed) along the vat surface or within the vat volume to crosslink a photoreactive resin into a 3D object in a linear (point-by-point) manner (Fig. 1). Note that we use the term ‘crosslink’ throughout this review to describe the solidification of the liquid resin, which describes either the polymerization of reactive monomers into polymer chains and crosslinking into a network or the reaction of functionalized polymers (i.e., macromers) directly into networks. Most SLA systems consist of a high intensity near-infrared (NIR) laser source, a focusing, high-numerical-aperture objective, and a translational stage to allow the building of 3D objects⁹. A controlled laser beam follows a predetermined path to crosslink the resin to a desired depth. The use of direct laser writing unlocks the potential for wavelength-sized, sub-micrometer (less than 100 nm) scale features with nanometer tolerance dictated by the laser source^10,11. However, this extraordinarily high resolution is often realized at the cost of long fabrication times (on the order of hours). Further, the fabrication of larger structures is limited by the working distance of the microscope objective used to focus laser pulses into the resin, which can dramatically reduce scalability. The laser path is controlled by deflection off a rapidly moving mirror galvanometer, whereby a single photon laser is focused through an objective lens to activate curing of the photocurable resin (e.g. collagen, gelatin, hyaluronan with or without cells)^12–15. 2PP-based 3D printing requires the use of specialized photoinitiators with high quantum yield and sufficient absorbance at the irradiation wavelength to achieve free-radical production within the desired volume of resin. As this review will not focus on SLA and 2PP printing, we direct the reader to other recent reviews focused on these methods^16,17.

Figure 1 |. Comparison of the resin requirements and print characteristics of various vat photopolymerization methods.

Differences between 2PP, DLP, and VAM/CAL based on mode of light projection, compatibility of resin (e.g., viscosity, volume needed), typical time required for printing an object with the same volume, and lowest feature size achievable. Print time and resin volume denote the theoretical time and amount needed to fabricate an object with the same dimensions across various vat photopolymerization methods. It is important to note that print time does not consider the total processing time, which includes pre- and post-print processing. Additionally, the listed volume does not consider modifications that can be made to the vat size to reduce resin use. The volume of resin needed depends on the volumetric size of the object to be printed, with vat photopolymerization methods typically requiring resins with volumes for VAM/CAL > DLP ≈ SLA ≥ 2PP. 2PP: two-photon polymerization, SLA: stereolithography, DLP: digital light processing, VAM: volumetric additive manufacturing, CAL: computed axial lithography.

Digital light processing (DLP) involves top-down or bottom-up (depending on position of the light source) light projection to enable the stepwise fabrication of 3D objects and is one of the most widely used lithography-based printing methods for hydrogels. Light-based (e.g., UV, 365 nm or visible, 405 nm) crosslinking of a liquid resin is achieved through patterned light projection via a digital micromirror device (DMD, 8-bit depth amplitude modulation) (Fig. 1). A 3D object is built layer-by-layer with projection of sequential 2D sliced images onto resin layers via a motorized build platform that moves in the build (z-) direction¹⁸. Each digitally displayed image contains an array of square pixels, which in turn result in the formation of a single-layer 3D rectangular brick called voxels. This method is often much faster than the use of a controlled laser beam as with SLA, resulting in the building of larger objects in much less time. The print time is solely dependent on the height (depth) and exposure time for each layer, rather than the size in the x-y direction.

In contrast to a layer-by-layer approach, continuous fabrication methods that involve steady, upward movement of the build platform with simultaneous projection of sliced images have also been developed (i.e., continuous liquid interface production, CLIP)^19,20. CLIP relies extensively on oxygen-inhibition to form a “dead-zone” (i.e., thin layer of uncured resin present between the oxygen permeable vat and the printed part) that permits continual replenishing of unreactive resin. This eliminates the staircasing effect that stems from layer-by-layer fabrication and also results in much faster fabrication speeds (upwards of 50 cm h⁻¹). Further improvements (5 to 10-fold) in fabrication speeds and compatibility with viscous resins can be realized via iCLIP that involves the injection of resin through microfluidic conduits into the dead zone during printing²¹. However, the need for the dead zone limits these methods to oxygen sensitive chemistries. To expand the printing method to oxygen insensitive chemistries, fluorinated oil has been employed to reduce adhesion between the printed part and the resin interface, but this has not yet been used for hydrogels²². Other approaches for continuous fabrication rely on the use of (i) an extremely soft, deformable hydrogel interface to reduce adhesion force²³, (ii) high velocity flow of resin for rapid replenishment,²⁴ or (iii) two-color irradiation of resins containing complementary photoinitiators and photoinhibitors, where one wavelength activates polymerization and the other inhibits polymerization to balance resin flow and part separation^25,26. Although bottom-up light projection is commonly employed with DLP, top-down projection allows a buoyant force to support printed structures, which is particularly useful for soft hydrogels. Nevertheless, with top-down projection, challenges regarding the possible size of a printed construct, oxygen inhibition at the surface from the air, and defects and inhomogeneities due to surface tension mismatch of the resin exist²⁷. The printing of soft hydrogels (less than 10 kPa) can be undertaken with a buoyancy-assisted continuous DLP printing method known as fluid-supported liquid interface polymerization (FLIP), which employs light projection on a support bath containing an immiscible photoreactive resin²⁸.

Volumetric additive manufacturing (VAM) or computed axial lithography (CAL) is a more recently developed lithography-based approach that relies on volumetric photopolymerization using light projection similar to 3D coherence tomography (CT) imaging (Fig. 1)²⁹. Rotation of the vat containing a photoreactive resin is synchronized with projection of sliced images to allow accumulation of a 3D light dose from patterns projected from all angles, which allows solidification of voxels rather than sequential or continuous curing of layers. Unlike slicing of objects in the x-y direction in the case of DLP, CAD models in VAM are built in the z-direction, as the z-axis is parallel to the axis of rotation of the material volume³⁰. To achieve appropriate light dose within the vat, the resin solution should have high optical transparency to prevent light scattering or attentuation through the build volume and changes in resin turbidity post-crosslinking should be avoided. Ideally, all parts of the print should polymerize at the same time; however, this is not always the case, especially for fine features (smaller than the oxygen diffusion length) that are often undercured and must be corrected for via a deconvolution method³¹. Additionally, the relative motion of the object being printed must be minimized during the vat rotation to avoid adverse effects on print resolution, shape, and fidelity, which requires the use of resins with high viscosity (higher than 10 000 Pa s) or solids³². Sacrifical, thermoreversible polymers (e.g., gelatin, agarose) have been added to resins to increase their viscosity to enable VAM printing, which are then removed through changes in temperature to recover the printed part^33,34.

Beyond the speed of printing, a major advantage of VAM compared to DLP is towards the printing of low modulus (i.e., 1–10 kPa) hydrogels. DLP-printing of soft hydrogels is challenging due to the drag or viscous forces exterted on overhanging features or on cells encapulated within resins, whereas VAM allows the printed object to remain relatively stationary during the print to minimize these forces. Some advantages of DLP printing over VAM include: (i) VAM generally requires larger build volumes that may limit scalability (especially in the use of high cell densities), (ii) printed resolutions are often greater with DLP, (iii) the fabrication of tall objects is limited with VAM, and (iv) optical modulation instabilities (OMI) of the light beam in VAM can result in defects. To address these limitations, helical additive manufacturing combines the rotation and translation of the resin vial to enable the fabrication of multi-centimeter structures that are much larger than traditional VAM without the need for magnifying projected images³⁵. Further, fabrication throughput can be scaled up with the use of continuously flowing photoresin through a specialized vat³⁶. To address resolution concerns, recent efforts have been focused on sequential correction where an object is printed, visualized, and the print adjusted to correct for overcured or undercured regions³⁷. Additionally, to avoid challenges with synchronous vat rotation and light projection in VAM, fabrication can also be achieved via non-rotational methods such as through the superposition of patterned optical fields from multiple projected beams, where their intersection creates holographically generated images within photoreactive resin vats³⁰.

