TOK TSMR-5000 is a positive-tone i-line photoresist based on a novolak resin and diazonaphthoquinone sensitizer system. The material is supplied as a solvent-borne liquid with a nominal kinematic viscosity of 5,000 mPa·s at 25°C, determined by cone-and-plate viscometry in accordance with ASTM D4287. The formulation is designed for single-spin coating of thick films in the 8–20 µm range after softbake, placing it between thin sub-micron i-line resists and very high-viscosity chemically amplified thick resists. Primary applications include electroplating molds for copper redistribution layers, gold bumping, and nickel under-bump metallurgy, as well as sacrificial layers in MEMS bulk micromachining. Unlike chemically amplified systems, TSMR-5000 does not rely on photoacid generation or acid diffusion. The DNQ-novolak dissolution inhibition mechanism provides stable post-exposure delay latitude and low sensitivity to airborne amines. The product is filtered to reduce trace metals; typical manufacturer specifications for sodium, potassium, iron, and chromium are below 50 ppb per element, although published data for this specific configuration is limited.

What Spin-Speed Regime Governs Single-Coat Film Thickness for TSMR-5000?

Representative spin-curve data for TSMR-5000 show a non-linear thickness response across the 800–3,000 rpm range. At 800 rpm, softbake film thickness reaches approximately 18–20 µm; at 1,500 rpm, thickness falls to 12–14 µm; at 3,000 rpm, thickness is approximately 5–6 µm. These values assume a 200 mm silicon wafer, a 23°C resist dispense temperature, and a closed-bowl spin coater with exhaust airflow of 0.3–0.5 m/s and relative humidity between 45% and 55%. Film thickness is measured by stylus profilometry after softbake, using a stylus radius of 12.5 µm and a scan length of 1 mm over a step created by solvent edge bead removal. Thickness non-uniformity across the wafer is typically within ±5% when acceleration from 300 rpm to final spin speed is set to 1,000 rpm/s and final spin time is 30–60 s. The resist is pseudoplastic; low-shear viscosity is 5,000 mPa·s, but high-shear viscosity during spin coating is lower. Dispense lines with an internal diameter smaller than 3.2 mm can create pressure drops above 70 kPa and should be avoided. Solvent loss during spin coating is rapid in the first 5 s; the drying front moves inward from the wafer edge and can form a surface skin if exhaust flow is too high. A maximum exhaust of 0.5 m/s at the coater cup is recommended. If the edge bead is not removed with PGMEA within 5–10 s after spin stop, bead height can be 1.5–2.5× the nominal film thickness and may crack during softbake. The edge-bead removal nozzle should be positioned 2–4 mm from the wafer edge and dispense solvent at 0.5–1.0 mL/s while the wafer rotates at 500 rpm.

Typical process parameters for TSMR-5000 on 200 mm wafers with copper seed layer

Process step	Parameter	Set point or range	Measurement method
Substrate dehydration	Hotplate	150°C for 60 s	Contact angle or visual inspection
Adhesion promotion	HMDS vapor prime	110°C for 60 s	Contact angle 65–75°
Coating	Static dispense	3–5 mL; spin 800–3,000 rpm	Stylus profilometry after softbake
Softbake	Hotplate	110°C for 120–180 s	Residual solvent ≤3 wt% by ASTM D2369
Exposure	i-line stepper or aligner	150–250 mJ/cm² at 365 nm	NIST-traceable radiometer
Post-exposure bake	Hotplate	110°C for 60 s	Thermal profile calibration
Development	Spray or puddle	2.38 wt% TMAH, 60–120 s at 23°C	Endpoint by laser interferometry
Hardbake	Convection oven	120–150°C for 30 min	Sidewall profile by SEM cross-section

