High-NA DUV Immersion Scanner

1. Theory

• Definition: A DUV immersion scanner images a reduced mask pattern onto resist with 193 $\mathrm{nm}$ ArF light through a projection lens whose last element is coupled to the wafer by water, which raises the numerical aperture above 1 (Smith et al.; ASML page).
• Resolution: Half pitch scales as $k_1\lambda /NA$, with $k_1$ limited to 0.25 because at least two diffraction orders must pass the pupil (ECE 695).
• Immersion: Water has an index of 1.44 at 193 $\mathrm{nm}$, shortening the wavelength in the gap to about 134 $\mathrm{nm}$; ASML’s tools reach NA 1.35 (Smith et al.; ASML page).
• Partial coherence: The illuminator fills a fraction $\sigma =N{A}_c/N{A}_p$ of the pupil, and the image is described by Hopkins transmission cross coefficients (Kirchauer).
• Polarization: At high angles the TM fringe contrast falls and vanishes at 45 degrees between beam and normal, so TE polarization matters for low $k_1$ (Estroff et al.; Smith et al.).
• Steppers and scanners: Steppers expose a whole reduced field at once at 5:1, while step-and-scan scanners sweep a slit through the lens sweet spot at 4:1; KrF (248 $\mathrm{nm}$) and dry ArF (193 $\mathrm{nm}$) tools precede immersion (NPTEL L14; Laser Focus World; ASML DUV page).

2. Working Principle

• Illuminator: Koehler illumination makes the field uniform to about 1 per cent, and the pupil shape is set to conventional, annular, dipole or quadrupole (Sasian; Kirchauer); ASML uses a programmable FlexRay illuminator (ASML page).
• Projection lens: A doubly telecentric reduction lens images the mask (Sasian).
• Immersion hood: Water fills the lens-to-wafer gap, with index held to tens of ppm and temperature to about 0.1 ${}^{\circ }\mathrm{C}$ (Smith et al.).
• Stages and sensors: Dual stages, alignment and levelling sensors keep overlay to 2.5 $\mathrm{nm}$ on product (ASML page).

3. Applications

• Critical layers: Single-exposure patterning down to 38 $\mathrm{nm}$ half pitch with dipole illumination (ASML page).
• Non-critical layers: Dry KrF and ArF scanners at 300 to 400 wafers per hour (ASML DUV page).
• Low-k1 patterning: Polarized, off-axis illumination for $k_1$ below 0.3 (Smith et al.; Kirchauer).
• Metrology link: Printed CD and overlay are verified on the CD-SEM and overlay tools (CD-SEM page; Overlay Metrology Tool page).

Resolution and Immersion

Name	Formula	Related Formulae	Variables/Constants
Rayleigh resolution	$HP=k_1\frac{\lambda }{NA},NA=n\sin \theta$	$k_1\ge 0.25\Rightarrow p_{min}=\frac{\lambda }{2NA}$	$HP:halfpitch$ $k_1:processfactor$ $\lambda :vacuumwavelength$ $NA:numericalaperture$ $\begin{array}{r}n:indexofthemediumfillingthelast\\ gap\end{array}$ $\begin{array}{r}\theta :largestrayangle;ECE695:R=k_1\lambda /NA,\\ k_{1}limit0.25,Abbepitch\lambda /(2NA),52\mathrm{nm}at\\ 193\mathrm{nm}andNA0.93;NA=n\sin \theta (myreading\\ oftheimmersionNA,withn=1.44from\\ Smithetal.)\end{array}$ $\begin{array}{r}(computed:at193\mathrm{nm}andNA1.35,k_1=0.25\\ gives35.7\mathrm{nm}halfpitchanda71.5\mathrm{nm}\\ minimumpitch;ASM{L}^{\prime }s38\mathrm{nm}dipoleand40\\ \mathrm{nm}C-quadresolutionsare{k}_1=0.266and\\ 0.280)\end{array}$
Immersion numerical aperture	$N{A}_{max}<n_{water}=1.44$	${\lambda }_i=\frac{\lambda }{n_{water}}$	$n_{water}:waterindexat193\mathrm{nm}$ $\begin{array}{r}{\lambda }_i:wavelengthinwater;Smithetal.:n=\\ 1.44,{\lambda }_{i}about134\mathrm{nm},absorptionbelow\\ 0.05\mathrm{c}{\mathrm{m}}^{-1};ECE695:drylimitabout0.93at\\ 68degrees;ASML:NA1.35\end{array}$ $\begin{array}{r}(computed:NA1.35inwaterisa69.6\\ degreerayangle,closetothe68.4\\ degreesofadryNA0.93lens)\end{array}$
Depth of focus	$DOF=k_2\frac{\lambda }{N{A}^2}(\text{paraxial})$	$DOF=\pm k_2\frac{{\lambda }_i}{{\sin }^2\theta }(\text{in the medium})$	$k_2:DOFfactor$ $\begin{array}{r}\theta :angleintheimagingmedium;ECE695:\\ theparaxialformforNAupto0.5;Smith\\ etal.:themediumform,DOFscaling\\ linearlywithindex(seetheQ/A)\end{array}$ $\begin{array}{r}(computed:fora110\mathrm{nm}pitch,\lambda /(2p)=\\ 0.877gives\pm 251k_2\mathrm{nm}dryand\pm 361k_2\mathrm{nm}in\\ water,afactorof1.44)\end{array}$