Similar to printing resolution, the overall size of the object to be printed depends highly on the printer specifications. For instance, in the case of SLA or DLP, the size in the xy direction is limited by the area of the build platform and the maximum travel distance of the build platform in the vertical direction controls the possible z-height. DLP-based 3D printers with build areas as high as 40 × 25 × 46 cm (bottom-up, Carbon Inc.) and 45 × 37.1 × 39.9 cm (top-down, Desktop Metal Inc.) enable the fabrication of large constructs, as well as arrays of multiple objects that are printed simultaneously. Similarly, the xyz dimensions in the case of VAM or CAL are defined by the volume of the vat. In addition to printer specifications, print size can also depend on the resin employed. For example, in the case of bottom-up DLP, hydrogels with low moduli can fail to sustain their own weight during fabrication, which can result in delamination and limit the maximum z-height achievable before the force of gravity dominates.

Printing an object with the same dimensions is generally much faster with VAM or CAL due to the volumetric light projection when compared to planar (in the case of DLP, CLIP) or point/linear (in the case of SLA, 2PP) light projection. While these differences in print time can be a defining factor for the method to be selected, it is also important to highlight that print time alone does not holistically capture the various steps typically involved in these approaches. The entire processing routine can involve steps towards the preparation of the resin vat (e.g., calibration of the build platform or rotation stage), print retrieval, washing, and post-curing (e.g., light flood cure of printed object), which can require significant amounts of time. Moreover, the total processing time can also depend on the type of hydrogel to be printed. For instance, pre- and post-processing times may vary depending on the handling requirements (e.g., thermoreversible gelation in the case of VAM) or when bioprinting with cells to account for additional steps to isolate cells and maintain their high viability. These considerations make it difficult to generalize the overall processing time for any of the lithography-based 3D printing methods described here and will depend on the user-defined system and application.

Resins for Vat Photopolymerization

This section highlights the various parameters with regards to resin formulations that must be considered for DLP or VAM-based printing of hydrogels (Fig. 2). Printer specifications (e.g., light source and intensity, optical resolution, speed of build plate movement or vat rotation) can also considerably influence lithography-based printing of hydrogels and are discussed within Box 1.

Figure 2 |. Various resin and print parameters that influence vat photopolymerization.

Schematic representation of resin formulation considerations and printing parameters that will influence the vat photopolymerization process and final printed object properties, such as the type of polymerization reaction, choice of reactive species (i.e., type and concentration of monomer/macromer, degree of modification, viscosity, molecular diffusion), optical properties of the resin (i.e., type and concentration of photoinitiator, photoinhibitor/photoabsorber, refractive index matching solution) and printer specifications (resolution, light intensity and wavelength, slicing layer thickness, and speed of build platform/vat movement).

Box 1 |. Printer Considerations.

There are many printers available for lithography-based printing of hydrogels. These range from commercial options for DLP (e.g., marketed by Asiga, Cellink, 3D systems, Formlabs, Carbon, Desktop Metal) and VAM (e.g., Readily3D) printing, as well as custom-built printers. The cost to purchase printers varies widely based on complexity and features such as resolution, grayscale capability; however, it is important to note that there are low-cost options available that make printing accessible^229,230.

Maskless light projection within commercial or custom-built printers is usually supported through LCD or DMD²³¹. LCD dynamic projectors use liquid crystal displays to generate images, whereas DMD uses an array of mirrors that function as light switches such that each mirror corresponds to a single pixel and the pixel dimensions can range from 5 to 25 μm. Although LCD projectors are relatively cheaper, their optical efficiency is limited and they exhibit large pixel sizes, low switching speeds, and low optical contrast. LCD-based DLP printers can exhibit variations in the light source between printers and hence to evaluate and ensure consistency, spectrophotometers should be used to probe irradiance uniformity. Within the light setup, the choice of light source can be LED or laser. LEDs are much more cost-effective than lasers; however, their high divergence may compromise resolution. While the choice of wavelength within commercial printers is largely limited to 385 or 405 nm light, they may also provide user control over the irradiant light intensity. For more complex workflows or to integrate other wavelengths (e.g., 520 nm), researchers can turn to custom-built printers^56,232.

The optical resolution (i.e., pixel size) of the digital projection plays a large role in the resolution of printed constructs; however, it is not the sole factor responsible for defining the smallest feature that is achievable during printing²³³. Physico-chemical properties such as resin reactivity, oxygen diffusion, scattering opacity, viscosity, extent of light illumination, and accuracy of the projected image can also contribute significantly to the fidelity of the print^79,234. In the case of DLP, layer thickness in the z-direction is dependent on both (i) optical penetration depth within the resin and (ii) positional accuracy of the vertical translation stage that controls up/down movement of the build platform.

For both DLP and VAM, software considerations for printing pertain to slicing algorithms (to convert encoded data in the form of CAD or STL file to a technique specific output like porjection image) and synchronization (to set up communication between projection system and motor to move build platform or vial)⁶.

Monomer and macromer chemistry:

Photoinitiated reactions are necessary in photo-based lithography, which are defined by the chemistry of the monomer or macromer used³⁸. Free radical chain growth polymerizations are commonly used and proceed upon radical formation with photoinitiator exposure to light, followed by chain propagation and termination³⁹. Monomers and macromers contain functional groups with varied reactivity (e.g., polymerization rates: acrylate > acrylamide > vinyl ester ≈ vinyl carbonate > methacrylate > methacrylamide)³². Although chain-growth reactions are oxygen inhibited, these can reach gelation even at low conversions (< 2%). Unlike chain-growth reactions, step-growth reactions (e.g., thiol-ene) are oxygen insensitive and result in increased network homogeneity and faster kinetics but require higher conversions to reach gelation. Thiol-ene reactions occur in two steps: i) propagation of a thiyl radical through an ene functional group (e.g., norbornene, allyl ether) to form a carbon radical and ii) chain transfer from the resulting carbon radical to a thiol functional group, regenerating the thiyl radical⁴⁰. Electron rich alkenes (e.g., norbornene, allyl ether, vinyl ether) react faster with thiols than electron poor alkenes (e.g., acrylonitrile) such that the reactivity is norbornene > allyl ether ≥ vinyl ether > maleimide > methacrylate⁴¹. Improvements in chain length uniformity are possible with living or controlled radical polymerization reactions (i.e., reversible addition/fragmentation chain transfer RAFT and atom transfer radical polymerization ATRP)^42,43. Other reaction chemistries (e.g., photoredox⁴⁴, ring-opening metathesis⁴⁵) have not been widely adopted for hydrogels yet. In addition to conventional chemical crosslinks, there is a growing interest in leveraging physical (e.g., polymer chain entanglements, guest-host interactions, hydrogen-bonding) and dynamic covalent chemistries (e.g., disulfide, boronate esters, Diels-Alder adduct) for vat photopolymerization of hydrogels⁴⁶. Incorporating such dynamic bonds allows for potential reshaping and reprocessing of 3D printed hydrogels⁴⁷. Further, dynamic interactions can also endow 3D printed hydrogels with self-healing ability. Another emerging approach has been the use of mechanophores (e.g., spiropyran, anthracene dimers) as latent crosslinkers within printed materials. These crosslinks can undergo triggered covalent rearrangement, allowing for polymer chain extensions to enhance material moduli in response to mechanical deformation^48,49. Although most mechanochemistry has been limited to organic solvents, the development of water soluble mechanophores will be beneficial for light-based 3D printing of mechanically tough hydrogels.

