Spin Rinse Dryer and IPA Vapor Dryer

1. Theory

• Definition: Spin rinse dryers rinse wafers in DI water and remove it by spinning under heated nitrogen, while IPA vapour and Marangoni dryers pull the water film off with a surface-tension gradient (NIST; Arizona; Marangoni patent).
• Why drying matters: Water that evaporates on the wafer leaves dissolved solids, and centrifugation can generate particles (Marangoni patent).
• Marangoni effect: IPA absorbed near the meniscus lowers local surface tension, so the film flows back into the bulk liquid (Marangoni patent).
• Pattern collapse: Capillary pressure $2\gamma \cos \theta /S$ between structures bends tall, narrow features as they dry (Watanabe et al.).
• Watermarks: On HF-last surfaces, dissolution and re-precipitation during drying leave watermarks linked to dissolved oxygen and HF vapour (Soitec patent).

2. Working Principle

• Spin rinse dry: A cassette or single wafer is rinsed at a few hundred rpm while resistivity is monitored, then spun at 1200 to 2200 rpm under nitrogen (Arizona; NIST; Soitec patent).
• Marangoni dryer: The wafer is withdrawn from water through a nitrogen stream carrying IPA vapour (Marangoni patent).
• Single-wafer IPA dry: IPA replaces water before spin-off, lowering the capillary force (Watanabe et al.).
• Hydrophobizing dry: A surface treatment raises the contact angle near 90 degrees, nearly cancelling capillary force (Watanabe et al.).

3. Applications

• End of every wet clean: After RCA and megasonic cleans (NIST; RCA Wet Bench page; Megasonic Single-Wafer Cleaner page).
• HF-last surfaces: Watermark-free drying before epitaxy or gate oxidation (Soitec patent).
• High-aspect-ratio memory structures: Collapse-free drying at aspect ratios near 19 (Watanabe et al.).

Spin Rinse Dry

Name	Formula	Related Formulae	Variables/Constants
Spin rinse dry recipe	${\omega }_{rinse}\approx 600\mathrm{rpm},{\omega }_{dry}=1200\mathrm{rpm}$	$t_{dry}\ge 240\mathrm{s}$	$\begin{array}{r}Arizona:SemitoolST-2700speedsand\\ minimumdrytimewith{N}_2,Tefloncassettes;\\ NIST:DIrinsewithresistivityprobeand\\ heated-nitrogenspot-freedry;Soitec:\\ single-waferspindryat2200rpmfor1\\ min\end{array}$ $\begin{array}{r}(myreading:batchcassettedryersspin\\ slowerandlongerthansingle-waferspin\\ dryers)\end{array}$
Rinse endpoint	${\rho }_{rinse}\to {\rho }_{DI}$	${\rho }_{DI}>18\mathrm{M\Omega }\mathrm{cm}$	$\begin{array}{r}NIST:resistivityprobemonitorschemical\\ concentrationintherinse;RCAWetBench\\ page:DIwaterabove18MOhmcm\end{array}$ $\begin{array}{r}(myreading:rinsecontinuesuntilthe\\ effluentresistivityrecoverstowardthat\\ oftheincomingDIwater)\end{array}$
Evaporative residue	$\text{evaporation leaves dissolved solids}$	$\text{centrifugation can generate particles}$	$\begin{array}{r}Marangonipatent:evaporativedrying\\ depositsdissolvedsolidsandsmall\\ particles,centrifugationknownto\\ generateparticles\end{array}$ $\begin{array}{r}(myreading:thegoalofanydryeristo\\ removewaterasliquidratherthanletting\\ itevaporateinplace)\end{array}$

IPA and Marangoni Drying

Name	Formula	Related Formulae	Variables/Constants
Marangoni drying	$\nabla \gamma \text{from IPA uptake near the meniscus}$	$\text{film flows from low to high}\gamma$	$\begin{array}{r}\gamma :surfacetension;Marangonipatent\\ citingLeenaarsetal.:solventvapourin\\ nitrogenabsorbsintothethinfilmnear\\ themeniscus,loweringlocalsurface\\ tension,sothefilmflowsbackintothe\\ bulkandthesurfaceisdrywithin\\ millimetres\end{array}$ $\begin{array}{r}(myreading:liquidispulledtoward\\ highersurfacetension,whichhereisthe\\ bulkwater)\end{array}$
Surface tension of drying liquids	${\gamma }_{H_{2}O}=73,{\gamma }_{IPA}=22,{\gamma }_F=14mN/m$	$\frac{{\gamma }_{IPA}}{{\gamma }_{H_{2}O}}\approx 0.30$	$\begin{array}{r}Watanabeetal.:valuesat20{}^{\circ }C,IPA\\ aboutonethirdofwater\end{array}$ $(computed:22/73=0.30)$
Vapour-dryer solvent use	$V_{solvent}:\text{large (vapour dryers) or a few mL (Marangoni)}$	$\text{dry zone}\approx \text{several mm behind the meniscus}$	$\begin{array}{r}Marangonipatent:earlierIPAor\\ fluorinatedvapourdryersneededlarge\\ solventamounts,thepatentedmethodafew\\ millilitres,surfacedrywithinseveral\\ millimetres\end{array}$ $\begin{array}{r}(myreading:Marangonidryingusesvapour\\ onlytosetagradient,nottoreplacethe\\ water)\end{array}$