Batch-to-batch viscosity variation is usually less than ±5% of the certificate-of-analysis value for TSMR-5000. A +5% viscosity shift at 1,500 rpm can increase film thickness by 0.4–0.6 µm, which may shift the bottom critical dimension of a plating mold by 0.5–1.0 µm if exposure dose is not adjusted. Incoming material should therefore be tested by ASTM D4287 and compared with the release certificate. If a new lot falls outside the ±5% band, spin speed is adjusted by approximately ±100 rpm per 5% viscosity shift as a first-order correction. The correction must be verified on a monitor wafer by profilometry before product lots are processed. Dissolution kinetics in 2.38 wt% TMAH follow the classic DNQ-novolak mechanism. Unexposed TSMR-5000 dissolves slowly, typically 0.1–0.3 µm/min in the first 60 s of development, because the diazonaphthoquinone sulfonate ester acts as a dissolution inhibitor. Upon exposure at 365 nm, the DNQ molecule undergoes Wolff rearrangement to a ketene intermediate and then hydrolyzes to indene carboxylic acid, which accelerates alkaline solubility. The photochemical contrast for TSMR-5000 class materials is approximately 2.5–3.5, and the clearing dose is 40–80 mJ/cm², but these values are proprietary and must be verified on the user’s aligner. Developer temperature has an Arrhenius-like effect: a 1°C increase raises development rate by 5–10%, so developer temperature must be held at 23±0.5°C. Softbake temperature above 120°C further crosslinks the novolak and reduces photospeed; a 5°C softbake increase can increase clearing dose by 15–25%. Residual solvent after softbake should be below 3 wt% by loss-on-drying in accordance with ASTM D2369; higher residual solvent causes pop defects during hardbake and lowers plating bath compatibility.

When TSMR-5000 Replaces Thin-Film Resists in Electroplating Molds

Substitution of TSMR-5000 for a conventional low-viscosity i-line resist changes the process architecture. A low-viscosity resist of 10–50 mPa·s must be coated in multiple cycles to reach 10 µm, and each additional layer introduces interfacial contamination, solvent re-dissolution, and cumulative edge-bead variation. TSMR-5000 reaches the same thickness in one spin pass, eliminating interlayer adhesion failures but increasing edge bead height and altering the spin curve. The higher viscosity requires a larger static dispense volume, typically 3–5 mL for a 200 mm wafer, and a slower initial spread step at 300–500 rpm for 5 s before ramp to final spin speed. A dispense rate of 1.0–1.5 mL/s is common; lower dispense rates can produce puddle asymmetry when the resist stream is not centered within 1 mm. Against chemically amplified thick resists, TSMR-5000 offers lower airborne amine sensitivity and wider post-exposure delay latitude. Chemically amplified resists require immediate transfer to post-exposure bake and can suffer T-topping if exposed to amine concentrations above 1 ppb. TSMR-5000 can tolerate post-exposure delay up to 24 h under controlled 23°C and 45–55% relative humidity without measurable linewidth shift, although this is application-specific. However, resolution is lower. At 15 µm film thickness, the minimum resolved line/space is approximately 5 µm/5 µm, whereas a chemically amplified thick resist may resolve 2–3 µm at equivalent thickness. The sidewall angle of TSMR-5000 after development is typically 75–85°, depending on exposure dose and developer agitation; chemically amplified resists can produce near-vertical sidewalls of 88–90°. Thermal stability after hardbake is another differentiator. TSMR-5000 does not reflow below 150°C; hardbake at 150°C for 30 min retains sidewall angle within 5° of the post-develop profile, as verified by cross-sectional SEM. This is sufficient for most electroplating baths. If the process requires hardbake above 180°C, the resist begins to decompose and is not recommended. Published data for this specific configuration is limited above 180°C.

Comparative properties of TSMR-5000, low-viscosity i-line resist, and chemically amplified thick resist

Parameter	TSMR-5000	Low-viscosity i-line resist	Chemically amplified thick resist	Method
Kinematic viscosity at 25°C	5,000 mPa·s	10–50 mPa·s	200–2,000 mPa·s	ASTM D4287
Single-coat film thickness	8–20 µm	0.5–3 µm	10–100 µm	Stylus profilometry
Resolution at nominal thickness	5 µm at 15 µm	0.5 µm at 1 µm	2–3 µm at 15 µm	SEM cross-section
Post-exposure delay CD shift at 24 h	≤5%	≤5%	≥15%	SEM linewidth metrology
Stripping after plating	Solvent or NMP plus O2 plasma	Solvent or NMP plus O2 plasma	Strong acid or piranha plus proprietary solvent	XPS residue analysis