Type, concentration, and degree of modification of reactive monomers and macromers:

Natural polymers (e.g., gelatin^50–52, hyaluronic acid^53,54, silk^55,56, alginate⁵⁷, cellulose⁵⁸, chitosan⁵⁹, κ-carrageenan⁶⁰, decellularized matrix^61–63, bovine serum albumin⁶⁴) are often modified with reactive groups and used in photolithography to form hydrogels, owing to their innate biodegradability, bioactivity, low immunogenicity, and biocompatibility⁶⁵. Disadvantages such as batch-to-batch variability and dispersity in molecular weights and functionalization can limit their ease of use with vat photopolymerization, which is highly sensitive to changes in crosslinking kinetics due to inconsistency in reactive group concentrations. In contrast, synthetic monomers or polymers (e.g., polyethylene glycol PEG^18,66, polyvinyl alcohol PVA^33,67, polyacrylamide^68,69) allow a high level of tunability and control over mechanical and biochemical properties and can be modified with cell-instructive cues to guide cell behaviors (e.g., attachment, growth, differentiation). Polymer concentrations and degrees of modification can impact important photolithography parameters such as time to gelation and cure depth. An emerging approach towards the formation of soft networks is the use of bottlebrush polymers, as compared to linear or branched polymers which could be extended towards 3D processing of bottlebrush hydrogels⁷⁰.

Resin viscosity:

Resin viscosity is another important parameter for vat photopolymerization, as it influences reaction kinetics, oxygen diffusion, and forces experienced by the object during printing. A low viscosity resin may enable faster reactivity of monomers but may also allow the rapid diffusion of oxygen and reaction inhibition within the resin. Nevertheless, low viscosity resins are better suited for DLP, owing to the low capillary forces exerted on the object during the build platform lift off and recoating process between printing of subsequent layers. However, cell sedimentation can occur within low viscosity resins, causing inhomogeneous cell distributions. To prevent cell settling, the viscosity of the resin can be modulated through addition of buoyancy modifiers such as Percoll, silk fibroin, and xanthan gum¹⁸. When printing viscous resins with DLP, it is recommended to use slower build platform speeds for detachment between layers, as the capillary forces scale linearly with separation speed. Increasing the viscosity of the resin can also reduce the diffusion length of radicals, thereby improving resolution. As mentioned earlier, higher viscosity resins are often desired with VAM to stabilize the printed part during vial rotation. The viscosity of the resin can depend on the polymer type (e.g., charged or neutral), concentration, molecular weight, and architecture (e.g., linear or branched). For example, high molecular weight, linear polymer chains can result in more entanglements and increase resin viscosity, whereas branched polymers are difficult to pack and result in fewer entanglements and hence, a lower resin viscosity.

Photoinitiators:

Photoinitiators are integral for photolithography-based printing and must match both the wavelength of the light source used (with high molar extinction coefficients for efficient radical generation) and the crosslinking mechanism employed⁷¹. Additionally, photoinitiators to be employed in the presence of cells must exhibit (i) high solubility in aqueous solutions, (ii) cytocompatibility, and (iii) high visible light absorption and quantum yields even under low-intensity irradiation. Type I photoinitiators are comprised of a single component and typically absorb in the UV or visible range, which allows cleavage into an excited, intermediate state followed by generation of free radicals⁷. Most widely used type I initiators include 2-hydroxy-l-[4-(hydroxyethoxy)phenyl]-2-methyl-l-propanone (Irgacure 2959), lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP), 2, 2′-azobis[2-methyl-n-(2-hydroxyethyl) propionamide] (VA-086), and 2-methyl-l-[4-(methylthio)phenyl]-2-(4-morpholinyl)-1-propanone (Irgacure 907), and TPO nanoparticles^7,32. Type II initiators consist of a photoinitiator in combination with a coinitiator and include riboflavin, Eosin-Y, and tris(2,2-bipyridyl) dichlororuthenium(II) hexahydrate (Ru)/ sodium persulfate (SPS) which all absorb in the visible light range⁷². As an example, when irradiated with visible light, the ground state Ru²⁺ gets photoexcited and oxidizes via donation of electrons to SPS (coinitiator). Increased photoinitiator concentrations can remarkably increase the rate of photopolymerization, but often at the cost of (i) over-curing due to the diffusion of radicals outside of the illuminated area, (ii) shorter polymer chains due to increased initiation centers, and (iii) light attenuation that reduces cure depths.

Photoinhibitors and photoabsorbers:

Photoinhibitors and photoabsorbers play a crucial role in controlling photopolymerizations during lithography-based printing. Photoinhibitors such as (2,2,6,6-tetramethylpiperidin-1-yl)oxidanyl (TEMPO) act as stable, free-radical quenchers that terminate growing polymer chains to impede polymerization in a concentration dependent manner⁷³. In contrast, photoabsorbers provide dose-dependent delays in the initiation of polymerization by competing with photoinitiators to absorb light^18,67. This reduces the light penetration depth (D_p) and increases the critical energy (E_c) required to reach gelation. Similar to photoinitiators, ideal photoabsorbers should exhibit low cytotoxicity, high absorption at the source wavelength, and undergo sufficient elution from the hydrogel post-fabrication. Photoabsorbers can also result in gradients in conversion within each printed layer that must be balanced to improve print resolution without the loss of mechanical strength^53,74. Specifically, the cure depth (C_d) must be slightly higher than the print layer thickness to prevent delamination, but not excessive to avoid inadvertent polymerization of negative features. Tartrazine, a yellow food dye with absorption maximum at 405 nm, is one of the most widely adopted photoabsorbers for 3D printing of cell-laden hydrogels¹⁸. Other photoabsorbers include natural and synthetic dyes such as Ponceau 4R, curcumin (from turmeric), anthocyanin or modified biomacromolecules⁷⁵. Biocompatible pigments or inorganic nanoparticles such as gold or titanium dioxide can also confer light attentuating properties, but may have undesirable effects of light scattering or hinder light transmission for imaging. Non-reactive light attentuating additives can be effective at preventing undesired curing in the z-direction (or along the axis of light exposure), but may also significantly slow down photopolymerization rates. Alternatively, a slice correction algorithm that modifies projected images based on modeling of the curing process can be used to mitigate overcuring without compromising the printing speed⁷⁶.

Oxygen inhibition:

Molecular diatomic oxygen (present in the air or in the dissolved state within the resin) can react with radicals generated (e.g., initiation or propagation steps) during photopolymerization^27,32. As a result, oxygen can quench radicals, which results in the termination of growing chains or reduction in polymerization rates. This is manifested as uncured layers that can adversely impact mechanical properties and resolution of the printed object. Oxygen inhibition can be overcome through various techniques such as increasing irradiation energy, crosslinking in an inert environment, degassing resins before printing, or the use of oxygen scavengers⁷⁷. Further, quantitative models to account for the effect of oxygen diffusion and inhibition during vat photopolymerization can help predict curing profiles for different resins^78,79. While oxygen can be a negative factor for free-radical polymerization via conventional DLP, oxygen inhibition is often desirable for methods such as CLIP and VAM^19,29. Within CLIP and VAM, oxygen acts as a radical scavenger to limit polymerization until the local oxygen concentration is depleted below a critical threshold to reach gelation.

Refractive index:

High cell densities are needed in the printing of various soft tissues (e.g., heart, pancreas, liver) to closely recapitulate native tissue level complexity and function; however, cells are highly heterogeneous with subcellular organelles that act as scattering elements. Scattering of light can reduce (i) lateral resolution during printing due to uncontrolled photopolymerization in the region surrounding the desired pattern and (ii) light penetration depths in the vertical direction, particularly in the case of bottom-up or top-down light projection. To overcome the refractive index mismatch between cells and the surrounding medium, iodixanol – a biocompatible refractive-index matching compound – has been utilized for 3D printing of cell suspensions and aggregates (e.g., organoids) at high densities via both DLP and VAM^80,81. For example, the inclusion of iodixanol resulted in a 10-fold reduction in light scattering and a subsequent improvement in resolution down to 50 μm for a bioink containing 0.1 billion cells per mL. Beyond cells, other scattering elements (e.g., titanium dioxide nanoparticles, hydrogel microparticles) may also be present. To account for this, projected patterns displayed onto a DMD can be modified to compensate for scattering by (i) correcting for the attenuation of light and (ii) increasing blur of the pattern with depth⁸². In addition to experimental manipulations, computational methods involving deep learning algorithms may be applied to generate print parameters to compensate for scattering⁸³. Refractive index matching can also circumvent the spontaneous break-up of a uniform optical beam into filamented light beams when propagating within optically nonlinear media to avoid undesired defects in the form of microchannels or microfilaments^84,85.

