Chapter 1: Microstereolithography Free

Fabrication is a critical process in making materials into functional parts and devices. Traditional fabrication technologies such as machining and molding are commonly used in macro-scale three-dimensional (3D) fabrication. However, they are not adequate for microscale fabrication. Modern micro- and nanoscale fabrication technologies such as photolithography, soft lithography, electron beam lithography, focused ion beam lithography, dip-pen lithography, and atomic layer deposition, are often 2-dimentional (2D) in nature for thin film and surface patterning.

3D printing, as an additive free-form 3D fabrication technology, has achieved great commercial success in the past decade, because of its low cost, simplicity, and versatility. Microstereolithography is a light-assisted 3D free-form fabrication technology. This technology utilizes photosensitive materials, which can solidify upon ultraviolet (UV) or short wavelength visible light exposure. By spatially controlling the exposure dose, the desired 3D structure can be fabricated. This technology evolves in two directions. One, scanning-based microstereolithography, which provides the extremely fine resolution (sub-micron scale) but slow fabrication speed (e.g., hours). Two, projection-based microstereolithography, which provides both fine resolution (micron scale) and fast fabrication speed (e.g., seconds to minutes).

Microstereolithography is a powerful tool for biofabrication.¹  It has successfully demonstrated its capability of fabricating a wide range of biomaterials such as hydrogels, proteins, and cell-laden materials. Both synthetic and natural materials can be used to print, each having different advantages for stability, mechanical properties, cytocompatibility, and printability. Material decisions must come from the eventual application for the printed structure, as even the choice of photoinitiator can greatly impact the print.^2,3 

This chapter covers basic physical and chemical mechanisms in photopolymerization, materials, devices, and systems of microstereolithography. The photopolymerization mechanism is detailed in Section 2. Materials for microstereolithography are discussed in Section 3. Scanning-based microstereolithography, including single-photon polymerization microstereolithography, and two-photon polymerization nano-stereolithography, is detailed in Section 4. Projection-based microstereolithography, including liquid–air interface polymerization and liquid–substrate interface polymerization, is discussed in Section 5.

In this section, we will focus on photopolymerization, a technique most often used to crosslink liquid state monomers or oligomers into solid state long-chain polymers. Photopolymerization uses free radicals to initiate and crosslink strands within a monomer solution to form a solid hydrogel. When paired with microstereolithography techniques, various complex structures can be fabricated.^4,5 

Two of the main types of polymerization observed in hydrogel scaffolds are step-growth and free-radical polymerization. Although photopoylmerization methods use free radicals to polymerize structures, the kinetics of step-growth polymerization can describe some of the more unique polymerizations, and thus it is important to cover.^6,7  Step-growth polymerization occurs when polymer chains grow in a stepwise fashion either by condensation reactions, in which water is removed, or when reactive end groups interact. When considering simple linear chain reactions, the mechanisms and rates of all polymerization steps can be assumed as equal. Moreover, the Carothers equation (eqn (1.1)) defines the level of completion of the step-growth polymerization

where x_n is the average chain length, and p is the conversion rate of the monomers into polymers. This is an incredibly useful equation for predicting how long the reaction needs to proceed to obtain the correct molecular weight.⁸  As one can see, this defines an exponential relationship (Figure 1.1).⁹  As one might imagine, high molecular weights become increasingly more difficult to achieve for three reasons: (1) the frequency of reactive end groups meeting decreases; (2) the frequency of side reaction interference increases; and (3) when two or more monomer types are used in the reaction, it is difficult to ensure that the starting material concentrations are equal. This third point is an issue for users creating copolymer hydrogels, such as those consisting of monomer A and monomer B, where A-A does not react, nor B-B, but only A-B. To stop the reaction at a lower molecule weight, the user has a few options. One, the reaction can be rapidly cooled at the correct time point to slow down the polymerization rate as many step reactions have high activation energies. Two, a monofunctional material can be added to “cap” the end of the polymer, preventing it from further reactions. Lastly, if making a copolymer, a stoichiometric imbalance of the starting materials can be used. For example, if more A groups are used than B, eventually the polymer will have two A end groups on a chain and the B monomer will be completely consumed preventing further reactions with that polymer. The Carothers equation can be expanded to define this scenario, shown in eqn (1.2),

where x_n is the average chain length, p is the conversion rate of the monomers into polymers, and r is the ratio of monomer A to monomer B (N_A/N_B).⁸ 