Illumination and Image Formation

Name	Formula	Related Formulae	Variables/Constants
Partial coherence factor	$\sigma =\frac{N{A}_c}{N{A}_p}$	$p_{min}=\frac{\lambda }{NA\left(1+\sigma \right)}$	$\sigma :partialcoherencefactor$ $\begin{array}{r}N{A}_c,N{A}_p:condenserandprojectionNA;\\ Kirchauer:\sigma =N{A}_c/N{A}_p,typical0.4to\\ 0.7;Sasian:\sigma =d/D;minimumpitchfor\\ illuminationtiltedto\sigma NA,sothezero\\ andonefirstorderpassthepupil\\ (computedfromtheECE695two-orderrule)\end{array}$ $\begin{array}{r}(computed:atNA1.35,p_{min}=102,84and75\\ \mathrm{nm}for\sigma =0.4,0.7and0.9,approaching\\ \lambda /(2NA)=71.5\mathrm{nm}as\sigma tendsto1)\end{array}$
Hopkins imaging	$I\left(x\right)={\sum }_{p^{\prime },p^{\prime \prime }}TCC\left(p^{\prime },p^{\prime \prime }\right)T_{p^{\prime }}{T}_{p^{\prime \prime }}^{\ast }{e}^{-j2\pi \left(p^{\prime }-p^{\prime \prime }\right)x/a}$	$TCC\left(p^{\prime },p^{\prime \prime }\right)={\sum }_p{A}_p^{2}P\left(p+p^{\prime }\right)P^{\ast }\left(p+p^{\prime \prime }\right)$	$I:aerial-imageintensity$ $T_p:maskdiffractionorders$ $A_p:source-pointamplitudes$ $\begin{array}{r}P:pupilfunctionincludingdefocusand\\ aberrations\end{array}$ $\begin{array}{r}a:maskperiod;Kirchauer,Eqs.4.89and\\ 4.90(onedimensionwrittenhere):TCCs\\ dependonlyonillumination,NA,defocus\\ andaberrationsandarecomputedonce\end{array}$ $\begin{array}{r}(myreading:changingtheilluminator,as\\ FlexRaydoes,changestheTCCs,whileOPC\\ changes{T}_p)\end{array}$
Off-axis illumination	$\sin {\theta }_{ill}=\sigma NA$	$\text{annular, dipole, quadrupole}$	$\begin{array}{r}{\theta }_{ill}:illuminationangleatthemaskside;\\ Kirchauer:annularandquadrupole\\ aperturesremovezero-orderwavesthat\\ carrynoinformationandraiseresolution\\ withoutreducingDOF;Sasian:annular,\\ dipoleandquadrupolesources;ASML:\\ dipoleandC-quadproductionresolutions\end{array}$ $\begin{array}{r}(computed:38\mathrm{nm}dipoleagainst40\mathrm{nm}\\ C-quadonthesamelensis5percent\\ finer)\end{array}$