Step-by-step Guide to DLP Printing

Here we provide a guide to the various steps that need to be considered for lithography-based printing of hydrogels (Fig. 3). Given the vast availability of low-cost, commercial DLP printers, we focus this guide on the DLP-based 3D printing of hydrogels and at a level helpful for first-time users.

Figure 3 |. Stepwise guide to digital light processing (DLP) of hydrogels.

Schematic overview outlining the approach to DLP from the identification of an appropriate resin through the 3D printing process. Step 1 involves selection of the type and concentration of photoreactive molecules for use as a resin. Step 2 requires tuning of optical properties of the resin. Step 3 determines the resin working curve and optimizing the print resolution. Steps 4 and 5 involve 3D printing and post-processing of the hydrogel construct. G′, G″ denote storage and loss moduli over time (t). C_d corresponds to the cure depth, E denotes the irradiant light energy, D_p is the light penetration depth for a resin formulation and E_c is the critical energy required to reach gelation.

Step 1:

The first step is the formulation of a reactive resin, including the selection of the reactive monomer or macromer to be used. The choice between natural or synthetic monomer or macromer can depend on the desired application, as discussed in the previous section. Natural (e.g., gelatin, hyaluronic acid) or synthetic (e.g., PEG, PVA) molecules must include photoreactive functional groups such as acrylates, methacrylates, norbornenes, or allyl glycidyl ethers. The concentration to be employed can influence viscosity, mechanical properties, as well as cell outcomes. For instance, a high polymer content increases the ability of the hydrogel to support higher loads, whereas a low polymer content may be more favorable for cell viability. Overall, increasing the concentration or molecular weight of the polymer can significantly increase the resin viscosity, making it challenging to adopt for DLP.

Step 2:

Once the monomer or macromer and its concentration is defined, the next step requires optimization of the resin optical properties. A photoinitiator is necessary to induce crosslinking of the resin upon light irradiation and the appropriate photoinitiator concentration can be identified through photorheological measurements. While a high photoinitiator concentration can considerably speed up the reaction rate, very high concentrations can be cytotoxic due to the generation of excessive radicals. Generally, for the most commonly used water soluble photoinitiator (i.e., LAP), the concentration required to DLP print cell-laden hydrogels is typically less than 34 mM¹⁸. Additionally, the use of photoabsorbers is highly recommended to avoid over-curing. There exists a tradeoff such that a high photoabsorber concentration increases resolution, but increased light attenuation reduces reactive group conversion and adversely affects the printed hydrogel mechanical properties. Photoabsorbers must possess high molar extinction coefficients at the wavelength of printing, which allows control over the light penetration depth (D_P). Printing of a 3D object requires specification of a layer thickness (z_d) (often, 25 to 100 μm), which should be lower than or comparable to D_P to allow integration between layers. Delamination of the printed object can occur if z_d is higher than D_P. In the presence of light scattering elements (e.g., cells, granular media), a water compatible refractive index matching fluid (i.e., iodixanol) can be added at a recommended concentration between 20 to 35 wt.%⁸⁰.

Step 3:

Although there is a need for an ASTM standard to normalize the steps to obtain working curves, there are commonly adopted protocols that involve measuring cure depths as a function of irradiant energy (i.e., exposure time, irradiation intensity). It is highly recommended to use a calibrated, filtered light source with a narrow spectral bandwidth to ensure consistency and reproducibility across DLP printers^86,87. Measurements of cure depths of hydrogel resins can be performed using contact (e.g., atomic force microscopy, AFM) or non-contact (e.g., laser scanning confocal microscopy) methods^88,89. Jacob’s equation $C_d=D_p\ln \left(\frac{E}{E_C}\right)$ can then be used to correlate the cure depth (C_d) and minimum energy required to reach gelation (E_C). This step involves iteration of the photoabsorber concentration to identify optimal exposure conditions for a given resin formulation. Deviations from an ideal exposure energy can result in under- or over-curing of the printed features as compared to CAD dimensions, which can adversely affect the print resolution. In the case of over-curing hollow channels or negative features, increased photoabsorber concentration or decreased exposure energy can help. For a DLP printer with a fixed light intensity source, estimation of the exposure time per layer may require trial and error to arrive at the optimum exposure condition. Assessment of a working curve for each resin formulation as described can be both time-consuming and wasteful of material; however, theoretical models based on the relationship between exposure time and cure depth can be used⁹⁰.

Step 4:

Once the ideal exposure conditions have been identified, the next step includes the execution of 3D printing. The 3D object desired is first sliced into 2D images. Most DLP printers require inputs about base and body layer exposure times, layer thicknesses, and build platform movement speeds. The base of the 3D object is often exposed at a 2–3 fold higher energy than the body layers to allow attachment to the build platform, which may also be rough to further facilitate mechanical adhesion. In some cases, the roughness can result in poor optical clarity of the layers closest to the build platform, which can be circumvented with the use of a smooth or functionalized glass coverslip attached to the build platform. When processing objects, the effect of print direction (0, 45, 90°) of the model on print accuracy should be evaluated. While the build platform movement speed is usually dependent on the DLP hardware, it is recommended to keep the speed low for viscous resins to prevent delamination of the printed object due to capillary drag.

Step 5:

Several post-print processing strategies⁹¹ have been developed to improve the stiffness of the printed object (defined by the green strength)⁶. Improved mechanical strength is often the result of increased conversions of unreacted monomers and crosslinkers aided by post-curing methods. Light post-curing is the most widely adopted method that involves UV or visible light flood cure of the printed part. Similarly, thermal curing can be employed through the use of a thermal initiator within the resin before 3D printing, followed by heating of the printed part. Light curing can introduce heterogeneity within large 3D printed objects due to non-uniform light penetration or optical hinderance by the presence of photoabsorber, whereas thermal curing may not be easily adaptable for hydrogels due to issues with water evaporation.

Unlike DLP printing that requires multiple steps, the key parameter governing printability in VAM is the threshold energy required to induce crosslinking. Identification of the workable light dose range can be conducted via projection of an array of similar objects (i.e., disc-shaped spots) with varying light intensities and exposure times and the lowest dose required to obtain an object across the entire thickness of the vat can be denoted as the optimum exposure condition for that resin formulation^52,92.

Lithography-Based Printing in Biomedicine

The processing of hydrogels with photolithography is used across many applications ranging from the production of micro-architected metal alloys⁹³ to cultivated meat⁹⁴. In this section, we focus on biomedical applications within the purview of (i) 3D scaffolds for tissue engineering and regenerative medicine and (ii) 3D in vitro models for disease modeling and drug screening. These applications may include the incorporation of multiple cell types or matrix components in a predefined spatial manner, the engineering of porosity or microchannels to guide cell infiltration, or the specific placement of therapeutic molecules. While not meant to be an extensive review on these topics, we provide a few key examples where hydrogels fabricated via DLP or VAM have been leveraged in biomedicine.

3D scaffolds for tissue engineering and regenerative medicine:

Lithography-based 3D printing can produce high throughput and highly precise scaffolds designed to either encapsulate or be seeded with cells (cellular approach) or facilitate cell recruitment (acellular approach). Such hydrogel-based medical devices and implants can have broad applications in the fields of tissue repair, tissue adhesives, personalized biomaterials, or organ-scale manufacturing (Fig. 4a). In the context of personalized medicine, 3D designs can be generated through high-resolution medical imaging (e.g., magnetic resonance imaging, MRI and computed tomography, CT) and then printed through vat photopolymerization in mere seconds to minutes with extremely high precision and applied as tissue replacements or templates for the fabrication of medical devices (e.g., adhesives, patches)^95,96.

Figure 4 |. Potential applications of lithography-based 3D printing of hydrogels.

Hydrogels fabricated from a range of natural or synthetic polymers and that either include cells or are meant to interact with cells have been printed with photolithography methods. Vat photopolymerization can be applied to a | create personalized biomaterials, design scaffolds for tissue repair, fabricate tissue adhesives, and manufacture synthetic organs within the realm of tissue engineering and regenerative medicine; b | develop 3D in vitro (healthy or diseased) models of (i) tissues (e.g., lung, intestine, heart, liver) in the form of (ii) patterns or devices to diagnose onset, study disease progression or screen therapeutics.

