Introduction
Photolithography is a sequence of processing steps that allows the deposition and patterning of a photosensitive polymer thin film, the photoresist, on a substrate. The photoresist (PR) is selectively exposed to ultraviolet (UV) light, triggering a photoreaction that modifies its solubility in aqueous base (developer) solutions. The resulting photoresist pattern is generally used as a mask for the consecutive additive (lift-off) or substractive (etching) processing steps, before being removed (stripped).
I – Photoresist chemistry
Before going into the details of image reversal lithography, it is important to clearly distinguish between the two main families of photoresists. Depending on the photoresist formulation and the nature of the photochemical reactions, exposure to radiation can either increase or decrease the resist's solubility in the developer solution. In positive-tone photoresists, the exposed regions become more soluble after UV exposure. Consequently, these exposed areas are removed during the development step, while the unexposed regions remain on the substrate and define the final pattern. In contrast, negative-tone photoresists undergo crosslinking upon exposure, significantly reducing their solubility in the developer. As a result, the exposed regions remain after development, whereas the unexposed areas are dissolved away. The choice between positive and negative photoresists strongly affects the lithographic performance and the final pattern quality. Some of the major differences include sidewall profile, achievable resolution, adhesion to the substrate, resistance to wet or dry etching, sensitivity, and process stability. In general, positive photoresists are preferred for high-resolution patterning due to their superior dimensional control, while negative photoresists often provide better mechanical robustness and chemical resistance because of their cross-linked structure.
Among the various lithographic approaches, image reversal photoresists represent a special class of materials that combine the advantages of both positive and negative tone processes. Image reversal lithography typically starts with a positive-tone photoresist, but through an additional reversal bake and flood exposure step, the final dissolution behavior is inverted. As a result, regions initially exposed through the mask can ultimately remain after development, producing a negative-tone image from a positive-tone resist system. Historically, the first commercial photoresist used for integrated circuit (IC) fabrication was developed by Kodak and branded KTFR, for “Kodak Thin Film Resist”. It was a negative photoresist that could be tailored to absorb UV light from ~300nm to ~435nm. From 1972, due to its limited resolution power (minimum critical dimension (CD) ~ 2μm), KTFR was phased out and replaced by the second generation of (positive) photoresists, based on novolak/DNQ material. Over the years, advances in IC manufacturing followed Moore’s law, which states that the number of transistors on microchips doubles approximately every 2 years. In order to follow this miniaturization trend, Manufacturing processes are scaled down to smaller and smaller design “nodes” (the lateral dimension of the active transistor channel); New photolithography equipment is developed, with a particular focus on reducing the exposure wavelength; a new generation of photoresists, with completely different chemical composition, are introduced to the market, with absorption properties matched to the exposure wavelength. For photoresists, the generations can be summarized in this way: • i-line photoresists: absorption from ~350nm to ~450nm • DUV (KrF) photoresists: absorption @ 248nm • ArF photoresists: absorption @ 193nm • Extreme UV (EUV) photoresists: absorption @ 13 nm Regarding chemical composition, photoresists are composed of the following components:
II – Photolithography steps
1. Surface cleaning:
While the cleanroom environment (filtered air, low gases and chemicals purity) strongly reduces the amount of substrate surface contamination, the operator should still take steps to ensure that the wafer surface is as clean as possible, prior to applying the photoresist. Contaminants may be present on the surface as adsorbed metal ions, organic or inorganic elements and gases, discrete or clusters of particles, as well as naturally grown thin films (native oxides). All of these might negatively impact the final device performance. Several methods are available for cleaning wafers, but not all are effective at removing different types of contaminants. The most common are RCA (for silicon and fused silica wafers) and piranha (for glass wafers) cleaning. • RCA: RCA is a series of standard wet cleaning steps: o In step 1 (SC-1), wafers are soaked in a solution of hydrogen peroxide, ammonium hydroxide and DI water (NH4OH/H2O2/H2O 1:1:5) heated to 80°C. This step removes particles and organic contaminants (but oxidizes the surface). o In step 2 (optional step for silicon wafers only), wafers are soaked in a 10% hydrofluoric acid solution (HF/H2O 1:10). This step removes the native oxide. o In step 3 (SC-2), wafers are soaked in a solution of hydrogen peroxide, hydrochloric acid, and DI water (HCl/H2O2/H2O 1:1:6) heated to 85°C. This step removes ions and metal contaminants. • Piranha: Piranha is a mixture of sulfuric acid, hydrogen peroxide and DI water. It is an alternative to RCA and is very effective in removing organic residues, particles and metal ions from the substrate surface. It is usually the recommended cleaning process for glass wafers (float, borofloat 33, borosilicate,…). While RCA and piranha are the optimal cleaning methods before entering new batches of bare silicon and glass wafers into the fabrication chain, they may not be compatible with processed wafers, especially those that are patterned with metal stacks. Several alternatives are possible: • SCROD cleaning: SCROD wet cleaning uses a combination of a highly reactive DI water/ozone (DIO3) solution and diluted HF (DHF) to efficiently clean the surface from organics, particles, adsorbed metals, and native oxide. SCROD was initially developed by Sony as a safe, eco-friendly alternative to RCA and Piranha. SCROD is generally used in single-wafer spin-cleaning tools, where DIO3 and DHF are alternately sprayed onto the wafer. • Acetone/ IPA/ DI water solvent cleaning: Organic contamination, such as grease, photoresist residues, and carbon compounds, can be cleaned efficiently by soaking wafers in acetone, followed by isopropanol rinse baths. This can be combined with ultrasonic or megasonic agitation to also remove most particles from the surface. As a final step, the wafers can be soaked in DI water, which can help to remove ionic species. Although not as effective, this cleaning method does not use any acid solutions.
