MTBE vs PGMEA as a Photoresist Solvent: Evaporation Rate, Polarity and Why MTBE Stays a Co-Solvent

Photoresist casting solvents for 248 nm and 193 nm chemically amplified resists are selected through spin-coating defect performance, polymer solubility, and post-apply bake retention. Methyl tert-butyl ether (MTBE, CAS 1634-04-4) and propylene glycol monomethyl ether acetate (PGMEA, CAS 108-65-6) occupy distinct positions in that selection. PGMEA is the primary casting solvent for many polyhydroxystyrene and methacrylate resist platforms because its evaporation rate, polarity, and hydrogen-bonding capacity maintain resin solubility until film solidification. MTBE is introduced as a co-solvent at 5–15 wt% of total solvent mass, not as a direct replacement, because its vapor pressure and low polarity create a narrow film-formation window when used alone.

In a 300 mm coat-develop track, resist dispensing occurs through a 0.05 µm PTFE filter into a static or dynamic dispense puddle. Solvent composition affects dispense nozzle drying, puddle spread, and edge bead removal. The comparison between MTBE and PGMEA therefore must consider evaporation rate under crossflow exhaust, polarity relative to protected resin blends, and safety limits in semiconductor chemical handling.

What Limits Solvent Retention in a Spin-Coating Track?

Solvent retention after spin coating is controlled by centrifugal drainage and convective evaporation. On an open-bowl spin coater at 1,500–3,000 rpm with exhaust face velocity of 0.3–0.5 m/s, a 0.5–1.0 µm film loses most of the dispensed solvent before the post-apply bake. The relative evaporation rate by ASTM D3539 for MTBE is reported near 2.6 (n-butyl acetate = 1), while PGMEA is near 0.34. The vapor pressures at 20°C are approximately 27 kPa for MTBE and 0.39 kPa for PGMEA. The difference means MTBE-rich films begin to form a surface skin before radial drainage is complete. In resist films containing 10 wt% MTBE as a co-solvent, the resulting film retains less PGMEA at the point of solidification, but the risk of striation and Marangoni-driven thickness variation increases.

At 25°C, MTBE viscosity is 0.34 mPa·s, which is lower than the 1.24 mPa·s of PGMEA. Adding MTBE at 10 wt% to a PGMEA-based resist with a solid content of 8 wt% reduces the dispensed viscosity before solvent evaporation by approximately 15–20%. The reduction lowers the spin speed required to reach a 300 nm dry film thickness and alters wetting on hexamethyldisilazane-primed silicon. However, the lower viscosity also shortens the drainage-dominated flow period. Engineering estimates based on evaporation-rate scaling suggest that a 10 wt% MTBE formulation at 2,500 rpm reaches 95% of final dry thickness in 0.8–1.0 s, while the PGMEA baseline requires 1.5–1.8 s. Published data for this specific track configuration is limited, so the comparison is used as a process window indicator rather than as a production acceptance criterion.

Evaporation Rate and Film Formation Window

Defects from rapid evaporation include surface striations, micro-bubble craters, and lower film thickness at the wafer center when spin speed is not reset. A production coater with a sealed cup and solvent-rich atmosphere can extend the leveling time by 0.3–0.5 s, but this benefit is reduced for high-vapor-pressure MTBE because the exhaust system must be set to 0.3 m/s or higher to keep solvent concentration below 25% of the lower flammable limit. MTBE’s flash point of -28°C and boiling point of 55.2°C require a closed solvent cabinet with nitrogen blanketing and a leak-rate specification of 0.05 mm/min. In contrast, PGMEA at 45°C flash point is handled as a Class II liquid under NFPA 30 and does not require the same exhaust override in most fab facilities.

When MTBE fraction exceeds 12–15 wt% of total solvent, film thinning at the edge becomes pronounced. The low surface tension of MTBE (18.3 mN/m at 25°C) versus PGMEA (26.9 mN/m at 25°C) increases the spreading coefficient on low-energy anti-reflective coatings and changes BARC wetting behavior. In an EBR process using a 2.5 mm edge exclusion on a 300 mm wafer, a MTBE-bearing EBR solvent can penetrate 0.3–0.5 mm into the active area by capillary action before evaporation. The EBR nozzle angle and flow rate must be re-qualified when MTBE concentration changes by more than 2 wt%. Coat-develop track supplier documentation warns against EBR solvent vapor pressure above 30 kPa because backside condensation can contaminate the chuck and downstream wafer handlers.