When cells are included, hydrogels allow the engineering of suitable cellular microenvironments, such as to mimic the mechanics of tissues (e.g., cornea^97,98, myocardium⁹⁹, osteochondral¹⁰⁰, fibrocartilage⁵¹). However, the direct 3D printing of exceptionally soft hydrogels (0.1 to 1 kPa) is difficult with traditional AM methods. Owing to low shear forces and the need for viscous resins that maintain cell suspensions, VAM has been popular in recent years to print cell-laden soft hydrogels^51,52,101. Soft hydrogels can also be printed via DLP, particularly by printing stiffer materials and then softening post-printing¹⁰². Printed hydrogels may be designed to permit cell spreading, such as presence of gelatin along with thiol-ene crosslinkable PVA that create stress-relaxing environments³³. In some applications, soft hydrogels may be challenging to implement as scaffolds (e.g., load bearing in musculoskeletal tissues¹⁰³) and therefore recent strategies have focused on mechanical reinforcement through 3D printed interpenetrating polymer networks (IPNs) or semi-IPN hydrogels that exhibit superior mechanical properties^47,53,104. For instance, biopolymer-based 3D IPN hydrogels were fabricated in a one-pot system where a single photoinitiating step could induce orthogonal chain-growth and step-growth reactions⁵³. Despite the dynamic nature of the tissue microenvironment beneficial for cell spreading and outgrowth, most hydrogels have been purely elastic¹⁰⁵. Therefore, future research can further delve into light-based 3D printing of hydrogels with viscoelastic, stress-relaxing behavior.

The tunable nature of hydrogels also allows for controlled release of therapeutics, where the benefits of 3D printing related to the degree of freedom over shapes, which surpasses the limitations of traditional molding techniques^106–108. Like cells, therapeutics to be delivered can be encapsulated during printing or incorporated post-printing¹⁰⁷. Therapeutics (e.g., small molecule drugs, proteins) can be dissolved directly (i.e., to form a solution) or dispersed (i.e., to form an emulsion) within the photocrosslinkable resin. In this case, additional characterization may be warranted to ensure stability of the therapeutic against photolysis and free radicals. Incorporation after printing can also occur through swelling (i.e., absorption) of the printed object in a solution or via dip or spray-coating (i.e., adsorption) of therapeutic. As an example of 3D printed hydrogel for delivery of therapeutics, DLP-printed and compacted drug-eluting tough hydrogel stents were delivered to the vascular site with a medical guide wire and expanded based on programmed temperature responsiveness¹⁰⁹.

Vascularization plays a pivotal role in nutrient and oxygen diffusion of tissues and is important to incorporate into tissue engineered constructs to prevent necrosis and cell death^110,111. Lithography-based printing provides a great tool in which to introduce vascular structures within hydrogels¹¹². For example, prevascularized hydrogels (e.g., tissue constructs¹¹³ or microgels¹¹⁴) with precisely controlled distributions of multiple cell types (e.g., HUVECs, 10T1/2, hMSCs) can be 3D printed without the need for sacrificial materials and show anastomosis when implanted or injected in vivo. Using a similar approach, printed hydrogel carriers with hepatocyte aggregates survived transplantation in a rodent model of chronic liver injury¹⁸. Angiogenic patches encapsulating multiple cell types (i.e., fibroblasts, myoblasts, endothelial, and bone marrow stromal cells) were also printed via a micro-stereolithography projection apparatus to promote growth of neovasculature¹¹⁵. Vascularization of implanted hydrogels in vivo is challenging and remains an area of open exploration, likely through harnessing complex hydrogel resins (e.g., type of polymer, use of ECM proteins, stiffness) that encourage cellular invasion and vascular network formation. For perfusion of channels within bioprinted structures, custom bioreactors may also be needed¹¹¹.

Porosity and surface topography are important design features of acellular tissue engineering scaffolds to facilitate recruitment and invasion of cells, allowing for the proper integration of scaffolds over time. In more traditional bulk hydrogels, the nanoscale mesh size restricts cellular invasion, constricts embedded cells, and limits matrix deposition and elaboration. With photolithography, porosity (pore size and distribution) can be easily engineered by designing features within the CAD file of the object to be printed that can be infilled with cells or matrix to encourage cellular invasion¹¹⁶. Other alternatives have used resins that include nanoparticles¹¹⁷, aqueous immiscible bioinks¹¹⁸, microgel-templated porogels¹¹⁹, or gas bubbles to introduce porosity¹²⁰. For instance, emulsion bioresins formed by two immiscible aqueous phases can result in the generation of pores within hydrogels after photocrosslinking when the sacrificial phase is subsequently removed¹¹⁸. Similar phase-separating resins that support cellular spreading have been utilized to form centimeter-scale macroporous hydrogels within seconds via tomographic VAM¹²¹. Other methods to create porosity include the printing of 3D templates that could be molded into soft hydrogel substrates and then seeded with cells¹²² or the printing of salt structures that can be dissolved away to introduce porosity¹²³. While engineering by design within the CAD file provides control over the geometry, alternative strategies based on resin formulation can obtain pore sizes difficult to achieve with 3D printing and this distinction should be taken into consideration before printing.

Towards the goal of organ-scale manufacturing, lithography-based 3D printing offers a means of constructing precise and mechanically robust hydrogel-based biomimetic tissues such as patient specific artificial heart valves with hemodynamic function¹²⁴. Beyond hydrogel-based prosthetics, light-based 3D printing can also be used to create organ phantoms that mimic the spatial structural complexity and mechanical properties of diverse soft tissues such as kidney, brain, heart, liver, stomach, lung, trachea, and intestine¹²⁵. Such phantoms can serve as effective surgical training platforms that simulate clinical interventions and in vitro medical device testing.

While implantation is the common approach for 3D printed hydrogels for tissue repair^126,127, this requires a surgical procedure. As an alternative, the groundbreaking potential of non-invasive 3D bioprinting in vivo was recently demonstrated through digital near-infrared (NIR) photopolymerization (DNP)¹²⁸ or NIR laser associated with intravital multiphoton microscopy¹²⁹. Both methods use irradiation of NIR (at wavelengths longer than 850 nm) owing to the high tissue penetration of light to allow spatial polymerization of injected monomer solutions. As a proof of principle, a bioink (with up-conversion nanoparticles, UCNP coated with LAP photoinitiator) containing chondrocytes was subcutaneously injected into mice, followed by crosslinking into an ear-shaped construct that could be maintained for up to 1 month in vivo¹²⁸.

The growing interest in programming cells with synthetic biology tools has opened new avenues towards designing engineered living materials^130,131. However, technologies to shape engineered living materials into complex 3D geometries are still rather limited. The synergy of synthetic biology and light-based 3D printing can be advantageous for creating open architectures for the growth and expansion of genetically engineered microorganisms (e.g., mycelium, yeast) into a functional material outside of the host environment using hydrogel resins^132–135. Interestingly, in addition to DLP, VAM can be used to create multi-bacteria functional devices for chemical sensing. As an example, bioluminescent bacteria Photobacterium kishitanii was encapsulated within the core of the hydrogel and surrounded by melanin producing E. coli¹³².

3D in vitro models:

The cost of drug discovery and development remains high, in part, because of the lack of translation between 2D in vitro cell cultures and animal testing and the response in human patients¹³⁶. Engineering cellular systems that cater to the physiological signatures of patients in 3D could help bridge this gap. Specifically, 3D bioprinted tissues can serve as a platform to screen drugs with patient-derived cells, to include transcriptional profiles that enable prediction of drug performance (Fig. 4b). In addition to screening, in vitro 3D models have gained popularity in modeling healthy or diseased tissues to study development and disease progression¹³⁷. Such platforms allow investigation of cellular crosstalk and cell–ECM interactions in a controlled manner, which is deconvoluted from the milieu of signaling cascades present in vivo to allow the uncovering of novel pathways.