In free-radical polymerization, a free radical interacts with reactive end groups to form a polymer. The basic structure of the active group is CH₂CR₁R₂, as the pi-bond in the carbon–carbon double bond allows it to be rearranged when exposed to a free radical. From here forward this will be referred to as the active center. In the polymerization scheme, there are three main steps: initiation, propagation, and termination. As will be expanded on later in the chapter, the initiation step is the activation of the active center when a free radical reacts with the carbon–carbon double bond. After initiation, the active center reacts with other double bonds, propagating the chain to form a polymer and transferring the free radical to a new active center. The termination step can then occur in several ways: (1) two active centers can react; (2) one active center and one free radical can react; (3) the active center transfers to another molecule; or (4) interaction with impurities or inhibitors. Chain transfer is another important mechanism that occurs in free-radical polymerization. This occurs when an active center collides with a molecule such as a solvent, initiator or monomer, transferring the free radical to the second species.

Free-radical polymerization can theoretically go to full conversion, but the interaction of free radicals with each other must be kept in mind. Eventually, free radicals will form covalent bonds with each other, stunting the chain propagation and preventing full conversion. As a general rule, the greater the free radical concentration, the shorter the chain length. The viscosity also has an impact on the rate of the conversion, as it impacts the diffusion of polymer chains through the medium. As the conversion increases, so does the viscosity, preventing chains from interacting and terminating the reaction at the same speed.⁸  This means the propagation of chains by free radicals increases towards the end of the reaction (Figure 1.1).⁹  This auto acceleration can be prevented by stopping the reaction before the initiation and propagation steps become diffusion-mediated. Regardless of conversion, the average chain length of the completed polymer shows little variation throughout polymerization. Moreover, longer reaction times may increase polymer yield but will not increase the chain length. Increasing the temperature can decrease the molecular mass, but the best way to control the polymer chain length is to alter the initiator concentration.⁸ 

In “living” polymerization, the termination step is suppressed, so that the free radical is continually recycled. This results in a much lower polydispersity. Most living polymerization methods use a reversible “cap” on the active site, preventing it from continuously reacting. This slows down the reaction and suppresses termination, which results in a steady increase in chain length, defined by the following eqn (Figure 1.1):

where x_n is the degree of polymerization, [M]_o is the initial monomer concentration, [I]_o is the initial initiator concentration, and p is the conversation rate. Studies have been performed to try to improve reaction schemes for living polymerization.⁸  Two of the most common techniques are atom transfer radical polymerization (ATRP) and reversible addition fragmentation chain transfer (RAFT) (Schemes 1.1 and 1.2).^10,11 

A typical ATRP reaction consists of a dormant species, such as an alkyl halide (R-X) that can be reversibly activated by a transition metal (Mt^zY/L) to form an active radical (R˙) and an oxidized metal complex (XMt^z+1Y/L). When the free radical is not capped with the metal complex, it is free to propagate until it is capped again. ATRP methods are fairly versatile and work best with monomers such as styrenes, methacrylates, methacrylamides, and acrylonitrile. ATRP can also be used in conjunction with an initiator in a similar process called “reverse ATRP”.