Polarization, Control and Throughput

Name	Formula	Related Formulae	Variables/Constants
Two-beam polarization contrast	$I_{TE}=2E^2\left(1+\cos \delta \right)$	$I_{TM}=2E^2\left(1+\cos \delta \cos 2\theta \right)$	$\delta :phasedifferenceacrossthefringe$ $\begin{array}{r}\theta :half-anglebetweentheinterfering\\ beamsinthemedium;Estroffetal.:these\\ intensities,TMcontrastzeroat45\\ degrees;Smithetal.:TMcontrastzeroat\\ \pi /4,over95percentofTEcoupledinto\\ resistinwater\end{array}$ $\begin{array}{r}(assumed:resistindex1.7;computed:for\\ a76\mathrm{nm}pitch,\lambda /(2p)=1.27,TMcontrast\\ |\cos 2\theta |is0.56inwaterbutonly0.12in\\ resist)\end{array}$
Immersion thermal tolerance	$\frac{dn}{dT}\approx 150\mathrm{ppm}{/}^{\circ }\mathrm{C}$	$\Delta T_{max}=\frac{\Delta n_{allowed}}{dn/dT}$	$\begin{array}{r}Smithetal.:about150\mathrm{ppm}{/}^{\circ }\mathrm{C},index\\ controlof28to42\mathrm{ppm}at1\mathrm{mm}working\\ distanceforNA1.2to0.7,0.1{}^{\circ }Ccontrol\\ adequate\end{array}$ $\begin{array}{r}(computed:28\mathrm{ppm}allows0.19{}^{\circ }C,so0.1\\ {}^{\circ }Ccontrolleavesafactorof1.9margin)\end{array}$
Scanner productivity and overlay	$\text{WPH}\approx \frac{4600}{24}$	$O{L}_{on-product}\le 2.5\mathrm{nm}$	$\begin{array}{r}ASML:4600wafersperdaychampion\\ productivity,2.5\mathrm{nm}cross-matching\\ on-productoverlay;OverlayMetrologyTool\\ pageforhowoverlayismeasured\end{array}$ $\begin{array}{r}(computed:4600perdayisabout192\\ wafersperhour)\end{array}$

Steppers, Scanners and Dry DUV

Name	Formula	Related Formulae	Variables/Constants
Stepper against scanner	$\text{stepper: whole field at once},M=5$	$\text{scanner: slit scanned with reticle and wafer},M=4$	$\begin{array}{r}M:reductionratio;NPTELL14:steppers\\ exposeandstepfieldbyfield,5:1most\\ common,eachfieldfocusedandaligned\\ separately;LaserFocusWorld:\\ step-and-scansendslightonlythroughthe\\ lenssweetspot,cuttingdistortionand\\ focaldeviation,4Xallowingalarger\\ image\end{array}$ $\begin{array}{r}(myreading:scanningtradesalarge\\ staticlensfieldforanarrowslitswept\\ acrossthefield)\end{array}$
Lens field limit	$a_{die,max}\approx 0.707D$	$A_{field}=26\times 33\mathrm{mm}(\text{scanner})$	$\begin{array}{r}D:lensfielddiameter;NPTELL14:largest\\ squaredie;EUVScannerpage(Fomenkovet\\ al.)forthe26x33\mathrm{mm}scannerfield\end{array}$ $\begin{array}{r}(computed:asquareinscribedinacircle\\ hassideD/\sqrt{2}=0.707D)\end{array}$
Dry KrF and ArF	${\lambda }_{KrF}=248\mathrm{nm},{\lambda }_{ArF}=193\mathrm{nm}$	$N{A}_{dry}\approx 0.8\text{to}0.93$	$\begin{array}{r}NPTELL14:KrFusedtoabout130\mathrm{nm}and\\ introducingchemicallyamplifiedresists,\\ ArFNAabout0.8withapracticallimit\\ about0.9;ECE695:drylimit0.93;ASML\\ DUVpage:KrFNXT:870B400wph,ArFdry\\ XT:1460K65\mathrm{nm}andNXT:1470above300wph\end{array}$ $\begin{array}{r}(computed:65\mathrm{nm}atNA0.93is{k}_1=0.31;\\ 130\mathrm{nm}KrFat0.93is0.49;KrFtoArFis\\ a1.28timeswavelengthreduction)\end{array}$

Important Questions

Ques: How does immersion improve resolution if the light is still 193 nm?