Collapse and Watermarks

Name	Formula	Related Formulae	Variables/Constants
Capillary pattern collapse	$\Delta P=\frac{2\gamma \cos \theta }{S}$	$F=\frac{2\gamma \cos \theta }{S}HD$	$S:spacebetweenstructures$ $H:height$ $D:depthofthestructure$ $\begin{array}{r}\theta :contactangle;Watanabeetal.:these\\ relationsasthemaincauseofcollapse,\\ contactangles10degreesforIPAand92\\ degreesforwateronahydrophobized\\ surface\end{array}$ $\begin{array}{r}(computed:atS=20\mathrm{nm},7.3\mathrm{MPa}forwater\\ at0degrees,2.2\mathrm{MPa}forIPAat10\\ degrees,1.4\mathrm{MPa}forF-solventand-0.25\\ \mathrm{MPa}forwaterat92degrees)\end{array}$
Hydrophobized surface dry	$\theta \to 90^{\circ }\Rightarrow F\to 0$	$A{R}_{crit,HSD}=1.7A{R}_{crit,IPA}$	$\begin{array}{r}Watanabeetal.:forcevanishingnear90\\ degrees,1.7timeshighercriticalaspect\\ ratio,nocollapseataspectratioabout\\ 19whereIPAdrystuckthreelines\end{array}$ $\begin{array}{r}(myreading:modifyingthesurfaceattacks\\ \cos \theta ,whereasIPAonlylowers\gamma )\end{array}$
Watermarks after HF	$N_{def}:2500\to 220(\text{megasonic rinse})$	$\text{HF}+\text{HCl}(1.5\%+1\%)\Rightarrow \text{no watermarks}$	$\begin{array}{r}Soitec:watermarksfromdissolutionand\\ re-precipitation,dissolvedoxygenandHF\\ vapourascauses,megasonicrinseat0.7\\ to1\mathrm{MHz},HCladdition,H_{2}bake\\ insufficient;MegasonicSingle-Wafer\\ Cleanerpage\end{array}$ $\begin{array}{r}(computed:an11-foldreductionin0.40\mu \mathrm{m}\\ defects)\end{array}$

Important Questions

Ques: How does a spin rinse dryer clean and dry wafers?

By rinsing until clean and spinning off the water. NIST describes its Semitool PSC-101 as rinsing with DI water and drying with heated nitrogen after wet processing, with a resistivity probe monitoring the chemical concentration in the bath and a nitrogen heater for spot-free drying. The Arizona procedure for a Semitool ST-2700 rinses at about 600 rpm and dries at 1200 rpm for at least 240 s, with nitrogen and Teflon cassettes, while Soitec spin-dries single wafers at 2200 rpm for 1 min; batch dryers thus spin slower and longer than single-wafer ones (my reading). Rinsing should continue until resistivity approaches the 18 MOhm cm of incoming DI water (RCA Wet Bench page; my reading). Reference: NIST CNST NanoFab, ‘Semitool PSC-101 Spin Rinse Dryer’: deionized water rinse and heated nitrogen dry after wet processing, resistivity probe to monitor chemical concentration in the bath, nitrogen heater for spot-free drying, 75 to 200 mm wafers (fetch-tool summary); University of Arizona Micro/Nano Fabrication Center, ‘Spin Rinse Dryer (Semitool ST-2700)’ procedure: rinse at about 600 rpm, dry at 1200 rpm, minimum dry time 240 s recommended, N2 supplied, Teflon cassettes, upper unit for 100 mm and lower for 150 mm wafers, resistivity monitor on manual (fetch-tool summary); K. Radouane (Soitec), ‘Reduction of watermarks in HF treatments of semiconducting substrates’, US Patent 8,076,219 (2011): watermarks from solution-surface dissolution and re-precipitation forming lines from centre outward, attributed to dissolved oxygen in rinse water and HF vapour during drying, H2 bake not fully solving them, megasonic final rinse at 800 to 1000 W and 0.7 to 1 MHz for 10 to 60 s cutting 0.40 um defects from 2500 to 220, 1.5 per cent HF with 1 per cent HCl eliminating watermarks, single-wafer spin dry at 2200 rpm for 1 min (fetch-tool summary); RCA Wet Bench page of this vault.