For RAFT reaction schemes, the “capping” species contains a dithio compound and a good leaving group within its structure. The compound reacts with the free-radical functionalized polymer chain to form a dormant chain, and the leaving group is released as another free radical in solution. At an equilibrium state, the compound can react reversibly with up to two chains, rendering them dormant. This yields a much lower polydispersity. The molecular weight of the polymer can be compared to the conversion to distinguish the two types of polymerization (Figure 1.1).^8,12  Translating to hydrogel scaffolds, where the free radicals create cross-linking between polymer backbones, this means that a lower free radical concentration is needed for living polymerization than free-radical polymerization.¹³ 

In order to create the free radicals for photopolymerization, photoinitiators are added to the prepolymer solution. The photoinitiators used in hydrogels usually generate free radicals by one of two methods: (1) photocleavage or (2) hydrogen abstraction. In photocleavage, the molecule undergoes bond cleavage at C–C, C–Cl, C–O, or C–S bonds when exposed to light, generating free radicals. In method two, the photoinitiator undergoes hydrogen abstraction from an H-donor molecule, forming a ketyl radical and donor radical. When selecting an appropriate photoinitiator, the user must consider several factors including biocompatibility, solubility in water, stability, and cytotoxicity. Photoinitiators can greatly impact the print resolution, as well as cell viability. As such, preliminary studies modulating the concentration and exposure time of the print are necessary for any new photoinitiator considered.^14–16 

Photoinitiators can be characterized by the wavelength at which they most strongly absorb. Although less popular, visible light photoinitiators can be advantageous if the user plans on exposing the pre-hydrogel solution for a longer period of time. Compared with other photoinitiators that activate in the UV-range, encapsulated cells can be exposed for a longer time with less risk. However, these photoinitiators are more difficult to work with, as ambient conditions require the user to work in the dark. One popular visible light photoinitiator is Eosin-Y, which generates radicals by hydrogen abstraction. It is activated at 490–650 nm and is a common choice as a cytocompatible molecule. Unfortunately, in order to generate enough free radicals, it often needs a coinitiator (e.g., triethanolamine (TEOA)) and a comonomer (e.g., 1-vinyl-2 pyrrolidinone (NVP)) included in the prepolymer solution, making modulation of concentrations more difficult to work with than some of the common UV-initiators.^17,18 

UV-activated photoinitiators can be advantageous for users who desire easier ambient conditions. However, because the pre-hydrogel solution is exposed to UV, care must be taken to limit the exposure time when working with encapsulated cells in the gel. Preliminary exposure studies with cells are recommended to ensure limited cell death. Two common cytocompatible photoinitiators are Irgacure-2959 (I-2959) and lithium arylphosphanate (LAP), both of which generate free radicals by photocleavage. I-2959 remains a popular choice among users, but has some drawbacks when compared to LAP. I-2959 is not very water soluble (<0.5 wt%), and has low molar absorption at 365 nm (ε<10 M⁻¹ cm⁻¹), the wavelength usually used to excite UV-activated photoinitiators. As discussed previously, the lower molar absorption greatly impacts the polymerization rate, so that large amounts of photoinitiator or long, strong exposures must be used (eqn (1.1)). Due to its low solubility, I-2959 is usually incorporated at high concentrations into solely synthetic polymer systems with non-water based solvents, or exposed for long periods of time in natural polymer systems without encapsulated cells. LAP, on the other hand, has a high absorption at 365 nm (ε≈200 M⁻¹ cm⁻¹) and is very water soluble (at least 8.5 wt%), making it more conducive for work with natural, composite natural/synthetic polymers solutions, and encapsulated cell solutions. A comparison by Fairbanks et al. between I-2959 and LAP at 365 nm with comparable intensities and initiator concentrations found that the time to gelation was nearly an order of magnitude higher using LAP as the photoinitiator. Although commercially available, LAP can also be synthesized in-house nearly overnight, making it a reasonable option for users.^16,18 

One of the most common materials for representing the extracellular matrix (ECM) is the hydrogel, a solid material consisting of cross-linked polymer strands that can be tuned for stiffness, adhesiveness, and cell signaling potential. Hydrogel scaffolds can be printed using natural polymers, synthetic polymers, combinations thereof, and combinations with other materials such as carbon nanotubes or nanoparticles.³  Natural polymers are often more cell compatible, can have cell-controlled degradability, low immune response, and are good choices for studies involving direct cell encapsulation within the gel. However, synthetic polymers offer a greater amount of control over the scaffold shape, have better batch consistency, a greater range of mechanical properties and are much more robust overtime, making them good options for seeded cell studies.^2,3  By combining synthetic and natural polymers or materials in specific concentrations, scaffolds can be generated that are both cell compatible and have the appropriate physical, chemical and mechanical properties to support tissue growth in a biomimetic fashion.