By raising the numerical aperture above 1. Resolution scales as $k_1\lambda /NA$, with $k_1$ no lower than 0.25 because two diffraction orders must reach the wafer (ECE 695). In air, NA is limited to about 0.93 (ECE 695); with water of index 1.44 at 193 $\mathrm{nm}$ between lens and wafer, the wavelength in the gap shortens to about 134 $\mathrm{nm}$ (Smith et al.). ASML’s NXT:2000i reaches NA 1.35 and prints 38 $\mathrm{nm}$ with dipole and 40 $\mathrm{nm}$ with C-quad illumination (ASML page), which correspond to $k_1$ = 0.266 and 0.280 (computed). At the 0.25 limit, NA 1.35 would give 35.7 $\mathrm{nm}$ half pitch (computed), so production tools already run within about 7 per cent of the single-exposure limit. Reference: ECE 695 Nanometer Scale Patterning and Processing, Lecture 7, ‘Optical Lithography - Lithography System’, Spring 2016, nanoHUB resource 24304 (author not listed): R = k1 lambda/NA with k1 limit 0.25 (52 nm at 193 nm and NA 0.93), DOF = k2 lambda/NA^2 for NA up to 0.5 with a different form above, Abbe minimum pitch lambda/(2NA) needing at least two diffraction orders, oblique illumination, dry NA limit about 0.93 (68 degrees), NA 1.4 relevant through 193 nm immersion, g-line to F2 wavelengths (fetch-tool summary); B. W. Smith, H. Kang, A. Bourov, F. Cropanese and Y. Fan, ‘Water Immersion Optical Lithography for the 45nm Node’, RIT (SPIE 5040): water index 1.44 at 193 nm, absorption below 0.05 per cm at 193 nm, wavelength in water about 134 nm, DOF = +/- k2 lambda_i/sin^2 theta scaling linearly with index, dn/dT about 150 ppm per degree C, 0.1 C control adequate, index control 28 to 42 ppm at 1 mm working distance, microbubbles and resist outgassing, TM contrast zero at pi/4, 40 per cent TE loss at NA 0.95 in air against over 95 per cent coupling in water, sub-0.3 k1 likely only with polarization (fetch-tool summary); ASML, ‘TWINSCAN NXT:2000i’ product page: 193 nm ArF, NA 1.35, dual-stage immersion, production resolution 38 nm with dipole and 40 nm with C-quad, 2.5 nm cross-matching on-product overlay with ORION alignment and UVLS-2 levelling sensors, 4600 wafers per day champion productivity, FlexRay programmable illuminator (fetch-tool summary).

Ques: Is the immersion NA limit 1.35 or 1.4?

The sources quote different figures. ECE 695 says NA 1.4 is relevant through 193 $\mathrm{nm}$ immersion, while ASML’s production NXT:2000i is specified at NA 1.35 (ASML page). Both lie below the water index of 1.44 (Smith et al.), which is the hard ceiling because NA equals $n\sin \theta$ (my reading). At NA 1.35 the ray angle in water is already 69.6 degrees, about the same as the 68.4 degrees of a dry NA 0.93 lens (computed), and NA 1.4 would need 76.5 degrees (computed). The production value of 1.35 is therefore the more likely practical limit, with 1.4 a design target where lens size, polarization loss and DOF become severe (my reading). Reference: ECE 695 Nanometer Scale Patterning and Processing, Lecture 7, ‘Optical Lithography - Lithography System’, Spring 2016, nanoHUB resource 24304 (author not listed): R = k1 lambda/NA with k1 limit 0.25 (52 nm at 193 nm and NA 0.93), DOF = k2 lambda/NA^2 for NA up to 0.5 with a different form above, Abbe minimum pitch lambda/(2NA) needing at least two diffraction orders, oblique illumination, dry NA limit about 0.93 (68 degrees), NA 1.4 relevant through 193 nm immersion, g-line to F2 wavelengths (fetch-tool summary); ASML, ‘TWINSCAN NXT:2000i’ product page: 193 nm ArF, NA 1.35, dual-stage immersion, production resolution 38 nm with dipole and 40 nm with C-quad, 2.5 nm cross-matching on-product overlay with ORION alignment and UVLS-2 levelling sensors, 4600 wafers per day champion productivity, FlexRay programmable illuminator (fetch-tool summary); B. W. Smith, H. Kang, A. Bourov, F. Cropanese and Y. Fan, ‘Water Immersion Optical Lithography for the 45nm Node’, RIT (SPIE 5040): water index 1.44 at 193 nm, absorption below 0.05 per cm at 193 nm, wavelength in water about 134 nm, DOF = +/- k2 lambda_i/sin^2 theta scaling linearly with index, dn/dT about 150 ppm per degree C, 0.1 C control adequate, index control 28 to 42 ppm at 1 mm working distance, microbubbles and resist outgassing, TM contrast zero at pi/4, 40 per cent TE loss at NA 0.95 in air against over 95 per cent coupling in water, sub-0.3 k1 likely only with polarization (fetch-tool summary).

