Grayscale lithography is a microfabrication technique that builds three-dimensional surface profiles in a single exposure step by varying the dose of light or electrons across a photosensitive material. Where conventional photolithography works like a stencil, producing flat features that are either fully present or fully removed, grayscale lithography acts more like sculpting: regions that receive less energy stay thicker, regions that receive more energy dissolve further during development, and the result is a continuous or stepped surface relief rather than a flat-bottomed trench. The technique has become a key tool for manufacturing microlenses, diffractive optics, microfluidic channels with sloped walls, and other structures whose function depends on smoothly varying height.
How Variable Exposure Creates a Surface Relief
In standard lithography, the entire thickness of a light-sensitive coating (the photoresist) receives enough energy to be chemically altered all the way through. After a chemical bath called development, each spot on the wafer is either still covered with resist or bare. That binary outcome is exactly what you want when etching circuit lines, but it cannot produce ramps, curves, or tiered platforms.
Grayscale lithography replaces that uniform blast of energy with a spatially varying dose. A low dose only changes the resist near the surface; a stronger dose penetrates deeper. After development, the resist that was lightly exposed remains thick, while the heavily exposed areas thin out or vanish entirely. The result is a three-dimensional topography whose height at each point corresponds to the local exposure dose.1Sensors and Actuators B: Chemical. One-step maskless grayscale lithography for the fabrication of 3-dimensional structures in SU-8 The coating and soft-bake steps before exposure are essentially the same as in traditional photolithography, so the technique plugs into existing fabrication workflows without requiring exotic materials or entirely new equipment.
Mask-Based Versus Maskless Approaches
The simplest way to deliver spatially varying exposure is through a physical mask that transmits different amounts of light at different positions. Early grayscale masks used a grid of tiny opaque and transparent pixels: by packing more transparent pixels into a given area, you let more light through, mimicking an intermediate gray tone the way a newspaper photograph uses dots of ink to suggest shading. More advanced masks use high-energy-beam-sensitive (HEBS) glass, in which the optical density can be written continuously by an electron beam, yielding smoother gradients than pixel-based designs.
Both approaches share a practical constraint. Each mask encodes one specific three-dimensional design, so changing the target shape means ordering a new mask. For research groups iterating on prototypes, that turnaround is a bottleneck. Testing with an i-line resist, one study divided a 2.5-micrometer-thick film into roughly 30 distinguishable gray levels and found that a dose variation of just three percent caused a 48-nanometer shift in the final resist thickness, with a practical minimum step size of about 80 nanometers before any dose-uniformity optimization was applied.2ResearchGate / SPIE Proceedings. Grayscale lithography: 3D structuring and thickness control Those numbers illustrate both the fine control that is possible and the extreme sensitivity of the process to dose precision.
Maskless methods eliminate the physical mask by steering a focused beam of light or electrons point by point across the resist. A direct-write laser system, for example, modulates laser intensity on the fly, painting each pixel of the design at a different power level. This makes changing designs trivial: you simply load a different digital file. One group used direct-write laser grayscale lithography to pattern both tiered and smoothly sloped structures in a photoresist layer just 2.55 micrometers thick, then transferred those shapes into multi-layer piezoelectric ceramic stacks with a single etching step.3PubMed. Direct-Write Laser Greyscale Lithography for Multi-Layer Lead Zirconate Titanate Thin Films The flexibility of maskless direct-write systems has made them the dominant tool for grayscale prototyping in academic and small-batch settings.