Production-scale electroplating tests with TSMR-5000 molds on 200 mm wafers have been reported for copper sulfate baths at 25°C and current densities up to 4 A/dm². The resist sidewalls remain intact through 60 min of plating when the hardbake temperature is 120°C or higher. Without hardbake, adhesion loss appears at the resist-seed interface once plated metal thickness exceeds 10 µm, especially at the wafer edge where residual stress concentrates. Gold bump plating in cyanide-bearing baths at 55–65°C and pH 4.8–5.5 is similarly dependent on hardbake. A pre-coat oxygen plasma ash at 100 W for 30 s on copper or titanium seed layers improves wetting and reduces post-develop scum. If the seed layer has been stored for more than 4 h in ambient air, a dilute sulfuric acid dip is recommended before dehydration. The resist is not compatible with strongly alkaline plating baths above pH 9 for extended periods, because the novolak matrix can swell and saponify at the developer interface. It is also not recommended for amine-containing electrolytic nickel baths unless a protective hardbake is used and bath temperature is held below 50°C. Adhesion loss at the resist-seed interface is the dominant failure mode in thick electroplating molds. It typically appears as bright metal underplating along the resist foot, visible after seed etch at 50× optical magnification. Root causes include residual water on the seed layer, incomplete HMDS vapor prime, softbake temperature below 100°C, or developer attack at the interface due to insufficient rinse. On copper seed layers, adhesion energy of a comparable DNQ-novolak film after HMDS priming is on the order of 0.1–0.5 J/m² by four-point bend fracture; without HMDS, it falls below 0.05 J/m² and underplating increases. The adhesion promoter should be applied immediately after dehydration because copper oxide regrowth begins within 10 min at 45% relative humidity.

Stripping and Post-Plate Surface Conditioning

Resist stripping after plating is performed with solvent blends containing N-methyl-2-pyrrolidone or dimethyl sulfoxide at 23–60°C. Because TSMR-5000 is a non-crosslinking DNQ-novolak system, the bulk film dissolves in 10–20 min under immersion with ultrasonic agitation at 40 kHz. Residue is then removed by oxygen plasma ashing at 150°C for 5–15 min, with endpoint determined by optical emission spectroscopy of the CO plasma line at 483 nm. If a chemically amplified epoxy-based resist were used in the same loop, solvent strip alone would leave cross-linked residue, and an aggressive piranha clean would be needed. TSMR-5000 avoids that failure mode. Solvent stripping will not remove plated metal residues or intermetallic compounds formed during gold or copper deposition. A subsequent dilute acid clean, such as 10 wt% citric acid at 40°C for 60 s, is used for copper traces where wire-bonding surfaces must meet cleanliness requirements of IPC-A-610. After strip, the underlying wafer surface is inspected by X-ray photoelectron spectroscopy. Carbon and nitrogen signal intensities must remain below 0.1 atomic% to prevent wire-bond lift-off. If the carbon signal exceeds this threshold, the plasma ash is extended at 50 W in 10 min increments until the O 1s peak area changes by less than 5% between successive measurements. This practice is recommended only for wafers without thermally sensitive low-k dielectrics. TSMR-5000 must be stored at 5–10°C and allowed to equilibrate to 23°C for 2–4 h before opening to avoid water condensation. Shelf life from date of manufacture is 12 months in the original amber glass bottle under nitrogen. If viscosity measured by ASTM D4287 deviates more than ±10% from the certificate of analysis, the spin-speed film thickness curve must be re-qualified. The solvent vapor concentration in the coating area should remain below the ACGIH TLV of 20 ppm for PGMEA. Developer waste containing TMAH and dissolved resist solids is corrosive and must be segregated from acid waste streams because TMAH reacts exothermically with mineral acids. These operational limits are part of the material control plan for high-volume bumping and MEMS production lines.