Contact and Proximity Mask Aligner

1. Theory

• Definition: A mask aligner copies a 1x mask onto a resist-coated wafer by shadow printing, with the mask touching the resist (contact) or held a few tens of micrometres above it (proximity) (Chen; Voelkel et al.).
• Diffraction limit: Resolution follows $\sqrt{\lambda (s+d/2)}$, so it degrades as the square root of the gap (Chen; Voelkel et al.).
• Contact against proximity: Contact reaches sub-micron resolution but damages and contaminates the mask; proximity protects the mask but resolves about 3 to 5 $\mu \mathrm{m}$ (Chen; Voelkel et al.).
• Illumination: Collimation and uniformity of the mercury-lamp illumination shape the diffraction pattern at the wafer (Voelkel et al.).

2. Working Principle

• Load and level: The wafer is brought parallel to the mask at a set gap or into contact (UCSB).
• Align: Top or backside microscopes align wafer and mask marks to about 0.5 to 1 $\mu \mathrm{m}$ (UCSB).
• Contact mode: Soft, hard or vacuum contact progressively improves resolution (UCSB).
• Expose: A mercury arc lamp delivers a calibrated dose through collimating and homogenizing optics (UCSB; Voelkel et al.).

3. Applications

• MEMS and packaging: Large-feature, low-cost layers and backside alignment (UCSB; Voelkel et al.).
• Thick resists and bumps: Proximity printing at micrometre scale (Voelkel et al.).
• Periodic patterns: Talbot printing of gratings and dot arrays (Voelkel et al.).

Shadow-Printing Resolution

Name	Formula	Related Formulae	Variables/Constants
Shadow-printing resolution	$b_{min}=\frac{3}{2}\sqrt{\lambda \left(s+\frac{d}{2}\right)}$	$b_{contact}=\frac{3}{2}\sqrt{\frac{\lambda d}{2}}(s=0)$	$b_{min}:minimumhalf-pitch$ $\lambda :wavelength$ $s:mask-resistgap$ $\begin{array}{r}d:resistthickness;Chen:\sqrt{\lambda (s+0.5d)}\\ dependencewithaconstantnear3\\ (garbled),g-lineexample;Voelkeletal.:\\ resolutiondegradingasthesquarerootof\\ thegap;3/2coefficient(computedto\\ reproduceChe{n}^{\prime }s0.5\mu \mathrm{m}contactexample;\\ seetheQ/A)\end{array}$ $\begin{array}{r}(computed:g-line,d=0.5\mu \mathrm{m}:0.49\mu \mathrm{m}\\ contactand3.17\mu \mathrm{m}ats=10\mu \mathrm{m})\end{array}$
Gap dependence	$\frac{b\left(s_2\right)}{b\left(s_1\right)}\approx \sqrt{\frac{s_2}{s_1}}(s\gg d)$	$s=30\text{to}100\mu \mathrm{m}\Rightarrow b\approx 3\text{to}5\mu \mathrm{m}$	$\begin{array}{r}Voelkeletal.:typicalgapsand\\ resolution,narrowwindowfor5\mu \mathrm{m}CDat30\\ \mu \mathrm{m};UCSB:10to300\mu \mathrm{m}programmablegap\end{array}$ $\begin{array}{r}(computed:i-linewith1.2\mu \mathrm{m}resistgives\\ 5.0\mu \mathrm{m}at30\mu \mathrm{m}and9.1\mu \mathrm{m}at100\mu \mathrm{m};\\ quadruplingthegapfrom10to40\mu \mathrm{m}nearly\\ doublesb)\end{array}$
Contact modes	$b:<0.8(\text{vacuum}),<1.5(\text{hard}),<2.5(\text{soft})\mu \mathrm{m}$	$b_{prox}\left(20\mu \mathrm{m}\right)<3.0\mu \mathrm{m}$	$\begin{array}{r}UCSB:SUSSMA6resolutionbymodeon1.2\mu \mathrm{m}\\ AZ4110;Chen:contactdamagesand\\ contaminatesthemask;Voelkeletal.:\\ frequentcleaningincontact\end{array}$ $\begin{array}{r}(computed:theformulagives0.70\mu \mathrm{m}for\\ i-linevacuumcontactand4.1\mu \mathrm{m}ata20\mu \mathrm{m}\\ gap)\end{array}$