Ques: Why do the two depth-of-focus formulas disagree at high NA?

Because one is paraxial. ECE 695 gives $DOF=k_2\lambda /N{A}^2$ for NA up to about 0.5 and a different form above. Smith et al. write $DOF=\pm k_2{\lambda }_i/{\sin }^2\theta$ with the wavelength and angle in the imaging medium, and state that DOF then grows linearly with index. For a 110 $\mathrm{nm}$ pitch, the medium form gives $\pm 251k_2$ $\mathrm{nm}$ dry and $\pm 361k_2$ $\mathrm{nm}$ in water, a gain of 1.44 (computed), while the paraxial form would predict no gain at a fixed pitch. At NA above 1 the paraxial expression is the less reliable, since the small-angle approximation fails at 60 to 70 degree rays, so Smith’s form is more likely correct for immersion (my reading). Reference: ECE 695 Nanometer Scale Patterning and Processing, Lecture 7, ‘Optical Lithography - Lithography System’, Spring 2016, nanoHUB resource 24304 (author not listed): R = k1 lambda/NA with k1 limit 0.25 (52 nm at 193 nm and NA 0.93), DOF = k2 lambda/NA^2 for NA up to 0.5 with a different form above, Abbe minimum pitch lambda/(2NA) needing at least two diffraction orders, oblique illumination, dry NA limit about 0.93 (68 degrees), NA 1.4 relevant through 193 nm immersion, g-line to F2 wavelengths (fetch-tool summary); B. W. Smith, H. Kang, A. Bourov, F. Cropanese and Y. Fan, ‘Water Immersion Optical Lithography for the 45nm Node’, RIT (SPIE 5040): water index 1.44 at 193 nm, absorption below 0.05 per cm at 193 nm, wavelength in water about 134 nm, DOF = +/- k2 lambda_i/sin^2 theta scaling linearly with index, dn/dT about 150 ppm per degree C, 0.1 C control adequate, index control 28 to 42 ppm at 1 mm working distance, microbubbles and resist outgassing, TM contrast zero at pi/4, 40 per cent TE loss at NA 0.95 in air against over 95 per cent coupling in water, sub-0.3 k1 likely only with polarization (fetch-tool summary).

Ques: How do sigma and off-axis illumination set the smallest pitch?

By tilting the zero order so a first order still fits. Kirchauer defines $\sigma =N{A}_c/N{A}_p$, typically 0.4 to 0.7, and notes that small $\sigma$ steepens edges but adds overshoot and proximity effects. With illumination at $\sin {\theta }_{ill}=\sigma NA$, the zero and one first order pass the pupil down to $p=\lambda /(NA(1+\sigma ))$ (computed from the two-order rule in ECE 695): 102, 84 and 75 $\mathrm{nm}$ for $\sigma$ = 0.4, 0.7 and 0.9 at NA 1.35 (computed). Annular and quadrupole apertures remove the zero-order light that carries no information and raise resolution without reducing DOF (Kirchauer). In the Hopkins model this source shape enters only the TCCs, which are computed once per illuminator (Kirchauer). Reference: H. Kirchauer, PhD thesis, Institute for Microelectronics, TU Vienna (1998), sections 2.3.3 and 4.3.2: sigma = NA_c/NA_p, typical 0.4 to 0.7, small sigma steepening edges but adding overshoot and proximity effects, annular and quadrupole apertures removing zero-order waves and raising resolution without reducing DOF, Hopkins imaging with transmission cross coefficients computed once for given illumination, NA, defocus and aberrations (fetch-tool summary); ECE 695 Nanometer Scale Patterning and Processing, Lecture 7, ‘Optical Lithography - Lithography System’, Spring 2016, nanoHUB resource 24304 (author not listed): R = k1 lambda/NA with k1 limit 0.25 (52 nm at 193 nm and NA 0.93), DOF = k2 lambda/NA^2 for NA up to 0.5 with a different form above, Abbe minimum pitch lambda/(2NA) needing at least two diffraction orders, oblique illumination, dry NA limit about 0.93 (68 degrees), NA 1.4 relevant through 193 nm immersion, g-line to F2 wavelengths (fetch-tool summary); J. Sasian, ‘Lithography optics’, OPTI 696A lecture, University of Arizona: g-line 436, i-line 365, KrF 248, ArF 193 nm, sigma = d/D, Koehler illumination uniform to about 1 per cent, annular, dipole and quadrupole sources, doubly telecentric projection lens (fetch-tool summary).

