What it is
Image-reversal resists are ordinary positive photoresists processed with an extra reversal bake and an unmasked flood exposure so the developed pattern comes out tone-inverted from a standard positive exposure, with a re-entrant, undercut sidewall profile built for metal lift-off.
How it patterns
- Substrate
- Resist
- Exposed
Positive-resist photochemistry, at a lower dose
The image-wise exposure uses the same photochemistry as a standard positive-tone process, usually at a lower dose (for AZ 5214E, about half): light through the mask converts the resist's photoactive compound where it lands, leaving a latent image in an otherwise unchanged film. Nothing about this step is negative-tone yet — the reversal happens entirely in what comes next. (The dissolution chemistry this latent image is built on is covered on the DNQ-novolak resist page, not repeated here.)
The reversal bake locks in the exposed areas
A post-exposure "image reversal bake" — driven by a crosslinking agent or amine catalyst built into the resist — converts the exposed regions into an insoluble, crosslinked state and, critically, renders them no longer photoactive. The previously exposed pattern is now chemically fixed and blind to further light, while the areas that were never exposed remain an ordinary, still-photoactive positive resist.
Flood exposure sensitizes everything else
A second, unmasked flood exposure blankets the whole wafer. Because the reversal bake already deactivated the originally exposed regions, only the areas that were untouched by the first exposure respond to the flood — they become soluble in exactly the way an ordinary positive exposure would make them. Development then dissolves those newly sensitized areas, so what remains is the resist that was NOT exposed the first time: a negative, tone-inverted copy of the original mask pattern.
Why the profile suits lift-off
The reversal process leaves a distinctive re-entrant (undercut) sidewall — the resist opening is narrower at the top than at the base. Deposited metal in a lift-off process breaks cleanly at that overhang instead of forming a continuous film down the sidewall, so the resist strips away cleanly with the unwanted metal, leaving sharp-edged features. It is this undercut, not a specific dose or bake number, that makes image reversal the standard choice for lift-off patterning.
What each step does
- 01 Spin coat
The film is a thin DNQ-novolak resist and is spun the same way, on a quick ramp to a flat, thin coat. Lift-off is the usual reason to use image reversal, so the thickness you want depends on the metal you will deposit.
Adjusting it: Pick a speed on the spin curve that makes the film comfortably thicker than the metal. Keep that thickness fixed once you have tuned the process, because the dose and bake that give a good undercut are tuned at that one thickness.
- 02 Soft bake
Drives out the casting solvent, as in any DNQ resist. The DNQ it leaves intact is what the flood exposure relies on later.
Adjusting it: Too hot or too long starts breaking down that DNQ, so the areas meant to wash away clear slowly after the flood. Under-baked film erodes and loses adhesion in the developer.
- 03 Image exposure
Writes the pattern, converting DNQ to acid in the areas that will remain after development. Resist absorbs light, so the bottom of each exposed area gets less dose than the top.
Adjusting it: This dose, more than any other setting, controls the undercut. Lower dose leaves the base of each feature less converted, so it gives a stronger overhang. Go too low and the features themselves wash away.
- 04 Reversal bake
Heat activates the additive in the resist. The exposed areas become insoluble and no longer respond to light, while the unexposed areas stay ordinary positive resist.
Adjusting it: This is the most temperature-critical step in the process. Too cool and the exposed areas only partly reverse and thin in the developer. Too hot and heat breaks down the DNQ in the unexposed areas, so they clear slowly or not at all after the flood. Hold the datasheet temperature on a calibrated hotplate.
- 05 Flood exposure
An unpatterned exposure of the whole wafer converts the DNQ in the areas that were never exposed, making them soluble. The reversed areas no longer react to light.
Adjusting it: Give a generous dose, enough to clear the full film thickness. Too little leaves resist in the openings. Extra dose does little harm because the reversed areas no longer react.
- 06 Develop
Dissolves the flood-exposed areas. The base of each reversed feature is less crosslinked than its top, so the developer cuts in underneath it, and that is the undercut.
Adjusting it: A longer develop deepens the undercut until narrow features lift. Too short leaves resist at the foot of the openings. Set the undercut with the image dose first and fine-tune it with develop time.
- 07 Deposit and lift off
Metal from a directional source such as evaporation breaks at the overhang. A solvent or remover then dissolves the resist, and the metal on top of it floats away.