2. Surface preparation:
Most inorganic materials used as substrates for microfabrication (silicon, glass) have hydrophilic surfaces, and after cleaning, a certain amount of moisture may remain on the wafer. If moisture is not removed, the adhesion of the photoresist (PR) to the surface will be insufficient, leading to the loss of fine structures after development. To solve the issue, wafers should be treated just before applying the PR. The following standard treatments are available to remove the wafer moisture: • Dehydration: Dehydration is a simple thermal treatment: The wafers are heated at a temperature ranging from 130°C to 160°C for a few minutes, and moisture evaporates. This will be the standard preparation method for coating PR on most metallic surfaces. • HMDS: The application of hexamethyldisilazane, [(CH3)3Si]2NH, is an industry standard silanization process, which is used mainly to prepare silicon, SiO2 , and Si3N4 surfaces, but can potentially be effective with most oxidized or nitrided surfaces. HMDS is applied from the gas phase on heated substrates (typically 120°C to 150°C). The reaction is depicted below: the silicon atom of the priming agent binds to the oxidized silicon surface, leaving hydrophobic methyl groups pointing towards the outside media (air). Consequently, the wettability of the surface is affected, and turns from hydrophilic to hydrophobic, as shown in the contact angle snapshots below
• Plasma O2: Using an oxygen (O2) plasma to treat the wafer surface is an alternative dry method that can potentially combine the cleaning and surface preparation steps in a single process. Oxygen ions are extremely effective in removing both organic contamination and humidity. Plasma O2 can not only clean but also remove (strip) several microns of spin-coated PR thin films, and for this reason, it is often used between process steps. Unfortunately, plasma O2 does not remove particles and metal contaminants from the surface. In addition, the plasma O2-activated surfaces will be extremely hydrophilic, and the photoresist coating should be performed promptly to avoid rehydration from ambient humidity.
Unfortunately, plasma O2 does not remove particles and metal contaminants from the surface. In addition, the plasma-O2-activated surfaces will be extremely hydrophilic, and the photoresist coating should be applied promptly to avoid rehydration from ambient humidity.
3. Photoresist Coating:
Following surface preparation, the sequence continues in the spin coating module: a small amount of PR, typically a few milliliters, is dispensed to the center of the substrate, either while the substrate is stationary (static dispense) or while it is already spinning (dynamic dispense). The substrate is then rotated, first at an intermediate speed of 500-1000 RPM to spread the photoresist by centrifugal force, and then at speeds of 1000-6000 RPM to thin it to the target thickness.
4. Soft Bake:
After coating, the resulting resist thin film will still contain between 20–40% by weight of solvent. To further stabilize the resist film, the coated wafer will be baked for a few minutes at a temperature between 90°C and 120°C, effectively removing excess solvent by evaporation, down to about 3-8% by weight. After the softbake, the photoresist reaches its final thickness, its adhesion to the substrate improves, and it is now dry enough to be handled without risk of contaminating the equipment.
The softbake conditions will strongly affect the exposure parameters of the resist. Since it can also negatively impact the PR (by potentially promoting the decomposition of the photosensitive compound and crosslinking/oxidizing compounds in the resist), a trade-off must be made between the temperature and duration of the bake to achieve optimal, stable process conditions.