Ques: Why does polarization matter in a high-NA immersion scanner?

Because TM fringes lose contrast at large angles. For two beams, Estroff et al. give $I_{TE}=2E^2(1+\cos \delta )$ and $I_{TM}=2E^2(1+\cos \delta \cos 2\theta )$, so TM contrast falls to zero when each beam is 45 degrees from the normal. For a 76 $\mathrm{nm}$ pitch, $\lambda /(2p)$ = 1.27, giving TM contrast of 0.56 in water and only 0.12 in an assumed resist of index 1.7 (computed). Smith et al. add that dry imaging at NA 0.95 can lose up to 40 per cent of the TE image at the air-resist interface, while water couples over 95 per cent, and that $k_1$ below 0.3 is likely only with polarization control. Estroff et al. also find that masks themselves polarize diffraction orders. Reference: A. Estroff, Y. Fan, A. Bourov, B. Smith, P. Foubert, L. H. A. Leunissen, V. Philipsen and Y. Aksenov, ‘Mask induced polarization effects at high NA’, RIT and IMEC (JM3 submission 2004): two-beam intensities I_TE = 2E^2(1 + cos delta) and I_TM = 2E^2(1 + cos delta cos 2 theta), TM contrast falling to zero at 45 degrees, higher NA giving greater polarization effects, binary masks polarizing the zero order TM and the first order mainly TE, over 25 per cent contrast gain from TE first order (fetch-tool summary); B. W. Smith, H. Kang, A. Bourov, F. Cropanese and Y. Fan, ‘Water Immersion Optical Lithography for the 45nm Node’, RIT (SPIE 5040): water index 1.44 at 193 nm, absorption below 0.05 per cm at 193 nm, wavelength in water about 134 nm, DOF = +/- k2 lambda_i/sin^2 theta scaling linearly with index, dn/dT about 150 ppm per degree C, 0.1 C control adequate, index control 28 to 42 ppm at 1 mm working distance, microbubbles and resist outgassing, TM contrast zero at pi/4, 40 per cent TE loss at NA 0.95 in air against over 95 per cent coupling in water, sub-0.3 k1 likely only with polarization (fetch-tool summary).

Ques: What immersion-specific defects and controls matter for yield?

Bubbles, outgassing and index drift. Smith et al. warn that resists can release gas under exposure, that microbubbles form where water traps air in surface cavities, and that degassed water or a retaining structure may be needed. The water index changes about 150 $\mathrm{ppm}$ per degree, and their table requires index control of 28 to 42 $\mathrm{ppm}$ at 1 $\mathrm{mm}$ working distance; 28 $\mathrm{ppm}$ allows 0.19 ${}^{\circ }\mathrm{C}$ (computed), so their 0.1 ${}^{\circ }\mathrm{C}$ control leaves a margin of about 1.9 (computed). ASML specifies 2.5 $\mathrm{nm}$ on-product overlay and 4600 wafers per day, about 192 per hour (computed). Defects printed by the scanner are then found by wafer inspection and review, overlay is measured on the overlay tool, and printed CD on the CD-SEM (Overlay Metrology Tool page; CD-SEM page). Reference: B. W. Smith, H. Kang, A. Bourov, F. Cropanese and Y. Fan, ‘Water Immersion Optical Lithography for the 45nm Node’, RIT (SPIE 5040): water index 1.44 at 193 nm, absorption below 0.05 per cm at 193 nm, wavelength in water about 134 nm, DOF = +/- k2 lambda_i/sin^2 theta scaling linearly with index, dn/dT about 150 ppm per degree C, 0.1 C control adequate, index control 28 to 42 ppm at 1 mm working distance, microbubbles and resist outgassing, TM contrast zero at pi/4, 40 per cent TE loss at NA 0.95 in air against over 95 per cent coupling in water, sub-0.3 k1 likely only with polarization (fetch-tool summary); ASML, ‘TWINSCAN NXT:2000i’ product page: 193 nm ArF, NA 1.35, dual-stage immersion, production resolution 38 nm with dipole and 40 nm with C-quad, 2.5 nm cross-matching on-product overlay with ORION alignment and UVLS-2 levelling sensors, 4600 wafers per day champion productivity, FlexRay programmable illuminator (fetch-tool summary); Overlay Metrology Tool page of this vault; CD-SEM page of this vault.