Adjusting it: Don't hard bake a lift-off stencil, and keep it cool during deposition. Extra heat hardens the resist and makes it hard to remove. Sputtered metal coats the sidewall and bridges the break.
When it goes wrong
How thick should the resist be compared with the metal?
Cause: Lift-off is cleanest when the resist is well thicker than the metal. A sputtered film, which coats the sidewall anyway, needs more margin than an evaporated one.
Fix: Coat the resist clearly thicker than the planned metal, with extra margin if the metal will be sputtered.
Sources: MicroChemicals, Image reversal resists and their processing, Sufficient resist film thickness; MicroChemicals, TI 35E datasheet, General information
Why is there no undercut, with sidewalls coming out sloped?
Cause: The image dose was too high, so the film was converted almost evenly through its depth and the reversal bake fixed top and bottom alike. A reversal bake at the hot end of its window also flattens the profile.
Fix: Lower the image dose to strengthen the undercut, and run the reversal bake toward the cool end of its qualified window.
Sources: MicroChemicals, Image reversal resists and their processing, The first exposure; influence of baking parameters; MicroChemicals, TI 35E datasheet, 1st exposure
Why are there craters or bubbles in the resist after development?
Cause: Nitrogen released by the image exposure hadn't escaped before the reversal bake softened the film, so it was trapped as bubbles that open into craters in the developer.
Fix: Wait between the image exposure and the reversal bake so the nitrogen can diffuse out, longer for thicker films. For very thick films, a resist that releases no nitrogen avoids it.
Sources: MicroChemicals, Image reversal resists and their processing, Formation of bubbles in the resist film; MicroChemicals, TI 35E datasheet, Delay after exposure
Why doesn't anything clear after the flood exposure?
Cause: The reversal bake ran too hot, and the areas that were never exposed lost their ability to respond to the flood, so nothing became soluble.
Fix: Lower the reversal-bake temperature. The datasheet gives a test for it: bake unexposed samples across a range of temperatures, flood and develop them, and set the process safely below the temperature where they stop clearing.
Source: Merck, AZ 5214 E datasheet, General information
Why do reversal results vary across the wafer or from wafer to wafer?
Cause: The reversal bake's temperature window is narrow, so ordinary hotplate non-uniformity, an oven in place of a hotplate, or a thick, poorly conducting or warped substrate shifts the resist temperature enough to change the result.
Fix: Use a hotplate, not a convection oven, for the reversal bake, and re-qualify the bake for each substrate type.
Source: MicroChemicals, Image reversal resists and their processing, Influence of substrate and equipment
Why did the resist develop as a positive image instead of reversing?
Cause: The reversal bake or the flood exposure was skipped or too weak, so the resist behaved as an ordinary positive resist.
Fix: Check that both steps ran before developing. The flood is forgiving, so err toward more dose rather than less.
Sources: MicroChemicals, Image reversal resists and their processing, Process sequence; Merck, AZ 5214 E datasheet, General information
Why do narrow features shrink or wash away in the developer?
Cause: Too little image dose leaves even the top of a feature under-reversed, so it keeps dissolving. Developing long to grow the undercut can also cut right under a narrow line from both sides.
Fix: Raise the image dose back toward the datasheet value, and keep over-development short, shorter still for narrow, tall lines.
Source: MicroChemicals, Image reversal resists and their processing, Low light doses; formation of the undercut
Why is residue left in the openings even though exposure looked right?
Cause: Resist that was partly crosslinked, often by a reversal bake running slightly hot, leaves a poorly soluble residue that an ordinary developer can't clear.
Fix: Check the reversal-bake temperature first, since the residue usually starts there. A developer formulated with a scum remover helps clear it.
Source: MicroChemicals, Lithography trouble-shooter, Compatible developer?
Should I sputter or evaporate the metal for lift-off?
Cause: Sputtering coats even an undercut sidewall, so lift-off depends on the remover soaking through that coating, which limits how thick a sputtered film can be lifted. Evaporation is directional and leaves the undercut clear.
Fix: Evaporate where you can, especially for thicker films. If you must sputter, keep the film thin and expect to need ultrasonic help during lift-off.
Sources: MicroChemicals, Lift-off processes with photoresist, Sputtering or evaporation?; MicroChemicals, TI 35E datasheet, Lift-off of sputtered layers
Why does the pattern look torn or wavy after deposition?
Cause: Used in positive mode, the developed resist is still light-sensitive. UV from a sputter, CVD or plasma process exposes it, and nitrogen released under the growing film tears or blisters it, more so as the substrate warms.