Electron Beam Grayscale Lithography
Electron beams can be focused to spots far smaller than optical wavelengths allow, which makes electron beam lithography the method of choice when the target features are on the nanometer scale. For grayscale work, the beam’s dwell time or current at each pixel is varied to create a dose map, and the resist develops into a height profile just as it does with optical grayscale exposure. The technique is used to write diffractive optical elements like Fresnel lenses, where the surface profile must follow a precise mathematical curve with sub-micrometer accuracy.4Journal of Vacuum Science & Technology B, Nanotechnology and Microelectronics: Materials, Processing, Measurement, and Phenomena. Efficient proximity effect correction method based on multivariate adaptive regression splines for grayscale e-beam lithography
A persistent headache in electron beam lithography is the proximity effect. Electrons scatter as they enter the resist and the substrate beneath it, depositing energy in an area wider than the intended spot. For binary patterning this causes line widths to swell, but for grayscale work the problem is worse because every pixel’s height is affected by scattered energy from its neighbors. Without correction, features can blur together or develop to the wrong depth. Multi-pass grayscale strategies combined with proximity effect correction algorithms have achieved uniform nanowire widths down to about 15 nanometers across entire arrays, a level of uniformity that was impossible without those corrections.5Microelectronic Engineering. Implementation of electron beam grey scale lithography and proximity effect correction for silicon nanowire device fabrication The trade-off is speed: electron beam writing is serial and slow compared to optical exposure, so it is generally reserved for master molds, prototype devices, and structures where no other technique can match the resolution.
Transferring Resist Profiles Into Substrates
A photoresist relief is fragile and temporary. Most practical devices need the three-dimensional shape carved into a durable material like silicon or glass. The standard transfer method is reactive ion etching, a plasma process that removes material from the wafer surface. If the etch removes the resist and the substrate at comparable rates, the height profile in the resist gets faithfully copied into the substrate below, scaled by the ratio of those rates.
Getting a faithful copy requires careful tuning. Any mismatch in etch selectivity will compress or stretch the vertical profile, and non-uniform plasma conditions across the wafer can introduce tilt or distortion. Still, the approach works well enough to produce silicon microlens mold cavities whose shapes deviate from the intended design by roughly ten percent, with curvature radius variation within about five percent.6Micro and Nano Engineering. Assessing tolerances in direct write laser grayscale lithography and reactive ion etching pattern transfer for fabrication of 2.5D Si master molds For high-volume production, those silicon master molds can then be replicated by nanoimprint lithography, stamping copies of the three-dimensional pattern into polymer films on additional wafers. Profilometry comparisons between masters and imprinted stamps have shown the dimensional match is excellent.7Journal of Manufacturing Processes. Nanoimprint lithography for grayscale pattern replication of MEMS mirrors in a 200 mm wafer
The same etch-transfer principle has been used to create passive microfluidic valves directly in silicon and glass. By sculpting sloped resist features and then etching proportionally, researchers produced three-dimensional valve geometries embedded in the substrate itself, removing the need for separately bonded or assembled valve components.8Microfluidics and Nanofluidics. 3D passive microfluidic valves in silicon and glass using grayscale lithography and reactive ion etching transfer
Surface Roughness and Standing Waves
One of the subtler challenges in grayscale lithography is unwanted surface roughness. When light enters a thin photoresist film, part of it reflects off the substrate and interferes with the incoming beam, creating standing waves: alternating bright and dark fringes stacked vertically through the resist. In binary lithography, those fringes mostly wash out during development because the resist is removed all the way down. In grayscale work, you stop development partway through the film, so the surface often lands right on a standing-wave fringe, leaving a rippled texture that can ruin optical performance.
Measurements on i-line resist show standing-wave-induced roughness peaks reaching about 6 to 7 nanometers at certain exposure doses, against a baseline roughness of around 3 nanometers in regions between the peaks. A post-exposure bake, which lets thermally activated diffusion smooth out the chemical gradients left by standing waves, reduces the roughness peaks by roughly 19 percent and baseline roughness by about 15 percent. Adding a bottom anti-reflective coating (BARC) beneath the resist eliminates the standing-wave peaks entirely, since the coating absorbs the reflected light that would otherwise cause interference.9Micro and Nano Engineering. Standing waves-induced surface roughness in grayscale lithography The practical takeaway is that anyone fabricating optical elements by grayscale lithography on a reflective substrate should budget time for BARC deposition and careful bake optimization, or accept roughness that can compromise device performance.