Ques: How do steppers and scanners differ?

In how the reticle field is exposed. NPTEL describes the step-and-repeat stepper of the early 1980s: the whole reticle field is imaged at once, usually at 5:1, the stage steps to the next field, and each field is focused and aligned separately, which makes wafer expansion and flatness unimportant; the largest square die is about $0.707D$ for a lens field of diameter D. Laser Focus World reports Nikon’s 1995 KrF step-and-scan system, which scans reticle and wafer together so light passes only the lens sweet spot, reducing distortion and focal deviation, and uses 4X to print a larger image; it resolved better than 0.25 $\mu \mathrm{m}$. Scanning thus replaces a large static field with a slit swept across a 26 x 33 $\mathrm{mm}$ field (my reading; EUV Scanner page). Reference: NPTEL course 103105065, Module 11 Lecture 14, ‘Photolithography - 4 (Continued)’: steppers introduced in the early 1980s stepping the wafer between exposures of a reticle field, 5:1 the most common reduction, largest square die about 0.707 D for lens diameter D, each field focused and aligned separately, KrF 248 nm introducing chemically amplified resists and used to about 130 nm, ArF 193 nm capable of about 40 nm, F2 157 nm, W = k1 lambda/NA with k1 about 0.25, NA about 0.4 for i-line and 0.8 for ArF with a practical air limit of about 0.9, water immersion n about 1.47 with NA 0.93 giving about 35 nm (fetch-tool summary); Laser Focus World, ‘Excimer-based step-and-repeat scanner achieves 0.25-um resolution’ (1 March 1995, no byline): Nikon step-and-scan system with a 248 nm KrF laser, reticle and wafer stages scanned together so light passes only the lens sweet spot, reducing distortion and focal deviation, 4X reduction against 5X for steppers allowing a larger image, resolution better than 0.25 um for 256-Mbit DRAM, throughput comparable to i-line steppers (fetch-tool summary).

Ques: Is the water index at 193 nm 1.44 or 1.47, and where do dry KrF and ArF tools fit?

1.44 is more likely. Smith et al. measured 1.44 at 193 $\mathrm{nm}$, whereas the NPTEL lecture quotes about 1.47 and estimates 35 $\mathrm{nm}$ half pitch for NA 0.93 in water; with 1.44 the same estimate is 36.0 $\mathrm{nm}$ against 35.3 $\mathrm{nm}$ (computed), so the difference matters little, but Smith’s value is a direct measurement (my reading). NPTEL places KrF at 248 $\mathrm{nm}$ down to about 130 $\mathrm{nm}$ and dry ArF NA near 0.8 with a 0.9 practical limit, while ECE 695 gives 0.93. ASML still sells dry tools: the KrF NXT:870B at 400 wafers per hour, the ArF XT:1460K at 65 $\mathrm{nm}$, $k_1$ = 0.31 at NA 0.93 (computed), and the NXT:1470 above 300 wafers per hour. Reference: B. W. Smith, H. Kang, A. Bourov, F. Cropanese and Y. Fan, ‘Water Immersion Optical Lithography for the 45nm Node’, RIT (SPIE 5040): water index 1.44 at 193 nm, absorption below 0.05 per cm at 193 nm, wavelength in water about 134 nm, DOF = +/- k2 lambda_i/sin^2 theta scaling linearly with index, dn/dT about 150 ppm per degree C, 0.1 C control adequate, index control 28 to 42 ppm at 1 mm working distance, microbubbles and resist outgassing, TM contrast zero at pi/4, 40 per cent TE loss at NA 0.95 in air against over 95 per cent coupling in water, sub-0.3 k1 likely only with polarization (fetch-tool summary); NPTEL course 103105065, Module 11 Lecture 14, ‘Photolithography - 4 (Continued)’: steppers introduced in the early 1980s stepping the wafer between exposures of a reticle field, 5:1 the most common reduction, largest square die about 0.707 D for lens diameter D, each field focused and aligned separately, KrF 248 nm introducing chemically amplified resists and used to about 130 nm, ArF 193 nm capable of about 40 nm, F2 157 nm, W = k1 lambda/NA with k1 about 0.25, NA about 0.4 for i-line and 0.8 for ArF with a practical air limit of about 0.9, water immersion n about 1.47 with NA 0.93 giving about 35 nm (fetch-tool summary); ECE 695 Nanometer Scale Patterning and Processing, Lecture 7, ‘Optical Lithography - Lithography System’, Spring 2016, nanoHUB resource 24304 (author not listed): R = k1 lambda/NA with k1 limit 0.25 (52 nm at 193 nm and NA 0.93), DOF = k2 lambda/NA^2 for NA up to 0.5 with a different form above, Abbe minimum pitch lambda/(2NA) needing at least two diffraction orders, oblique illumination, dry NA limit about 0.93 (68 degrees), NA 1.4 relevant through 193 nm immersion, g-line to F2 wavelengths (fetch-tool summary); ASML, ‘DUV lithography systems’ product page: KrF TWINSCAN NXT:870B at 400 wafers per hour, ArF dry XT:1460K at 65 nm and NXT:1470 above 300 wafers per hour, ArF immersion NXT:1980Fi to 2150i, immersion systems above 6000 wafers per day (fetch-tool summary).