Exposure and Alignment

Name	Formula	Related Formulae	Variables/Constants
Exposure dose and time	$t_{exp}=\frac{D}{I}$	$I_{365}=9mW/c{\mathrm{m}}^2,I_{405}=15mW/c{\mathrm{m}}^2$	$D:dose$ $\begin{array}{r}I:irradiance;UCSB:calibratedchannels,\\ driftcompensationasthelampages;\\ Voelkeletal.:mercurylines365,405and\\ 436\mathrm{nm}\end{array}$ $\begin{array}{r}(assumed:100mJ/c{\mathrm{m}}^2;computed:11sat9\\ mW/c{\mathrm{m}}^2)\end{array}$
Illumination quality	${\theta }_{coll}=\pm 3^{\circ }\text{to}\pm 5^{\circ }$	$\Delta I/I<\pm 2\%(\text{microlens integrator})$	$\begin{array}{r}{\theta }_{coll}:collimationhalf-angle;Voelkelet\\ al.:divergence,about300000microlens\\ channels,uniformitybetterthan\pm 2per\\ centagainst\pm 3to\pm 5percentforolder\\ integrators\end{array}$ $\begin{array}{r}(myreading:theangularspectrumsetshow\\ diffractionfringesfromthegapblur,so\\ itistunedalongwiththegap)\end{array}$
Alignment	${\Delta }_{align}:0.5\mu \mathrm{m}(\text{top}),1\mu \mathrm{m}(\text{back})$	${\Delta }_{prod}\approx \pm 2\mu \mathrm{m}$	$\begin{array}{r}UCSB:topsidedownto0.5\mu \mathrm{m},backsidedown\\ to1\mu \mathrm{m}at5xto20x;Voelkeletal.:\pm 0.5\\ \mu \mathrm{m}best,\pm 2\mu \mathrm{m}inproductionfrommask\\ heatingrun-out\end{array}$ $\begin{array}{r}(myreading:1xshadowprintingcopies\\ maskexpansiondirectly,somaskheating\\ becomesoverlayerror)\end{array}$

Context and Extensions

Name	Formula	Related Formulae	Variables/Constants
Comparison with projection	$R=0.61\frac{\lambda }{NA},b=k_1\frac{\lambda }{NA}$	$DOF=K_2\frac{\lambda }{N{A}^2}$	$\begin{array}{r}Chen:projectionforms,k_{1}0.3to0.9,DOF\\ example{K}_2=0.5at435\mathrm{nm}andNA0.6;\\ contactcheap,proximitylow\\ contamination,projectionexpensivebut\\ highresolutionandthroughput\end{array}$ $\begin{array}{r}(computed:Che{n}^{\prime }sDOFexamplegives0.60\\ \mu \mathrm{m})\end{array}$
Mask-aligner enhancements	$\text{OPC assist features, AAPSM}$	$\text{Talbot printing for periodic patterns}$	$\begin{array}{r}Voelkeletal.:OPCand\\ alternating-aperturephase-shiftmasksfor\\ aligners,Talbotlithographyprinting\\ sub-microngratingsacrossfullwafers,\\ wavelength-dependentTalbotplanes\end{array}$ $\begin{array}{r}(myreading:theseborrowthe\\ projection-lithographyenhancementideas\\ ontheResolutionEnhancementpage)\end{array}$
Wavelength options	$\lambda =254,365,405,436\mathrm{nm}(\mathrm{Hg})$	$\lambda =193,248,308\mathrm{nm}(\text{excimer})$	$\begin{array}{r}Voelkeletal.,Table1;UCSB:broadband\\ 350to450\mathrm{nm}\end{array}$ $\begin{array}{r}(computed:goingfrom436to254\mathrm{nm}\\ reducesbbyafactorof1.31atfixed\\ gap)\end{array}$

Important Questions

Ques: What limits the resolution of a proximity mask aligner?