Ques: How does Marangoni drying work?

By a surface-tension gradient that pulls the water back. The Marangoni patent, citing Leenaars et al., describes withdrawing a wafer from water through nitrogen carrying solvent vapour: the vapour absorbs into the thin film near the meniscus and lowers its surface tension, so the film flows back into the bulk water, which has higher surface tension, and the surface is dry within several millimetres of the meniscus. Because the water leaves as liquid, dissolved solids and small particles flow off instead of being deposited as in evaporative drying, and the patent notes that centrifugation is known to generate particles. Earlier IPA vapour dryers needed large amounts of solvent, while the patented method uses only a few millilitres. Reference: US Patent 5,660,642, ‘Moving zone Marangoni drying of wet objects using naturally evaporated solvent vapor’ (Google Patents): Leenaars et al. (Langmuir 1990) Marangoni drying with solvent vapour in nitrogen while withdrawing from water, solvent absorbing near the meniscus and lowering local surface tension so the film flows back into the bulk, dissolved solids and particles carried off instead of deposited by evaporation, centrifugation known to generate particles, prior vapour dryers needing large solvent volumes, dry within several millimetres of the meniscus (fetch-tool summary).

Ques: Why do tall, narrow structures collapse during drying, and how is it prevented?

Capillary pressure pulls neighbouring lines together. Watanabe et al. give $\Delta P=2\gamma \cos \theta /S$ and $F=(2\gamma \cos \theta /S)HD$ as the main cause of pattern collapse. At a 20 $\mathrm{nm}$ gap this is 7.3 $\mathrm{MPa}$ for water at zero contact angle, 2.2 $\mathrm{MPa}$ for IPA at 10 degrees and 1.4 $\mathrm{MPa}$ for a fluorinated solvent (computed), so IPA cuts the force to about 30 per cent. Watanabe et al. call IPA drying close to its limit and show that hydrophobizing the surface, so water stands near 92 degrees, nearly cancels the force, to -0.25 $\mathrm{MPa}$ (computed). Their method raised the critical aspect ratio 1.7 times and avoided collapse at an aspect ratio of about 19, where IPA drying stuck three lines together. Reference: T. Watanabe, T. Orii, T. Toshima, M. Nakamori, K. Egashira and Y. Ido (Tokyo Electron Kyushu), ‘Advanced wafer drying technology for 1x node and beyond using surface modification method’, ECS abstract E8-2117: Laplace pressure Delta P = 2 gamma cos theta/S and force F = (2 gamma cos theta/S) H D as the main cause of pattern collapse, surface tension 73 (water), 22 (IPA) and 14 (F-solvent) mN/m at 20 C, contact angles 10 degrees for IPA and 92 degrees for water on hydrophobized surfaces, IPA drying replacing spin drying in single-wafer tools and nearing its limit, hydrophobized surface dry with 1.7 times higher critical aspect ratio and no collapse at aspect ratio about 19 (fetch-tool summary).

Ques: What causes watermarks after an HF clean, and how are they avoided?

Dissolution and re-deposition during drying of a bare surface. The Soitec patent attributes watermarks to solution-surface dissolution and re-precipitation forming lines from the centre outward, with dissolved oxygen in the rinse water and HF vapour near the wafer as suspected causes, and notes that a hydrogen bake does not fully remove them. Two remedies are claimed: a megasonic final rinse at 0.7 to 1 $\mathrm{MHz}$ for 10 to 60 s, which cut 0.40 $\mu \mathrm{m}$ defects from 2500 to 220, an 11-fold drop (computed), and adding about 1 per cent HCl to 1.5 per cent HF, which eliminated watermarks in their tests. The megasonic rinse uses the same physics as the single-wafer cleaner (Megasonic Single-Wafer Cleaner page). Reference: K. Radouane (Soitec), ‘Reduction of watermarks in HF treatments of semiconducting substrates’, US Patent 8,076,219 (2011): watermarks from solution-surface dissolution and re-precipitation forming lines from centre outward, attributed to dissolved oxygen in rinse water and HF vapour during drying, H2 bake not fully solving them, megasonic final rinse at 800 to 1000 W and 0.7 to 1 MHz for 10 to 60 s cutting 0.40 um defects from 2500 to 220, 1.5 per cent HF with 1 per cent HCl eliminating watermarks, single-wafer spin dry at 2200 rpm for 1 min (fetch-tool summary); Megasonic Single-Wafer Cleaner page of this vault.