References

1. ECE 695 Nanometer Scale Patterning and Processing, Lecture 7, “Optical Lithography - Lithography System,” Spring 2016, nanoHUB resource 24304. Available: https://nanohub.org/resources/24304/download/2016.01.27-ECE695Q-L07.pdf
2. B. W. Smith, H. Kang, A. Bourov, F. Cropanese and Y. Fan, “Water immersion optical lithography for the 45nm node,” Proc. SPIE 5040, Rochester Institute of Technology. Available: https://rit.edu/~w-lithography/research/immersion/SPIE_5040_58_immersion.pdf
3. A. Estroff, Y. Fan, A. Bourov, B. Smith, P. Foubert, L. H. A. Leunissen, V. Philipsen and Y. Aksenov, “Mask induced polarization effects at high NA,” Rochester Institute of Technology and IMEC. Available: https://rit.edu/~w-lithography/research/polarization/5754-52%20Revision%201.pdf
4. ASML, “TWINSCAN NXT:2000i.” Available: https://asml.com/products/duv-lithography-systems/twinscan-nxt2000i
5. H. Kirchauer, PhD thesis, Institute for Microelectronics, TU Vienna, 1998, sections 2.3.3 and 4.3.2. Available: https://www.iue.tuwien.ac.at/phd/kirchauer/node21.html and https://www.iue.tuwien.ac.at/phd/kirchauer/node62.html
6. J. Sasian, “Lithography optics,” OPTI 696A lecture, University of Arizona. Available: https://wp.optics.arizona.edu/jsasian/wp-content/uploads/sites/33/2018/11/Lithography-optics-basics.pdf
7. NPTEL, course 103105065, Module 11, Lecture 14, “Photolithography - 4 (continued).” Available: https://archive.nptel.ac.in/content/storage2/courses/103105065/M11l14.pdf
8. “Excimer-based step-and-repeat scanner achieves 0.25-um resolution,” Laser Focus World, 1 March 1995. Available: https://www.laserfocusworld.com/lasers-sources/article/16554120/excimerbased-stepandrepeat-scanner-achieves-025181m-resolution
9. ASML, “DUV lithography systems.” Available: https://www.asml.com/en/products/duv-lithography-systems
10. Overlay Metrology Tool page of this vault.
11. CD-SEM page of this vault.

Related pages in this vault: Overlay Metrology Tool, CD-SEM, Optical CD Scatterometer (OCD), Patterned-Wafer Optical Inspection, E-Beam Inspection and Defect Review, Photoresist and Process Window, Resolution Enhancement and Multiple Patterning, Light-Matter Interaction for Optical Inspection, Defect Sources and Classification, EUV Scanner, Spin Coater and Resist Track, Contact and Proximity Mask Aligner, Nanoimprint Lithography Tool.