Fresnel diffraction across the gap. Chen gives half-pitch resolution proportional to $\sqrt{\lambda (s+0.5d)}$, with s the gap and d the resist thickness, and Voelkel et al. state that resolution degrades with the square root of the gap, quoting about 3 to 5 $\mu \mathrm{m}$ at typical gaps of 30 to 100 $\mu \mathrm{m}$. With a coefficient of 3/2, i-line exposure of 1.2 $\mu \mathrm{m}$ resist gives 5.0 $\mu \mathrm{m}$ at 30 $\mu \mathrm{m}$ and 9.1 $\mu \mathrm{m}$ at 100 $\mu \mathrm{m}$, and quadrupling the gap from 10 to 40 $\mu \mathrm{m}$ nearly doubles the minimum feature (computed). Voelkel et al. add that the process window for 5 $\mu \mathrm{m}$ features at a 30 $\mu \mathrm{m}$ gap is very narrow. Shortening the wavelength from 436 to 254 $\mathrm{nm}$ helps only by $\sqrt{436/254}$ = 1.31 (computed). Reference: E. Chen, ‘Introduction to Nanofabrication’, Applied Physics 298r, Center for Imaging and Mesoscale Structure, Harvard University (12 April 2004): contact half-pitch b = k sqrt(lambda d) and proximity b = k sqrt(lambda (s + 0.5d)) with k about 3 (equations garbled), g-line example d = 0.5 um and s = 10 um giving about 0.5 um contact and 2.6 um proximity, projection R = 0.61 lambda/NA and b = k1 lambda/NA with k1 0.3 to 0.9, DOF = K2 lambda/NA^2, contact printing cheap with high resolution but mask contamination and damage, proximity low contamination but poor resolution (fetch-tool summary); R. Voelkel, U. Vogler, A. Bramati and W. Noell, ‘Micro-optics and lithography simulation are key enabling technologies for shadow printing lithography in mask aligners’, Adv. Opt. Technol. 4, 63 (2015): contact printing sub-micron but needing frequent mask cleaning, proximity resolution about 3 to 5 um at typical 30 to 100 um gaps, resolution degrading with the square root of the gap, narrow window for 5 um CD at 30 um gap, collimation +/-3 to 5 degrees, microlens integrators with about 300000 channels giving better than +/-2 per cent uniformity, mercury lines 254, 365, 405, 436 nm and excimer 193, 248, 308 nm, alignment +/-0.5 um best and +/-2 um in production from mask heating, Talbot lithography for periodic patterns, OPC assist features and AAPSM for aligners (fetch-tool summary).

Ques: Is the shadow-printing coefficient 3 or 3/2, and does Chen’s example fit?

Only partly. Chen’s text, garbled in extraction, gives a constant near 3, but his own contact example, 0.5 $\mu \mathrm{m}$ at 435 $\mathrm{nm}$ with 0.5 $\mu \mathrm{m}$ resist, is reproduced by $\frac{3}{2}\sqrt{\lambda d/2}$ = 0.49 $\mu \mathrm{m}$ (computed), so 3/2 for half-pitch is more likely, with 3 describing the full period (my reading). His proximity example, 2.6 $\mu \mathrm{m}$ at a 10 $\mu \mathrm{m}$ gap, does not fit either: 3/2 gives 3.17 $\mu \mathrm{m}$ and 3 gives 6.3 $\mu \mathrm{m}$ (computed); 2.6 $\mu \mathrm{m}$ would need a gap of about 6.7 $\mu \mathrm{m}$ (computed). The SUSS MA6 figure of below 3 $\mu \mathrm{m}$ at 20 $\mu \mathrm{m}$, against 4.1 $\mu \mathrm{m}$ from the formula (computed), also beats the estimate. The formula is therefore a conservative guide, and measured tool data are the more reliable figures (my reading). Reference: E. Chen, ‘Introduction to Nanofabrication’, Applied Physics 298r, Center for Imaging and Mesoscale Structure, Harvard University (12 April 2004): contact half-pitch b = k sqrt(lambda d) and proximity b = k sqrt(lambda (s + 0.5d)) with k about 3 (equations garbled), g-line example d = 0.5 um and s = 10 um giving about 0.5 um contact and 2.6 um proximity, projection R = 0.61 lambda/NA and b = k1 lambda/NA with k1 0.3 to 0.9, DOF = K2 lambda/NA^2, contact printing cheap with high resolution but mask contamination and damage, proximity low contamination but poor resolution (fetch-tool summary); UCSB Nanofabrication Facility wiki, ‘Contact Aligner (SUSS MA-6)’: 350 W Hg arc lamp with UV400 optics (350 to 450 nm), 9 mW/cm2 at 365 nm and 15 mW/cm2 at 405 nm, vacuum contact below 0.8 um, hard contact below 1.5 um, soft contact below 2.5 um, proximity at 20 um below 3.0 um (1.2 um AZ 4110), gap 10 to 300 um, topside alignment down to 0.5 um and backside down to 1 um (fetch-tool summary).