Ques: When is IPA preferred over plain spin drying?

When surfaces are hydrophobic or structures are fragile. Watanabe et al. report that IPA drying replaced spin drying because it is easy to install in single-wafer tools, and that IPA’s surface tension, 22 $mN/m$ against 73 $mN/m$ for water, about 0.30 of it (computed), lowers capillary forces. The Marangoni patent adds that removing water as liquid avoids the residues of evaporative drying and the particles generated by centrifugation. Plain spin drying with heated nitrogen remains common for robust wafers after cleans (NIST; Arizona). For HF-last wafers, where watermarks form during drying, Marangoni-type or IPA drying is the safer choice, while for very high aspect ratios even IPA is near its limit and surface modification is needed (my reading). Reference: T. Watanabe, T. Orii, T. Toshima, M. Nakamori, K. Egashira and Y. Ido (Tokyo Electron Kyushu), ‘Advanced wafer drying technology for 1x node and beyond using surface modification method’, ECS abstract E8-2117: Laplace pressure Delta P = 2 gamma cos theta/S and force F = (2 gamma cos theta/S) H D as the main cause of pattern collapse, surface tension 73 (water), 22 (IPA) and 14 (F-solvent) mN/m at 20 C, contact angles 10 degrees for IPA and 92 degrees for water on hydrophobized surfaces, IPA drying replacing spin drying in single-wafer tools and nearing its limit, hydrophobized surface dry with 1.7 times higher critical aspect ratio and no collapse at aspect ratio about 19 (fetch-tool summary); US Patent 5,660,642, ‘Moving zone Marangoni drying of wet objects using naturally evaporated solvent vapor’ (Google Patents): Leenaars et al. (Langmuir 1990) Marangoni drying with solvent vapour in nitrogen while withdrawing from water, solvent absorbing near the meniscus and lowering local surface tension so the film flows back into the bulk, dissolved solids and particles carried off instead of deposited by evaporation, centrifugation known to generate particles, prior vapour dryers needing large solvent volumes, dry within several millimetres of the meniscus (fetch-tool summary); NIST CNST NanoFab, ‘Semitool PSC-101 Spin Rinse Dryer’: deionized water rinse and heated nitrogen dry after wet processing, resistivity probe to monitor chemical concentration in the bath, nitrogen heater for spot-free drying, 75 to 200 mm wafers (fetch-tool summary); University of Arizona Micro/Nano Fabrication Center, ‘Spin Rinse Dryer (Semitool ST-2700)’ procedure: rinse at about 600 rpm, dry at 1200 rpm, minimum dry time 240 s recommended, N2 supplied, Teflon cassettes, upper unit for 100 mm and lower for 150 mm wafers, resistivity monitor on manual (fetch-tool summary).

References

1. University of Arizona Micro/Nano Fabrication Center, “Spin rinse dryer” procedure (Semitool ST-2700). Available: https://www.cores.arizona.edu/sites/default/files/attachments-service/spinrinsedryer100-150_0.pdf
2. NIST CNST NanoFab, “NanoFab tool: Semitool PSC-101 spin rinse dryer.” Available: https://www.nist.gov/node/413766
3. “Moving zone Marangoni drying of wet objects using naturally evaporated solvent vapor,” US Patent 5,660,642. Available: https://patents.google.com/patent/US5660642
4. K. Radouane, “Reduction of watermarks in HF treatments of semiconducting substrates,” US Patent 8,076,219, Soitec, 2011. Available: https://patents.google.com/patent/US8076219
5. T. Watanabe, T. Orii, T. Toshima, M. Nakamori, K. Egashira and Y. Ido, “Advanced wafer drying technology for 1x node and beyond using surface modification method,” ECS meeting abstract E8-2117. Available: https://ecs.confex.com/ecs/224/webprogram/Abstract/Paper21076/E8-2117.pdf
6. RCA Wet Bench page of this vault.
7. Megasonic Single-Wafer Cleaner page of this vault.

Related pages in this vault: RCA Wet Bench, Megasonic Single-Wafer Cleaner, Defect Sources and Classification, Laser-Scattering Particle Inspection, Vapor HF and Plasma Dry Clean Tool, Wet Etch Bench.