Design Workflow and Software
Designing a grayscale exposure requires translating a three-dimensional shape into a two-dimensional dose map. Each pixel in the map specifies how much energy the resist should receive at that location, which in turn determines the final height. The mapping between dose and height is not linear; it follows a contrast curve specific to the resist and processing conditions. Getting the curve right requires a calibration run: expose a series of test patches at known doses, measure the resulting thicknesses, and fit the relationship.
Traditionally, this calibration and map generation relied on specialized lithography software, which limited access for groups without expensive tool suites. A recent approach demonstrated that the open-source 3D modeling program Blender can handle the entire design chain. Users build the target three-dimensional structure in Blender, apply a height-based color gradient mapped through the experimentally measured calibration curve, and render the result as a grayscale image from directly above. That rendered image becomes the digital mask file fed to the lithography tool.10Micro and Nano Engineering. Rapid prototyping of 3D microstructures: A simplified grayscale lithography encoding method using blender Because Blender is free and widely known in the maker and design communities, this lowers the barrier considerably for research groups and teaching labs that do not have access to commercial mask-design packages.
Applications in Optics and Photonics
The earliest and still most prominent use of grayscale lithography is fabricating micro-optical components whose function depends on precisely shaped curved or graded surfaces. Microlens arrays, Fresnel lenses, blazed diffraction gratings, and beam-shaping elements all require smooth, continuous height variations that binary lithography simply cannot produce in a single step. Making a blazed grating by binary methods, for instance, would require dozens of sequential aligned exposures to approximate a sawtooth profile; grayscale lithography does it in one pass.
Recent work has pushed the technique into diffractive optical elements for research microscopy. A consumer-grade digital light processing (DLP) projector was repurposed to project a grayscale pattern onto a droplet of UV-curable resin, producing finished diffractive elements including vortex plates and Zernike polynomial phase masks in about ten minutes. The resulting optics showed strong agreement with simulated performance.11ACS Photonics. Fabrication of Customized Diffractive Optics in under 10 Minutes via Single-Shot Grayscale Projection on a Consumer-Grade DLP System That speed and simplicity is a dramatic departure from the multi-day fabrication cycles traditionally associated with custom diffractive optics.
At the other end of the complexity spectrum, electron beam grayscale lithography has been combined with atomic layer deposition to create dielectric metasurfaces, arrays of sub-wavelength structures whose optical properties can be engineered pixel by pixel. By varying the height of individual nanostructures across the metasurface, researchers gained independent control over both the phase and the dispersion of visible light passing through the device.12Nano Letters. Metasurfaces with Freely Varying Height in the Visible Using Grayscale Lithography That additional degree of freedom is difficult or impossible to achieve with fixed-height metasurface designs, and it opens the door to broadband flat optics with performance closer to traditional bulk lenses.
Microfluidic and Biomedical Uses
Microfluidic devices, the miniature plumbing networks used in lab-on-a-chip diagnostics and chemical analysis, benefit from three-dimensional channel geometries that guide fluid flow in ways flat channels cannot. Sloped walls can passively steer droplets, funnel cells toward a sensor, or create valves that open and close without any moving parts. Grayscale lithography’s ability to define such shapes in resist and then etch them into silicon or glass has made it a practical fabrication route for passive three-dimensional microfluidic valves.13Microfluidics and Nanofluidics. 3D passive microfluidic valves in silicon and glass using grayscale lithography and reactive ion etching transfer
Beyond channel geometry, the concept of spatially graded exposure has been extended to biological materials. By combining projection lithography with engineered proteins containing photosensitive amino acids, researchers have used gradient light patterns to create surfaces where both the stiffness and the chemical identity of an artificial extracellular matrix vary across the substrate. Cells cultured on these gradient surfaces experience a continuously changing mechanical and chemical environment, making the platform useful for studying how cells respond to stiffness gradients and chemical cues simultaneously.14Biomedical Microdevices. Gradient lithography of engineered proteins to fabricate 2D and 3D cell culture microenvironments This is a less obvious extension of the grayscale principle: instead of sculpting a polymer resist, the variable exposure dose controls the degree of crosslinking in a biological hydrogel, tuning its mechanical and biochemical properties in space.