At the end of the whole sequence, the two factors that will have the most impact on the final thickness of the photoresist are the photoresist's initial viscosity and the rotation speed in the spincoating unit.
Consequently, the photoresist end user can only change the spin-coating speed, which is why photoresists are usually described by a spin curve, as shown below.
5. Expose:
Once the photoresist has been applied to the substrate and baked, it must be selectively exposed to light. The exposure triggers a chain of chemical reactions in the photoresist, to make it either soluble or insoluble in an aqueous developer solution.
The following types of equipment are used to expose the photoresist:
Mask-aligner:
Mask-aligners are the oldest type of exposure equipment. The first commercial mask-aligner was introduced in 1965 by Kulick & Soffa, and then the market was dominated by companies such as Kasper, Cobilt , and Canon for the next ~20 years. Mask-aligners are robust, fast, reliable, and cost-effective, and, for this reason, they are still in use to this date, mainly for R&D and small-scale production devices.
The mask is a UV-transparent glass or quartz square plate coated with an opaque material, usually chromium. During the mask fabrication process, the wafer layout pattern is precisely transferred into the chromium (more on how to create a mask here). In the mask aligner, the mask is usually clamped, the wafer is brought into contact or close proximity, precisely aligned using an incorporated microscope, and finally, a broad beam of collimated UV light is projected through the mask, exposing a 1:1 image of the layout pattern into the photoresist.
Most maskaligners use a high-pressure mercury (Hg) vapor lamp as a light source. The Hg lamp emits narrow spectral lines, covering both the i-line and DUV spectral range:
Direct Laser writer:
Direct laser writing is a maskless lithography (MPL) technology that was initially developed in 1995 at the Microelectronics and Computer Technology Corporation (MCC) in Austin, Texas. In a direct laser writer, a DUV or UV laser is directly focused into a very small focal spot (500nm) on the substrate without passing through a mask. The substrate moves under the objective lens, and the laser replicates the layout by exposing only the digitized area. The stage movement can be either: 1) rasterized (line by line), which is faster, more accurate but with possible stripe overlap (“stitching”) defects; 2) vectorized, which is slower but can result in stitching-free exposure.
Furthermore, in the latest generation of direct laser writing equipment, multiple laser beams are used, combined with the use of 1D or 2D micro-arrays of spatial light modulators (SLMs) to no longer expose a single spot on the substrate, but a projected image of the SLM device, greatly enhancing the throughput of the equipment.
Due to their flexibility, ease of use, and cost-effectiveness, direct laser writers are very popular in research and development (R&D) facilities, when processing is limited to small batch production.
In production facilities, however, they are still limited in terms of throughput and suffer from data-handling issues with large-polygon-count layouts, i.e., when structuring 300mm-diameter wafers with a high density of high-resolution patterns. As a result, their use is mainly limited to mask/reticle writing equipment where critical dimensions, throughput, and rapid data handling are less of an issue.
6. Development:
After exposure, the photoresist (PR) needs to be developed. The development is critical for determining the shape of the PR profile and controlling its linewidth.
Post-exposure bake (PEB)
Prior to development, some photoresists (mainly negative resists and chemically amplified resists) require a post-exposure bake (PEB) to complete the photoreaction and convert the PR into a soluble or insoluble compound.
Another positive aspect of PEB is that it improves the quality of the resist sidewalls (reduction of standing waves and better line edge roughness (LER)) due to the diffusion of the photoactive compound into the base polymer.
Development
Most developers are aqueous solutions of tetramethyl ammonium hydroxide (TMAH), potassium hydroxide (KOH), or sodium hydroxide (NaOH). Since the last two contains metallic ions (K+, Na+), they might not be compatible with micro-electronic applications.
For standard-size SEMI wafers (4 and 6 inches), several development techniques can be employed (illustrated below):
• Puddle: the wafer is static with a layer of developer covering the whole surface
• Spray: the wafer is rotating, and the developer is sprayed onto it
• Immersion: the wafer is submerged in a bath containing the developer solution for a controlled period of time.
For all techniques, the development process is typically followed by a rinsing step using deionized (DI) water while the wafer is spun at high speed to remove residual developer and dry the surface.
Users should consider the following important information:
• Substrate compatibility: Most TMAH and KOH-based developers attack alkaline-sensitive materials, such as aluminum or copper. Alternative photoresist developers are available for manual development to minimize aluminum etching.