Towards these models, lithography-based bioprinting enables the rapid patterning of cells (e.g., hepatocytes, neurons, ventricular cardiomyocytes, endothelial cells, macrophages) into distinct regions that recapitulate the functional unit of native tissues (e.g., lung¹³⁸, liver^139–141, myocardium¹⁴², bone^143,144). Further, changes in light exposure allow for spatial changes in mechanical stiffness within these in vitro models, towards simultaneous control over both biophysical properties and biochemical cues. Beyond direct cell encapsulation during printing, a cell-laden matrix can also be seeded within 3D printed hydrogel structures to create microtissues, such as with the seeding of fibroblasts or MSC laden collagen gels within DLP-printed micropillar hydrogel molds^53,145. The cells then remodel the surrounding matrix to form the microtissue along the predefined boundary imposed by the micropillar. Towards cancer models, the co-encapsulation of patient-derived tumor or stem cells can help elucidate the cellular crosstalk important for therapeutic design¹⁴⁶. For example, a glioblastoma model was created with three distinct zones of defined stiffnesses: (i) patient-derived tumor region, (ii) acellular ECM region (i.e., HA) to mimic the brain parenchyma, and (iii) human endothelial cells to mimic blood capillaries¹⁴⁷.

Beyond single cells, organoids or self-assembled aggregates can be included during printing to better model tissue morphogenesis and the corresponding cell–cell interactions and cellular self-assembly¹⁴⁸. For instance, hepatic organoids were VAM printed into mathematically defined lattices with varying degrees of pore network, tortuosity, and cultured under perfusion⁸¹. These structures acted as liver-like metabolic biofactories for albumin synthesis and ammonia detoxification, opening new possibilities for 3D drug testing. Liver-inspired devices fabricated from 3D printed hydrogel nanocomposites can also be utilized for the collection and sensing of toxins¹⁴⁹. Maskless projection afforded by vat photopolymerization techniques can be a promising tool for rapid micropatterning of hydrogel substrates to study stem cell organization and signaling in developmental biology¹⁵⁰.

Organs-on-a-chip platforms have revolutionized research in the fields of drug testing and disease modeling¹⁵¹. However, these have heavily relied on the use of multi-layer lithography, followed by often multiple rounds of casting with poly(dimethyl siloxane) (PDMS) elastomer to produce features of interest. This process is time-consuming, not compatible with large scale manufacturing, limited for complex geometries, and the PDMS must be treated to avoid non-specific absorption of drugs or small molecules¹⁵². To overcome these limitations, hydrogels and vat photopolymerization provide an attractive opportunity^153,154. Additionally, the alignment of dissimilar layers can be performed automatically during the printing process and optical clarity for imaging is apparent with many hydrogels¹⁵⁵. As one example of DLP-printed microfluidics, the 3D patterning of gene expression was realized through fluid-based heat transfer from volumetric networks to activate transgenes expressed by cells embedded within the hydrogel¹⁵⁶. With these advantages in mind, light-based printing of high resolution, transparent, and ready-to-use hydrogel devices with integrated capillary circuits also hold great promise for immunoassays and point-of-care diagnostics^157,158.

Biological structures often interface with different cell populations and microenvironmental features. Although recapitulating native tissue-like complexity remains a major challenge in the field of bioengineering, there is still fundamental understanding (i.e., cell signaling, crosstalk, regulation) that can be gained from current 3D printed in vitro platforms. An important biofabrication question is how complex is complex enough and researchers must discern the level of biological complexity needed based on the desired application. Additionally, in vitro models fabricated via lithography-based 3D printing should allow the integration of external (e.g., mechanical, electrical) stimulation and the direct measurement of functional outcomes.

Enhancing Complexity in Lithography-Based Printed Structures

Despite a desire to mimic complex features across tissues and cellular environments, most examples of lithography-based printed hydrogels discussed thus far are still comprised of a single material. To address this, multi-material vat photopolymerization techniques are now emerging to provide additional degrees of freedom with respect to the patterning of cells and materials in 3D (Fig. 5a)^159–161. The simplest approach is through print-pause-print methods where resin baths are switched between layers¹⁵⁵. While this approach can be directly adopted without the need for printer modifications, it requires significant user attention during the multiple resin switching steps. Methods to streamline this process include use of a motorized silicone sled¹⁶² or automatic switching of resins via dynamic fluid control with microfluidic devices¹⁶³, integrated fluidic cells¹⁶⁴, or rotation of the vat-containing resins^165,166. This automated method was successfully leveraged to create bilayer, multi-responsive (i.e., thermo-responsive and electroactive) hydrogel structures that undergo varied degrees of bending deformation depending on the stimuli applied. To improve washing between subsequent resins, processes that leverage mechanical agitation or centrifugal force driven material switching have been adopted¹⁶⁷. Further, microfluidic chaotic mixers can be integrated to form continuous or discrete compositional gradients of resins in real-time during printing¹⁶⁸.

Figure 5 |. Enhancing the complexity of 3D printed hydrogels.

a | Schematic representation of approaches towards multi-material (in the xy and z-direction) vat photopolymerization. b | Examples of lithography-based printing combined with other fabrication techniques (e.g., microfluidics, extrusion printing, melt-electrowriting MEW, multi-photon based laser ablation). c | Hydrogels can be post-functionalized after printing with biochemical cues (e.g., growth factors, peptides) or undergo spatial stiffening or softening to design responsive materials. d | Grayscale 3D printing of hydrogels to control mechanical heterogeneity or create gradients in biochemical cues. e | Illustration of vat photopolymerization with multiple synergistic or orthogonal wavelengths of light. f | Designing responsive materials through 4D printing of hydrogels that undergo dynamic changes in shape or mechanical properties over time via an external trigger (e.g., heat, pH, secondary reaction).

In addition to multi-material structures, hydrogels with multi-directional organization are needed to better recapitulate heterogenous tissue architectures¹⁶⁹. Thus, methods to control directional stiffness (i.e., tunable negative Poisson’s ratio) can be leveraged to design auxetic patches that conform to curved surfaces and dynamic movements of organs (e.g., heart, lung)^170,171. Multi-directional alignment can also be accomplished with a modified approach of multiple, consecutive filamented light (Cross or Multi-FLight) beam projections from different directions into a photoresin vat, which has been used to fabricate bilayer hydrogels with distinct zones of alignment to direct anisotropy of engineered cartilage in vitro^84,172. Additionally, multi-axis 3D printing can be undertaken with integration of a six-axis robotic arm into microcontinuous liquid interface production (μCLIP) for freeform fabrication of vascular scaffolds to fit patient anatomy¹⁷³.

The combination of multiple biofabrication techniques is now being used to expand printed material properties, such as anisotropic mechanics and multi-material structures (Fig. 5b). One simple approach has been the use of surface-tension assisted coating of 3D printed supports with cell-laden polymer solutions, essentially dipping printed structures into the solutions and crosslinking to produce multi-component constructs with mechanical anisotropy¹⁷⁴. As another approach, structures fabricated with melt-electrowriting (MEW) have been introduced into resin baths and hydrogels have been printed around the structure, both for mechanical reinforcement and to mimic structures commonly present in the renal glomerulus, intestinal blood vessels, and nasal mucosa¹⁷⁵. This technique harnesses the unique ability of VAM or CAL to support overprinting: a method of crosslinking material around a complex, preexisting 3D structure. Due to the presence of highly opaque materials such as poly(caprolactone) (PCL) that can block the projection path and result in significant light attenuation, refractive index matching solutions may often be needed. Suspension baths have also been used recently to print cells or materials in 3D, followed by VAM printing to define the 3D shape. For example, granular hydrogels – jammed hydrogel microparticles – that undergo shear-thinning and self-recovery supported the deposition of cells for subsequent VAM printing¹⁷⁶. As a proof-of-concept, optogenetic-engineered pancreatic β-cells were co-cultured with pre-adipocytes to investigate intercellular communication in the context of metabolic regulation. Gelatin has been used for suspension bath printing of materials, prior to VAM printing³⁴. Extrusion-based deposition of multiple photo-crosslinkable hydrogels, thermoplastic melts, or fugitive inks can be combined with high resolution DLP-based patterning to create heterogeneous multi-layer structures that mimic the esophagus¹⁷⁷, hybrid vascular conduits¹⁷⁸, or channels for nutrient diffusion²⁸. Subtractive manufacturing can also be combined with vat photopolymerization to fabricate multi-scale constructs with centimeter (VAM) to micrometer (two-photon ablation) features, such as vessel networks with branching microcapillaries (up to 2 μm) that are ablated and perfused within VAM printed hydrogels⁸⁵. Similarly, hybrid laser printing converges multiphoton ablation with CLIP via sequential additive-subtractive steps. This step-by-step technique can create hollow features within a 3D hydrogel at any depth, which is an improvement over the flood illumination of optically thick 3D hydrogels that can limit light penetration¹⁷⁹. Combining vat photopolymerization methods with advanced patterning techniques such as those based on electrical¹⁸⁰, magnetic¹⁸¹, or acoustic¹⁸² can also have significant impact on improving the functionality of the hydrogel.