Fix: Give the pattern a generous flood exposure before deposition so it is no longer photoactive. In reversal mode the resist is already inert, so this applies only to positive-mode use.
Source: MicroChemicals, Lift-off processes with photoresist, Nitrogen formation in case of positive resists
Why are there metal fences along the edges after lift-off?
Cause: Metal coated the resist sidewalls during deposition, so the remover could only get in at random points, and the sidewall coating was left standing as a fence.
Fix: Evaporate rather than sputter where you can, strengthen the undercut with a lower image dose, and keep the resist from softening during deposition, which closes the undercut.
Source: MicroChemicals, Lift-off processes with photoresist, Fences after lift-off
Why won't the metal lift off, or the resist strip, after deposition?
Cause: Heat from the source, the condensing film or ion bombardment softened the resist so the metal sealed over it, and may have crosslinked it as well.
Fix: Keep the resist below its softening point: improve thermal contact with the holder, deposit more slowly, or deposit in stages with cooling pauses. A hard-baked resist has lost its undercut and needs a stronger stripper.
Sources: MicroChemicals, Lift-off processes with photoresist, Thermal effects on the resist structures; MicroChemicals, Lithography trouble-shooter, Lift-off does not work (well)
Where it fits
- For metal lift-off, the reversal bake and flood exposure leave a re-entrant, undercut sidewall that breaks the deposited film cleanly at the resist edge, without a separate release/undercut layer.
- For a dark-field pattern (small isolated features surrounded by a large exposed field), image reversal inverts a positive-tone mask's contrast without designing or ordering a separate negative mask.
- For a process line already built around a positive resist and an aqueous TMAH developer, image reversal adds the negative-tone option on the same coating and develop chemistry, rather than switching to a different resist family.
- For a simple etch mask with no lift-off requirement, the added reversal-bake and flood-exposure steps cost process time with no benefit over a single positive exposure and develop.
- For dose and bake control, the reversal step narrows the process window: the image-wise dose and the reversal-bake temperature have to be tuned together to hit a target undercut without eroding the finest features.
- For very thick single-coat films — tens of micrometres and up — image-reversal resists are a poor fit: they coat in the same one-to-few-micrometre range as the parent positive resist they are built from.
Datasheet-cited recipes
Every value on these pages is cited to the manufacturer’s datasheet — spin curve, bake schedule, exposure dose and developer.
Further reading
- H. Klose, R. Sigush, W. Arden. Image reversal of positive photoresist: Characterization and modeling. IEEE Transactions on Electron Devices (1985). doi:10.1109/t-ed.1985.22175
- K. Chiong, H. Hiraoka. Sub-Halfmicron Negative Resist Systems By Image Reversal. SPIE Proceedings, Advances in Resist Technology and Processing IV (1987). doi:10.1117/12.940327
- Hua-yu Liu, E.D. Liu. Diffusion in AZ-5214 image reversal process and its application to e-beam proximity effect correction. Microelectronic Engineering (1987). doi:10.1016/0167-9317(87)90071-2
Common questions
Why use image reversal for lift-off?
Because the reversal bake and flood exposure leave a re-entrant, undercut sidewall instead of the vertical or slightly sloped wall of a standard positive exposure. Deposited metal breaks at that overhang rather than coating the sidewall in a continuous film, so lift-off gives cleaner, sharper edges with fewer torn or stringy features.
What does the reversal bake actually do?
It crosslinks the just-exposed areas of the resist — via a crosslinking agent or amine catalyst built into the formulation — and, in the same step, deactivates their photoactive compound so they no longer respond to light. That is what lets a second, unmasked flood exposure sensitize only the areas the first exposure missed.
Is an image-reversal resist positive-tone or negative-tone?
Both, depending on how it is processed. The same coated film can be developed right after the first exposure as an ordinary positive resist, or carried through the reversal bake and flood exposure to develop as a negative-tone, undercut image — the final tone is a processing choice on one resist, not a fixed property of the material.
Expose it at 365 and 405 nm
NANYTE BEAM is a desktop maskless lithography system with software-selectable dual-wavelength exposure and 16-bit grayscale — no photomask, no mask cost, same-day iteration.
General photolithography reference material, not a specification of any particular NANYTE BEAM configuration. Product names and trademarks belong to their respective owners; NANYTE is not affiliated with the manufacturers mentioned.