• Developer selectivity and dark erosion: The photoresist development rate will strongly depend on the developer concentration, which is optimized to achieve a reasonably fast development rate (typically around 2μm/min) while minimising the erosion of unexposed resist region (dark erosion). Still, the selectivity of the developer is not infinite, and the PR thickness will be reduced during the development step by ~10%.
• Developer exhaustion: In the case of manual immersion development, the photoresist development rate will slowly decrease: 1) Over time due to the absorption of CO2 from the ambient air that reduces the activity of the developer; 2) For each consecutive wafer, especially with thick photoresist and large cleared surfaces, due to the saturation of the developer solution with dissolution byproducts. A refresh of the developer might be needed after a certain duration/ number of wafers.
7. Hard Bake:
To increase photoresist stability in certain applications, it is possible to further bake the developed photoresist for a longer duration than during the softbake. The hardbake has various effects, but mainly: 1) a reduction of the remaining PR solvent concentration, 2) a reduction in the DNQ PR concentration, and 3) an increase in the degree of crosslinking.
The hardbake can improve resistance to adhesion to the substrate, enhance chemical stability during certain wet etching processes (alkaline wet etching, electroplating with pH > 7), and improve resistance to certain dry etching plasma compositions.
The use of a hardbake step should be carefully evaluated, and its parameters optimized, since it also has potential drawbacks: The hardbake step, by reducing the solvent content in the resist, can make it more brittle and more difficult to strip.
8. Reflow:
In addition to the above effects, positive photoresists heated above their softening point (ranging from 110°C to 130°C, depending on the resist) will begin to reflow. The resist profile is modified and rounded (sidewalls are less vertical) to reduce its surface energy. A reflow treatment can be recommended in combination with certain etching processes, such as physical argon ion beam etching, to prevent the redeposition of material on the PR sidewalls.
Furthermore, reshaping the resist profile has been used in many microfabrication processes to obtain spheres or tilted profiles, which are then transferred to the underlying substrate to use as micro-optics components.
The dimensions of the design and reflow process parameters (i.e., duration and temperature) have a significant impact on the final resist profile, in addition to the intrinsic properties of the resist. The overall volume of the photoresist shrinks only slightly during the reflow; however, one must also take dimensional restrictions into account. To round the square sidewall shapes, there must be enough area for the photoresist to flow into. Typically, a width 3 times larger than the height as illustrated below.
9. Stripping/ Lift-off
Photolithography, which is generally used only as a temporary masking process, requires an effective method for removing the resist. The two main methods are:
Wet chemistry
The cross-linking degree of the photoresist significantly affects its solubility. Positive photoresist tends to be highly soluble. Crosslinking, an inherent part of negative ones, results in lower solubility and is generally slightly more difficult to strip.
• Solvents: Solvents are the go-to option for stripping due to their compatibility to most substrates. However, users should be aware that some transition metals (in particular Ni, Cu, Co, In, Al, Fe, Ti, Zn) can be oxidized/corroded by strong solvents under specific conditions.
o Acetone – Not ideal because of its high volatility.
o N-methylpyrrolidone (NMP) – It has been the standard for many years but is slowly phased out after having been classified as toxic (organ and reproductive toxicity).
o Dimethyl sulfoxide (DMSO) – An alternative to NMP which show comparable dissolution results and is not classified as toxic.
o N-ethyl pyrrolidone (NEP) – Used in certain stripping formulations and a little bit less hazardous than NMP according to most classifications.
• Bases: Alkaline solutions, including the standard PR developers (with lesser dilution), can be an alternative to solvent stripping in specific cases (e.g., for thick negative resist, after dry etching, etc). However, metals and even crystalline silicon can be attacked by high pH solutions.
• Acids: Strong acids such as sulfuric acid, nitric acid, aqua regia, and piranha can also be used to remove photoresists. This is used as a last resort when safer alternatives are not working or are not possible due to compatibility issues.
Plasma asher (O2)
Molecular oxygen gas is introduced into a plasma chamber, where a microwave plasma is generated under high-vacuum conditions. High-energy electrons in the plasma collide with oxygen molecules, providing in some cases enough energy to split them into oxygen radicals.
These oxygen radicals are pumped into the process chamber, where they react with organic hydrocarbon residues, such as photoresists, on the heated substrate surface. The reaction produces volatile compounds, such as carbon oxides and water, which are then removed by the vacuum system.
Post a Comment