In addition to the direct printing of multi-material structures or patterns, the post-functionalization of printed objects is possible to further expand on the biochemical and biophysical signals within hydrogels (Fig. 5c). For example, when using thiol-ene chemistry, the degree of crosslinking (via thiol crosslinker concentration) or incorporation of a thiolated peptide or protein can be controlled through the thiol concentration relative to the alkene functional groups. This permits precise spatiotemporal patterning of unreacted alkenes (e.g., norbornene modified HA, gelatin, or PVA) with thiolated macromolecules (i.e., fluorescent peptides, growth factors) through maskless projection facilitated by DLP or VAM^{33,54,183,184}. As a proof of concept, vascular endothelial growth factor was locally patterned via VAM towards spatially selective cell adhesion and network formation of endothelial cells¹⁸³. Other chemistries can include photocages with sortase A-based enzymatic coupling or photomediated oxime ligation for 3D patterning of bioactive proteins within hydrogels^185,186. Dynamic mechanical properties (i.e., spatial softening) of printed hydrogels is also possible, such as with hydrolytically degradable hydrogels that soften over time for tissue integration or controlled molecule release⁵⁴. Future research in this area is warranted as selective photocrosslinking and photodegradation can be a powerful tool to dynamically control printed hydrogel properties¹⁸⁷. Lithography-based printing with photodegradable moieties (e.g., ortho-nitrobenzyl, coumarin) can be challenging due to the high optical absorption of such chemical groups and potential degradation during crosslinking with light exposure. Hence, there is a need for orthogonal printing strategies that incorporate such light-sensitive groups within hydrogels. In addition to softening, dynamic stiffening of hydrogels post-printing can be realized through the formation of secondary crosslinks or incorporation of a secondary polymer network¹⁸⁸.

Biomimetic structures (e.g., bone to tendon enthesis) consisting of soft and hard phases with composite-like mechanical property mismatch are challenging to print via simple vat photopolymerization. Tuning the exposure time for light illumination permits regional variations in stiffness or functionalization of biochemical cues without the need for changing the geometric architecture of the printed object. For example, a grayscale gradient in light intensity can locally increase the conversion of resin to form (highly stretchable) soft to (highly brittle) stiff regions within a single print layer of a hydrogel (Fig. 5d). These encoded structure gradients can be leveraged for swelling-induced morphing of hydrogels into complex configurations^189,190. Furthermore, grayscale printing can overcome the challenges of multi-material processing that require low throughput and complex vat switching and cleaning steps. Towards further complexity, the incorporation of dyes that change color in the presence of free radicals (e.g., anthraquinone-based dye) dependent on the light dose during printing can enable a wide spectrum of properties¹⁹¹.

Another interesting way to control chemical and mechanical properties in 3D involves orthogonal multi-color mode polymerization, where multiple light sources are used with wavelength-responsive resins to spatially guide curing (Fig. 5e)^192,193. For instance, the use of orthogonal acrylate and epoxide chemistry allows selective free-radical polymerization of acrylates at wavelengths higher than 385 nm, whereas UV irradiation (below 385 nm) results in cationic polymerization of the epoxides¹⁹⁴. Spatial control over chemical composition also enables mechanical heterogeneity, which drives swelling anisotropy towards shape-morphing applications¹⁹⁵. In addition to orthogonal crosslinking chemistries, multi-functional printing can also be undertaken with linear excitation of photoinitiators with two different wavelengths through an emerging technique known as xolography¹⁹⁶. Xolography uses a dual-color photoinitiator that is activated by the first wavelength, while the second wavelength results in initiation and inhibition of the reaction, respectively. Wavelength-selective light response within hydrogels can also be designed through post-printing incorporation of chemical moieties or by embedding various dyes that can absorb in different spectral regions (e.g., azo dye, 459 nm; IR780 dye, 850 nm)¹⁹⁷. These technologies are promising for the rapid production of 4D functional and multi-responsive materials using multi-color or multi-wavelength projection; however, challenges remain in the design of aqueous, cytocompatible resins, which has limited their adoption for hydrogels. Further, most methods for light-based printing of hydrogels have been limited to 385 nm and 405 nm wavelengths, with very few reports of vat photopolymerization at 525 nm⁵⁶. Shifting the 3D printing spectrum to longer wavelengths (e.g., 615 nm) can expand the scope of processable objects; however, it is currently limited by the available aqueous photo-chemistries and slow reaction times^198,199. Hence, there is a need for alternatives (e.g., use of coinitiators) to address this limitation.

There is growing interest in moving the complexity of printed hydrogels to the fourth dimension (i.e., time) (Fig. 5f). Post-printing, a secondary monomer can be incorporated via swelling within a 3D printed hydrogel and crosslinked through secondary reactions (e.g., alkoxyamine chemistry) to create adaptable materials that can undergo growth or degrowth with corresponding changes in size and mechanical properties²⁰⁰. Another important area for 4D printing is the design of intelligent actuators, artificial muscles, and grippers for biomedical applications²⁰¹. Using multi-material DLP, thermo-responsive, microporous hydrogel actuators can be 3D printed with N-isopropylacrylamide and acrylamide monomers²⁰². These actuators exhibit autonomic perspiration such that at temperatures lower than the LCST (lower critical solution temperature), the pores are sufficiently closed to facilitate pressurization but at higher temperatures (> 30 °C) the pores dilate to enable ‘sweating.’ In addition to external triggers (e.g., temperature²⁰³, pH^204,205, swelling^206,207, or dehydration^208,209) for lithography-based 3D printed hydrogels, responsiveness can also be programmed through diffusion path architecture to tune swelling kinetics²¹⁰ or encoded mechanical anisotropy during printing²¹¹. For instance, printed hydrogel scaffolds with body temperature-triggered shape memory and water-triggered programmable deformation can be delivered through a transcatheter for implantation in a minimally invasive manner²¹².

Outlook

Despite the many great examples of lithography-based printing of hydrogels mentioned throughout this review, there is still much ahead for this field. While vat photopolymerization techniques are attractive to print materials as in vitro models or for therapeutics, there are still many challenges to mimic the complexity of biology (e.g., incorporation of multiple, tissue relevant cell types, 3D architectures) in a scalable and reproducible manner. To address these and other challenges, the following sections highlight areas where we foresee advances in the near future.

Expanding the library of available hydrogel resins.