There are several common natural polymers used for scaffolds, most common of which are hyaluronic (HA) and gelatin derivatives. Hyaluronic acid, or hyaluronan, is a polysaccharide that plays a key structural role within cartilage, as well as promoting cell motility and differentiation, and naturally degrades over time.¹⁹  Gelatin, or denatured collagen, is a major ECM component that is biocompatible and biodegrades overtime. Although denatured, it maintains some cell binding moieties including arginine-gylcine-aspartic acid (RGD) sequences, which promote cell attachment, and target sequences of matrix metalloproteinase (MMP) for cell remodeling.^3,14  One popular derivative for these materials is the functionalization of methacrylate groups to form methacrylated hyaluronic acid (meHA) and gelatin methacrylate (GelMA).^3,14  These natural polymers undergo radical polymerization at rather mild conditions; room temperature, neutral pH, and aqueous environments that are favorable for stereolithography with cell encapsulation (i.e., 3D bioprinting).³ 

GelMA has been used fairly prominently by itself and with encapsulated cells to form complex and biomimetic structures.^20–22  Although it can physically crosslink at low temperatures or high concentrations, a GelMA solution exposed to UV light will covalently crosslink in a more stable fashion, and in user-defined shapes. As GelMA is synthesized in-house, either adding different amounts of methacrylic anhydride to the reaction, or changing the pH to make substitution more or less favorable can modulate the amount of methacrylation on the polymer. This in turn impacts the GelMA properties. In polymer scaffolds with more methacrylation groups, there is a greater degree of cross-linking, and thus smaller pore sizes and less swelling. Similarly, fewer methacrylation groups results in greater pore sizes and greater swelling. In one study, the average pore sizes of the gel, after freeze drying, were 50, 30, and 25 μm for 49.8%, 64.8%, and 73.2% substitution, respectively. In addition, the compressive modulus of the hydrogels increased with substitution rate, ranging from 2 kPa, to 3.2 kPa, to 4.6 kPa for 49.8%, 64.8%, and 73.2%, respectively.^3,23  Thus, the user can modulate the mechanical properties of the gel simply by changing the amount of methacrylic anhydride within the reaction. Moreover, the user can modulate stiffness and pore size using exposure time and strength (Figure 1.2).²⁴ 

Although not as conducive to cells, synthetic polymers are a favorable option when the user wants to gain additional printing control or durability. Synthetic materials have a narrower range of molecular weights, resulting in greater batch-to-batch printing consistency. In addition, the length and shape of the polymer can be chosen specifically to modulate porosity and stiffness of the scaffold. For example, a short polymer with two functional groups on either end will form smaller pores and be stiffer than a very long chain polymer with two functional groups on either end. Synthetic polymers can also be purchased in more complex, branched shapes, such as “stars”, to increase cross-linking density.^2,25  One of the most common polymers is the commercially available hydrophilic polymer polyethylene glycol (PEG), which can be functionalized with different reactant groups to make it sensitive to free-radical polymerization. As with GelMA and meHA, a common choice is to functionalize acrylate groups onto either end of a PEG chain to form PEG-diacrylate (PEGDA).¹⁴  As PEG scaffolds contain no natural bind moieties, either adhesive peptide sequences such as arginine–glycine–aspartate–serine (RGDS) and tyrosine–iso-leucine–glycine–serine–arginine (YIGSR) or adhesive proteins such as fibronectin or laminin must be grafted on to aid in cell growth.²⁶  Other synthetic polymers for photopolymerization include poly(N-isopropylacrylamide) (PNIAAm), poly(acrylic acid) (PAA), polymethylmethacrylate (PMMA), polyacrylamide (PAam), and poly(dimethylaminoethylmethacrylate) hydrochloride (PDMAEM).² 

Using composite solutions of natural polymers and synthetic polymers, or even natural polymers and various synthetic materials, can provide improved mechanical properties, structures, and even add additional capabilities to the scaffolds.