Where the Process Remains Difficult
For all its versatility, grayscale lithography demands tighter process control than binary lithography across nearly every step. A few areas give fabricators consistent trouble.
Dose sensitivity is the most fundamental. Because the final height is a continuous function of exposure energy, any drift in lamp intensity, laser power, or electron beam current translates directly into a height error. The three-percent dose variation producing a 48-nanometer thickness change mentioned earlier is not an extreme case; it is a routine sensitivity that must be managed through careful power monitoring and stabilization.15ResearchGate / SPIE Proceedings. Grayscale lithography: 3D structuring and thickness control
Resist contrast is another constraint. The number of distinguishable height levels you can achieve depends on the resist’s contrast curve: a high-contrast resist snaps sharply between fully exposed and unexposed states, which is ideal for binary patterning but gives very few usable gray levels. A low-contrast resist offers more gray levels but at the cost of reduced edge sharpness. Choosing or formulating a resist with the right contrast for the application is a non-trivial part of process development, and the optimal resist for one target shape may not be optimal for another.
Standing-wave roughness, discussed earlier, adds a further layer of complexity for optical applications where surface smoothness matters. The BARC and post-exposure bake mitigations help, but they also add process steps and narrow the operating window. Each mitigation introduces its own sensitivity: BARC thickness must be tuned to the exposure wavelength, and bake temperature and duration must be balanced against resist reflow that could distort the intended profile.
Accessibility and the Shrinking Cost Barrier
Grayscale lithography was once confined to facilities with mask-aligner suites, custom HEBS glass blanks, or million-dollar electron beam writers. Several developments are changing that picture. Maskless direct-write laser systems have become commercially available at price points accessible to well-funded university labs. The DLP-projector-based approach for diffractive optics requires only a consumer projector, a UV-curable resin, and a calibration procedure, reducing the equipment cost by orders of magnitude and the fabrication time to minutes.16ACS Photonics. Fabrication of Customized Diffractive Optics in under 10 Minutes via Single-Shot Grayscale Projection on a Consumer-Grade DLP System
On the software side, replacing proprietary design tools with free alternatives like Blender removes a cost and training barrier that historically kept smaller groups from experimenting with grayscale processes.17Micro and Nano Engineering. Rapid prototyping of 3D microstructures: A simplified grayscale lithography encoding method using blender Combined with the growing availability of nanoimprint replication, which lets a single grayscale master be stamped into thousands of copies, the economics of the technique are shifting. What was once a custom, one-off fabrication method is increasingly viable for moderate-volume production of components like MEMS mirrors and microlens arrays.18Journal of Manufacturing Processes. Nanoimprint lithography for grayscale pattern replication of MEMS mirrors in a 200 mm wafer
The resolution frontier continues to move as well. Electron beam grayscale writing now enables height-variable nanostructures for visible-wavelength metasurfaces, a regime where the lateral feature size and the height both need to be controlled at the tens-of-nanometers level.19Nano Letters. Metasurfaces with Freely Varying Height in the Visible Using Grayscale Lithography Proximity effect correction algorithms continue to improve, which gradually relaxes the trade-off between writing speed and pattern fidelity. Whether the application is a quick prototype diffractive lens printed in ten minutes on a desktop projector or a nanometer-precise metasurface written overnight on an electron beam tool, the underlying idea is the same: instead of treating lithography as an on-off switch, treat it as a dimmer, and use the full range of the dial.