Most vat photopolymerization resins have been limited to chain-growth crosslinking with meth(acrylates) and meth(acrylamides) (e.g., GelMA, PEGDA); hence, the development of new hydrogel resins that leverage underexplored chemistries (e.g., photo-mediated strain-promoted azide–alkyne cycloadditions and photo-redox reactions) would be beneficial. Moreover, the development of resins that result in properties (e.g., stiffness, toughness, dynamic reversible interactions) beyond the typically soft and brittle nature of hydrogels would unlock avenues for a wide range of applications^213,214. There is also a need for highly efficient, rapid photopolymerization chemistries and post-curing strategies to achieve near complete conversion of reactive species. This is to avoid any potential toxicity of unreacted monomers and macromers due to incomplete conversion. Further, most vat photopolymerization strategies described in this review require the use of a photoinitiator to generate free radicals, which can cause potential cytotoxicity in the context of hydrogel bioresins. Thus, investigation into radical-free photocrosslinking mechanisms would be highly beneficial. Some recent examples include two-photon uncaging of thiols and subsequent Michael addition with alkenes²¹⁵ or the use of UV-induced photodimerization of lipoic acid functional groups²¹⁶. Further, transformative impact in the development of new resins lies at the unexplored nexus of biomaterials with synthetic biology (e.g., protein, DNA, genetically engineered cells or matrix)²¹⁷. As new photo-chemistries are being developed, special emphasis should also be placed on sustainable synthesis routes to create resins and then test their cyto- and biocompatibility both in vitro and in vivo. It is also extremely important to consider the age and sex of both cells (in vitro) and animals (in vivo) used in the studies to evaluate printed hydrogels.

Harnessing the power of light towards precision biomaterials.

The unique ability of 3D printing to spatially control the positioning of cells and materials best positions the technology to recapitulate key functions of biological systems. When designing hydrogels for a specific application, the choice remains between cell-instructive (i.e., top-down fabrication of tissues) or cell-permissive (i.e., allow spontaneous self-assembly) approaches. Taking a developmental biology-inspired approach of allowing a system to build itself autonomously, through the remodeling and self-organization of cells and matrix into complex patterns can be huge step forward. On the other hand, the combination of medical imaging with high-speed, high-resolution lithography-based printing will be a defining factor towards personalized biomaterials⁹⁵ and biofabrication of functional organs in the future²¹⁸. This will require the development of hydrogels that undergo minimal swelling, possess strengths similar to native organs, and facilitate housing of various cell types, including towards vascularization and innervation. Exploration of on-demand, in situ or in vivo bioprinting of cells and hydrogels based on the medical images of the defect in real time will also be a paradigm shift, which will require wavelengths such as NIR due to its high light penetration and low scattering within deep tissues²¹⁹.

While the biological compatibility of commonly used biomaterials (e.g., PEG, gelatin, hyaluronic acid) has been evaluated in the past, it is equally important to confirm that the printing process does not adversely impact cell viability and cellular behaviors (e.g., differentiation, proliferation, quiescence). The major advantage of 3D printing to impart complex architectures (e.g., surface topography, curvature, pore size, shape) provides another avenue to control biological response in vivo (e.g., host-immune response, tissue integration). This structure-property-response has been understudied but should be evaluated in the future. Additionally, combining light-based 3D patterning with optogenetic tools can enable real-time monitoring of changes in gene expression in microphysiological setting to study fundamental pathways involved in disease.

Navigating the regulatory and commercialization barrier.

Commercialization will require accessible methods to source and synthesize biopolymers and their quality assurance (e.g., testing for endotoxins) to ensure regulatory compliance, as well as large sources of cells for bioprinting^220–222. Scalability, an integral consideration for additive manufacturing, is currently limited by the size of the build platform for DLP or the volume of the rotation vial for VAM. In the case of DLP, large xy- prints can be obtained with a large projection area by stitching projected images from multiple DMDs; however, this may result in undesired heterogeneity within the printed object. One approach has been the combination of a spatial light modulator with an optical scanning system to produce microscale architectures over a substantially larger area²²³. An alternate approach would be the translation of the DMD over the build platform that can raster and stitch projected images into the resin vat as it moves in the xy- direction. To enable high-throughput screening or manufacturing of identical copies of hydrogels, vat photopolymerization can be adapted to (i) automate printing into well plates using top-down light projection through a hollow guidance probe²²⁴ or (ii) emerging roll-to-roll rapid production via CLIP²²⁵. Although the latter has not been implemented for hydrogels yet, it has the potential to revolutionize manufacturing throughput.

Commercialization and large-scale lithography-based additive manufacturing of hydrogels can benefit from efforts to automate print preparation (with user-friendly design engines and machine learning algorithms to optimize layout of multiple parts within the build volume), real-time monitoring of the printing process, and post-processing (with seamless integration of washing and curing of the printed object). These analytic and hardware approaches can save time, reduce costs, and boost efficiency. Further, implementation for clinical use will require the printing process to abide by good manufacturing practice (GMP) requirements as well as appropriate infrastructure for sterilization of 3D printed hydrogels. The unreacted resin (not incorporated into the printed object) can often be reused to improve efficiency and reduce waste; however, the printing process may expose the resin to conditions (e.g., light, heat, oxygen, humidity) that may alter the functionality compared to the virgin or freshly prepared state. Hence, characterization must be performed to test that the resin reuse does not adversely impact the properties of the printed object.

There are numerous opportunities ahead towards clinical translation of lithographically-printed hydrogels; however, there are hurdles to overcome (Box 2). US Food and Drug Administration (FDA) has already provided regulatory guidelines for tissue-engineered constructs²²⁶ as well as additive manufactured medical devices²²⁷ that outlines best practices for CAD design considerations, material controls, and print validation. Similar framework and legislation are needed for the large-scale manufacturing process, including reproducibility, manufacturing tolerances, use of biopolymers or biologics, and sterility for implants²²⁸. Quality control and assurance can rely on non-destructive testing (e.g., ultrasound, micro-computed tomography, or confocal imaging) to evaluate the geometry and morphology of the printed hydrogel.

Box 2 |. Translational Considerations.

Medical products fabricated with vat photopolymerization have been approved for use; however, none of these products have been comprised of hydrogels or bioprinted “living” tissues. Lithography-based printed hydrogels will likely find greatest impact in point-of-care applications (e.g., personalized implants) and as in vitro screening platforms, owing to their high resolution and rapid printing times. Additionally, lithography-based printing will be most useful when traditional methods of molding or implant sizing by surgeons is not possible²³⁵.

There are advances being made in printing processes that will help translation. For example, new methods such as CLIP and VAM have expedited the time required for printing objects, making it theoretically possible to print human organ scale materials during a surgical procedure (minutes to hours). Additionally, the development of resins with highly efficient chemistries and methods that eliminate the need for small molecules (e.g. photo-initiators, absorbers), will simplify printing and potentially improve the safety profile of printed hydrogels by eliminating purification or even possibly allowing in vivo printing.

Beyond advances in lithography-based printing processes, our increased understanding of printed material structures and the types of materials that can be printed will help expand their use in medicine. For example, immune responses can be helpful or harmful depending on the context and the high resolution afforded by lithography-based methods should enable control over implant microstructure to modulate the immune response²³⁶. Additionally, many printed hydrogels have weak mechanical properties that make it difficult for surgeons to handle, so an increased focus on the fabrication of tough hydrogels with lithography-based printing could greatly expand their translation. As a last example, lithography-based printing can allow for single-step integration of multiple functionalities (e.g., conductive materials) into hydrogels, which could be used as sensors²³⁷ or for tissue adhesion²³⁸, which should open new translational applications.

Although there is limited clinical exploration of lithography-based hydrogels, government support from a range of initiatives is likely to strengthen efforts across many tissues and diseases. Importantly, the US government has recently launched programs (e.g., AM Forward) to facilitate the growth of small and medium-sized 3D printing companies, invested in funding transformative biomedical breakthroughs (e.g., ARPA-H program), and passed the FDA Modernization Act 2.0 which identifies in vitro models (e.g., predictive microphysiological systems) as sufficient alternatives to pre-clinical animal testing in certain situations²³⁹. Indeed, with continued regulatory oversight lithography-based printing will likely play a role in the clinical translation of new medical devices and therapies.

Key Points.

1. Lithography-based 3D printing can achieve high-resolution structures without compromising on the speed of fabrication.

2. Hydrogels can be processed through a range of lithography-based methods to control hydrogel structure, as well as biochemical and biophysical properties.

3. Lithography-based printing involves a sequence of steps from resin formulation through post-processing.

4. Printed hydrogels are promising for applications from tissue engineering to in vitro models for drug screening.

5. Emerging advanced lithography-based methods are increasing the complexity of printed structures to expand to new applications.