Most often, synthetic materials are used to add additional stability and durability to natural polymers. A common application can be to add a PEG polymer to a GelMA or HA polymer to create a stronger, interpenetrating network (IPN).²⁷  Moreover, synthetic materials can also be functionalized with groups for controlled degradation. For example, thiol-ene click-chemistry reactions have been employed for polymerizing binding moieties such as the RGD motif to a PEG scaffold, allowing cells to attach and locally degrade the gel, resulting in a spreading morphology.^28,29  In addition, recent studies exploring various click chemistries have been able to incorporate photodegradable groups onto PEG, making the possibility for more complex structures even greater.^30,31  According to studies, click reactions require less photoinitiator and are less oxygen sensitive, leading to more consistent print batches. In addition, “-ene” molecules do not react with itself, so by functionalizing a natural material with norbornene and a synthetic material with a thiol group, the user can be sure they have a well-mixed scaffold.³²  Thiol-ene reactions are often referred to as step-growth polymerizations (Section 1.2.1), although the polymerization results from free-radical reactions, as the rate kinetics resemble that of step-growth (Figure 1.1).⁶  Norbornene in particular has fast polymerization due to ring strain relief, making it an excellent choice for fast printing.⁷ 

Beyond PEG-GelMA and PEG-HA polymer networks, various nanoparticles have been incorporated into GelMA systems to enhance mechanical properties, including carbon nanotubes (CNT), graphene oxide (GO), inorganic nanoparticles, and other biopolymers.

Although increasing the GelMA concentration in a pre-hydrogel solution will lead to enhanced mechanical stiffness (Section 1.3.1), the resulting increase in cross-linking impacts degradability, porosity, cell spreading, and cell growth. Similar to synthetic polymers, synthetic nanomaterials can increase the mechanical stiffness of the scaffolds, but they have the added benefit of not compromising the basic cross-linking structure. For example, Shin et al. have carried out studies exploring both the addition of CNTs and GO to GelMA pre-hydrogel solutions.^33,34  When incorporating CNTs, the nanoparticles were first coated with GelMA (for better dispersion), mixed into a GelMA pre-hydrogel solution, and then polymerized. This resulted in an increase in the elastic modulus from 15 kPa in a 5% GelMA hydrogel to ∼60 kPa in a 5% GelMA – 0.5 mg mL⁻¹ CNT hydrogel for the same curing time. Moreover, the gels exhibited a higher toughness and stronger tensile strength when the CNTs were incorporated. The solutions were also shown to be printable and supportive to encapsulated cell culture.³³  The same group followed up this study by incorporating GO into the pre-hydrogel solution. The experiment was expanded to a variety of UV curing times; it was found that the compressive modulus range for 5% GelMA was 5 to 9 kPa and the range of 5% GelMA – (0–2.0) mg mL⁻¹ GO was 4 to 24 kPa. In addition, the GelMA–GO gels exhibited enhanced electrical conductivity, and cells were found to proliferate more quickly on the GelMA–GO scaffolds, possibly due to stronger cell adhesion.³⁴  As nanotechnology continues to advance, the availability and innovation of printable materials advances as well.

Photopolymerization can only be triggered by light intensity over a certain threshold. The idea of scanning-based microstereolithography is to use a lens to focus the beam, so that the intensity is over the polymerization threshold at only a specific focal point. A desired 3D structure can be made by scanning the focal point through the volume of the whole design. Clearly, this is a serial process because 3D fabrication relies on point scanning, which limits the throughput of this method. Scanning-based microstereolithography is also referred to as direct laser writing.