When MTBE Loading Rises Above 15 wt% in Polyhydroxystyrene Resists

MTBE is a low-polarity solvent with a dielectric constant near 2.6 at 20°C and dipole moment near 1.36 D. PGMEA has a dielectric constant near 8.3 and dipole moment near 1.7 D. Polyhydroxystyrene resins used in 248 nm photoresists have hydroxyl groups that require hydrogen-bonding solvency. MTBE has a Hansen hydrogen-bonding term of 5.0 MPa^0.5, while PGMEA is near 9.0 MPa^0.5. At MTBE additions above 15 wt% of total solvent mass, solution clarity can be lost for resin batches with high hydroxyl content or high molecular weight. The dispersed phase then increases pressure differential across a 0.05 µm dispense filter; an inline pressure rise of more than 20 kPa over 10 L dispensed indicates precipitation or microgel formation. The threshold is resin-dependent, and published data for specific resin grades is limited.

MTBE’s Low Dielectric Constant Moves the Solubility Boundary

Hansen solubility parameters explain why MTBE remains a co-solvent rather than a primary casting solvent. For a typical methacrylate resist resin, the solubility sphere includes a dispersion term near 16–18 MPa^0.5, polar term near 5–8 MPa^0.5, and hydrogen-bonding term near 6–10 MPa^0.5. MTBE’s polar term of 2.5 MPa^0.5 places it near the lower boundary of that sphere. Blending 10 wt% MTBE into PGMEA shifts the polar and hydrogen-bonding components by approximately 5–8%, sufficient to reduce viscosity without crossing the solvent-nonsolvent boundary for most resin molecular weights. A 10,000 g/mol acrylate copolymer may tolerate 15 wt% MTBE, while a 30,000 g/mol polyhydroxystyrene may be limited to 10 wt% before precipitation is observed by turbidimetry per ISO 7027.

Unlike PGMEA, MTBE is a peroxide-forming solvent. In semiconductor-grade packaging, peroxide content is specified below 1 mg/kg by iodometric titration according to ASTM E298. Storage in open containers under UV light increases peroxide formation. Peroxide accumulation above 10 mg/kg can oxidize resist additives or generate particles. For that reason, MTBE-containing resist solvents are stored in amber glass or fluoropolymer containers under nitrogen with headspace oxygen below 1 vol%. This storage boundary is a further restriction on the use of MTBE as a primary casting solvent, because PGMEA does not carry the same peroxide-control requirement.

Edge Bead Removal, BARC Wetting, and the Co-Solvent Compromise

EBR solvents use high-evaporation solvents to remove resist from the wafer edge and backside without leaving residue. MTBE’s evaporation rate of 2.6 (ASTM D3539, n-BuAc = 1) and low surface tension of 18.3 mN/m allow rapid wetting of the edge exclusion zone. However, the same properties can pull solvent into the active area when the EBR nozzle is misaligned or the exhaust is below 0.25 m/s. Production EBR blends therefore contain 10–30 wt% MTBE in PGMEA or 2-heptanone to balance penetration and drying. Backside rinse formulations may use a similar ratio. Coat-develop track documentation advises keeping the EBR solvent vapor pressure below 30 kPa; MTBE’s vapor pressure of 27 kPa at 20°C places it near this operational limit, which is why the solvent is diluted with PGMEA.

Top anti-reflective coating (TARC) formulations use a high-volatility solvent fraction to reduce intermixing with the underlying resist. MTBE is added at 20–30 wt% in some TARC solvent packages because its surface tension of 18.3 mN/m alters coverage on hydrophobic resists. The evaporation rate of 2.6 (ASTM D3539, n-BuAc = 1) reduces residual solvent before exposure. However, TARC film thickness uniformity measured by spectroscopic ellipsometry can degrade when MTBE exceeds 30 wt% because the film solidifies before complete leveling. Published data for this specific configuration is limited, and the range is qualified on a per-track basis.

Do Solvent Blends Alter Post-Apply Bake Uniformity?

Post-apply bake uniformity is measured by thermocouple-instrumented wafers at 110°C or 130°C on a proximity bake plate. Solvent evaporation consumes heat from the film. MTBE heat of vaporization is approximately 28 kJ/mol at the boiling point, while PGMEA is approximately 42 kJ/mol. At 10 wt% MTBE, the total evaporative heat demand during the first 2–3 s of PAB is reduced, but the temperature shift is within the repeatability of contact thermocouple measurements. Film thickness uniformity after PAB for a 300 mm wafer is typically specified as 3σ = 1.5 nm for 248 nm resist; achieving this with MTBE requires solvent composition control within ±0.5 wt% of total solvent. No published standard test method exists for solvent-limited PAB uniformity; the value is equipment-specific and must be characterized on the target coat-develop track.

In a fab chemical cabinet, MTBE-containing resist solvents are stored as Class IB flammable liquids under NFPA 30, requiring exhaust flow sufficient to maintain vapor concentration below 25% of the lower flammable limit. PGMEA is stored as a Class II combustible liquid. The storage and flammability boundary does not prevent MTBE use as a co-solvent, but it makes 100% MTBE primary-solvent operation impractical on existing PGMEA-qualified coat-develop tracks.