Typical single-photon scanning-based stereolithography uses a microscopic objective lens to focus the UV laser beam (Figure 1.3). The photocurable prepolymer solution is loaded in a vat, a focused UV laser then induces photopolymerization at its focal point. The motion of the fabricated structure is controlled by a 3D computer numeric control platform, and the on/off state of the laser beam is manipulated by a computer-controlled shutter. Hence, photopolymerization can happen only at desired positions. During fabrication, the platform first moves to a position which is slightly under the liquid surface, then translates in a 2D horizontal plane, thus forming a thin layer of 2D solid structure made from the liquid solution. Then the platform brings the structure down a certain distance in the z-direction, allowing a thin layer of unpolymerized solution to cover the top of the polymerized structure. Another round of 2D translational scanning follows, and a new layer of 2D structure is stacked on top of the previous layer. Therefore, a 3D structure can be fabricated by this layer-by-layer scanning process.³⁵ 

An alternative way of scanning is to introduce a 2D galvo scanning mirror to the system. In this setup, the platform can only move vertically, and the 2D horizontal scanning is performed by using galvo scanning mirrors to steer the focal point of the laser. Generally, galvo mirrors has a higher scanning speed but smaller scanning range than computer numeric control platform.

The resolution of single-photon stereolithography is on the micron scale, which is ideal for making biomimetic scaffolds.³⁶  It can be used for fabrication of cell encapsulation structures, tissue engineering scaffolds, implantation parts, drug delivery devices, and so on.^35,37–40 

Two-photon absorption is a physical process where a molecule absorbs two photons simultaneously to transition to a higher energy electronic state. The energy difference between the two states is equal to the sum of the two photons. That means that in a two-photon absorption scenario, the photon wavelength is two times as long as in a one-photon absorption scenario. Two-photon polymerization refers to photopolymerization initiated by two-photon absorption, which can happen to some photoinitiators that have a large two-photon absorption cross section (molecular probability of two-photon absorption event).^41–43  For example, bis(2,4,6-trimethylbenzoyl)-phenylphosphineoxide (product name: Irgacure 819) can initiate photopolymerization with irradiation under 440 nm wavelength. It can also initiate photopolymerization with 800 nm ultrafast laser, where the sum of the energy of the two photons is the same as the energy of a 400 nm photon.

Table 1.1 lists the two-photon absorption cross section of common photoinitiators.⁴¹  Z-scan and WLC-2PA are two different measuring methods. λ_max(SPA) and λ_max(2PA) denotes the wavelengths (in nm) of absorption peaks for single-photon absorption and two-photon absorption, respectively. σ_2PA is the measured two-photon absorption cross section, the unit is in Goppert Mayer unit (GM), where 1 GM=10⁻⁵⁰ cm⁴ s photon⁻¹.

Two-photon polymerization requires extremely high power density. Thus an ultrafast pulsed laser should be used, and the laser beam should be tightly focused by a high numerical aperture objective lens. Figure 1.4 shows a typical two-photon stereolithography setup. A Ti sapphire pulsed laser (wavelength=800 nm, pulse width=100 fs) and a high numerical aperture objective (NA=1.3) are used. The working average power is 5 mW, and the writing speed is 50 microns per second.^44,45  Galvo scanning mirrors can also be used in two-photon stereolithography to enhance scanning speed.

The light intensity profile at the focal spot is a Gaussian function, and the focal spot size, i.e. the full-width-at-half-maximum (FWHM) of the Gaussian function, can be estimated by D=0.61λ/NA. For an 800 nm laser and NA 1.3 objective, the focal spot size is around 375 nm. Due to the nonlinear nature of two-photon absorption, the two-photon polymerization initiation probability is proportional to the square of the light intensity. Hence, the FWHM of the two-photon polymerization initiation probability profile is narrower than that of light intensity profile. Therefore, the line width of two-photon stereolithography is smaller than the focal spot size.⁴⁶  Sub-micro resolution down to 100 nm can be achieved by two-photon stereolithography, which is much finer than single-photon stereolithography.⁴⁷ 

An important feature of two-photon stereolithography is that it can write inside the prepolymer solution, instead of being limited to surface. The prepolymer solution, which consists of monomer and photoinitiator, has a high absorbance in UV-range but very low absorbance in the visible and near-infrared (IR) range. As a result, in single-photon stereolithography, UV light can only polymerize the surface of the prepolymer solution; in two-photon stereolithography, the near-IR light can penetrate into the prepolymer solution and induce two-photon polymerization inside the solution. Compared to single-photon stereolithography, there are no movable parts, such as a motorized platform, invading in the solution. Therefore, solid state prepolymer solutions such as soft-baked negative photoresists are also compatible with this technique.

Research on two-photon absorption for stereolithography was first published in 1997.⁴⁸  Its free-form fabrication capability and sub-microscale resolution have attracted much research attention in the following decades. This technique has been reported to be used in the research of photonic crystals, photonic metamaterials, mechanical metamaterials, micromachines, microrobots, protein microstructure, cell behavior study, tissue engineering, and so on.^49–55 

Projection-based microstereolithography is a parallel process. Instead of point scanning, this method projects an image plane in the prepolymer. By scanning this plane in the z-direction, a 3D structure can be fabricated.

Traditional 2D photolithography is the most important technology in the modern semiconductor industry. A typical projection-based exposure system for photolithography contains a UV light source, photomask, lens, and a photoresist. The UV light is patterned by the photomask and is then projected onto the photoresist by a lens.

Recent advances of digital light processing (DLP) devices have allowed maskless photolithography. These devices include liquid crystal display (LCD) and digital micromirror devices (DMDs).

A LCD DLP device is an electrophotonic device. Typical transmitting LCD devices have a multilayer structure: one polarizer layer, one electrode array layer, one liquid crystal layer, one common electrode layer, and one polarizer layer (Figure 1.5(a)). The two polarizers has perpendicular polarizing axis; thus, the device has no transparency when no voltage applied. The liquid crystal has birefringence property. When voltage is applied, the liquid crystal is reoriented, and its extraordinary refractive index is changed. Therefore, the liquid crystal can rotate the polarization axis of a linear polarized light. At certain voltage, the polarization direction of light can be rotated by 90 degrees, so the light can pass through the second polarizer. Thus, by tuning the voltage on each pixel of the electrode array, the transmittance of each pixel on the LCD can be tuned, and light can be patterned.

DMD DLP device is a micro-electro-mechanical system (MEMS), which consists of millions of micro mirrors that can flip to two different angles (Figure 1.5(b)). By flipping the mirrors, the incoming light can either be directed into the projection light path, or be deflected out of the projection light path. By individually controlling each micro mirror on the DMD chip, a desired pattern can be projected onto the photosensitive polymers.

Inspired by maskless photolithography, microstereolithography methods using LCD and DMD were invented in the 1990s and 2000s, respectively.^56,57  Compared to the scanning-based microstereolithography, these projection-based methods provide extremely fast fabrication speeds since it is a parallel process. A complex 3D structure on the millimeter scale can be fabricated in mere seconds. This high throughput feature is very attractive for mass production in industry. Furthermore, fabrication of cell-laden biomaterials become much easier by this method, because the cells only have to endure the out-of-incubator environment for a very short time.^20,58 

Projection-based microstereolithography setup can be categorized into two classes, based on the position where photopolymerization takes place.⁵⁹ 

In the first class of setup, photopolymerization happens at liquid–air interface.^60–63  The prepolymer solution is loaded in a vat (Figure 1.6(a)). The UV light is modulated by a DLP device and projected from the top of the vat. Hence, photopolymerization happens at the liquid–air interface. During fabrication, a motorized platform first moves to a position slightly under the liquid. A pattern is then loaded on the DLP device and projected onto the prepolymer solution, fabricating a 2D structure in one exposure. The platform then brings down the structure to allow a thin layer of unpolymerized liquid to cover the fabricated structure, and the DLP device loads the mask for the next layer. It is then followed by another exposure to create another layer of 2D structure. By this layer-by-layer exposure process, a 3D structure is fabricated.

In order to achieve a high quality fabrication, the liquid–air interface should maintain good flatness. Therefore, the meniscus caused by surface tension and any ripples caused by motion should be avoided after the platform moves down to get ready for the next layer. There are two ways to make a flat surface, as outlined below.

One, instead of directly moving to the desired z position, the platform first moves to a z position which is much lower than the target position, then rises back to the target position. This roundabout motion ensures the unpolymerized liquid can efficiently cover the polymerized structure. The liquid will calm down after a few seconds, making a flat surface for the next exposure.

Another method is to use a recoating blade. The platform directly moves to the target position, then the recoating blade skims through the liquid surface to help making a thin layer of unpolymerized liquid on top of the fabricated structure.

Both ways introduce a time interval between two exposures; therefore, slow down the whole fabrication process, and also introduce visible “interfaces” between layers due to the discrete motion and exposure.⁶³  These interfaces are less obvious than interfaces produced by inkjet-based and extrusion-based 3D printing, yet it could still degrade the structural integrity.

In the second class of setup, photopolymerization takes place at the liquid–substrate interface.^64,65  The prepolymer solution is loaded into a vat (Figure 1.7(a)).⁶⁶  The UV light is modulated by the DLP device and projected from the bottom of the vat. A transparent anti-adhesion substrate is installed at the bottom of the vat. During fabrication, the platform first moves to a position very close to the anti-adhesion substrate. After exposure, the polymerized structure fills the space between the platform and the anti-adhesion substrate, and adheres to the platform. The motorized platform then moves up to allow a thin layer of unpolymerized liquid to flow into the space above the substrate, and the mask of the next layer is loaded on the DLP device. A second layer of structure can then be fabricated by another exposure. Therefore, a new layer of polymer is fabricated beneath the structure, and eventually a 3D structure is printed in a continuous fashion.

Since the substrate helps to maintain good surface flatness, there is no time interval required between motion and exposure. Hence, the platform motion and UV exposure can both be performed in a continuous manner instead of a layer-by-layer manner as described above. Thus, the fabrication time is significantly reduced. Furthermore, the “interface” between layers is eliminated, resulting in a smooth and layerless surface.

A typical anti-adhesion substrate can be a polydimethylsiloxane (PDMS) membrane and hydrophobic molecular layer coated glass, such as silane. An oxygen permeable window is also reported as anti-adhesion substrate, which allows excessive concentration of oxygen to deplete free radicals and impede polymerization in a thin layer near the substrate.⁶⁴ 

Projection-based microstereolithography is able to fabricate a complex 3D structure with micron scale resolution in several seconds to several minutes. Thus, intense research interest has focused on this technique. It is finding extensive applications in microrobots, electronic devices, imaging phantoms, tissue engineering, medical implantation, and so on.^20,67–70 

Microstereolithography is a promising technology for free-form 3D fabrication in micrometer and sub-micrometer scale. A wide variety of materials, especially biomaterials, are compatible to this method.

Scanning-based microstereolithography is a time consuming 3D fabrication method due to its point scanning process. Yet this method provides microscale fabrication resolution. Two-photon scanning-based stereolithography can achieve 100 nm resolution, and even finer resolution can be achieved by super-resolution stereolithography (tens of nm).

Projection-based microstereolithography has developed into the most attractive microstereolithography method. This method offers a remarkable fast fabrication speed (within a few seconds to minutes) as well as fine resolution (lateral resolution at micrometer scale), which are both better than traditional inkjet-based and extrusion-based 3D printing. Although the utilization of UV light may be considered harmful to cells, the fast fabrication speed can minimize the exposure time. Thus cell-laden materials are also compatible to this method.

In light of the rapid evolution of microstereolithography technology, it is reasonable to believe that faster fabrication speed, finer resolution and more compatible materials can be achieved in the future.