Ques: How do soft, hard and vacuum contact differ?

In how tightly the mask meets the resist. The UCSB SUSS MA6 page gives resolution below 2.5 $\mu \mathrm{m}$ in soft contact, below 1.5 $\mu \mathrm{m}$ in hard contact and below 0.8 $\mu \mathrm{m}$ in vacuum contact on 1.2 $\mu \mathrm{m}$ AZ 4110 resist, against below 3.0 $\mu \mathrm{m}$ in proximity at 20 $\mu \mathrm{m}$. The vacuum figure agrees with the contact formula, which gives 0.70 $\mu \mathrm{m}$ for i-line and that resist thickness (computed), because vacuum removes the residual gap. The cost is the mask: Chen notes that contact printing damages and contaminates masks and transfers defects, and Voelkel et al. say contact masks need frequent cleaning. Proximity printing is therefore chosen when mask life and defectivity matter more than resolution (my reading). Reference: UCSB Nanofabrication Facility wiki, ‘Contact Aligner (SUSS MA-6)’: 350 W Hg arc lamp with UV400 optics (350 to 450 nm), 9 mW/cm2 at 365 nm and 15 mW/cm2 at 405 nm, vacuum contact below 0.8 um, hard contact below 1.5 um, soft contact below 2.5 um, proximity at 20 um below 3.0 um (1.2 um AZ 4110), gap 10 to 300 um, topside alignment down to 0.5 um and backside down to 1 um (fetch-tool summary); E. Chen, ‘Introduction to Nanofabrication’, Applied Physics 298r, Center for Imaging and Mesoscale Structure, Harvard University (12 April 2004): contact half-pitch b = k sqrt(lambda d) and proximity b = k sqrt(lambda (s + 0.5d)) with k about 3 (equations garbled), g-line example d = 0.5 um and s = 10 um giving about 0.5 um contact and 2.6 um proximity, projection R = 0.61 lambda/NA and b = k1 lambda/NA with k1 0.3 to 0.9, DOF = K2 lambda/NA^2, contact printing cheap with high resolution but mask contamination and damage, proximity low contamination but poor resolution (fetch-tool summary); R. Voelkel, U. Vogler, A. Bramati and W. Noell, ‘Micro-optics and lithography simulation are key enabling technologies for shadow printing lithography in mask aligners’, Adv. Opt. Technol. 4, 63 (2015): contact printing sub-micron but needing frequent mask cleaning, proximity resolution about 3 to 5 um at typical 30 to 100 um gaps, resolution degrading with the square root of the gap, narrow window for 5 um CD at 30 um gap, collimation +/-3 to 5 degrees, microlens integrators with about 300000 channels giving better than +/-2 per cent uniformity, mercury lines 254, 365, 405, 436 nm and excimer 193, 248, 308 nm, alignment +/-0.5 um best and +/-2 um in production from mask heating, Talbot lithography for periodic patterns, OPC assist features and AAPSM for aligners (fetch-tool summary).

Ques: How are exposure and alignment set on a mask aligner?

From the lamp irradiance and the alignment microscopes. The UCSB MA6 uses a 350 W mercury arc with broadband 350 to 450 $\mathrm{nm}$ optics, calibrated to 9 $mW/c{\mathrm{m}}^2$ at 365 $\mathrm{nm}$ and 15 $mW/c{\mathrm{m}}^2$ at 405 $\mathrm{nm}$ with drift compensation, so an assumed 100 $mJ/c{\mathrm{m}}^2$ dose takes about 11 s at 365 $\mathrm{nm}$ (computed). Topside alignment reaches 0.5 $\mu \mathrm{m}$ and backside 1 $\mu \mathrm{m}$. Voelkel et al. quote plus or minus 0.5 $\mu \mathrm{m}$ at best and about plus or minus 2 $\mu \mathrm{m}$ in production, because mask heating causes run-out; in 1x printing any mask expansion maps directly onto the wafer (my reading). Their microlens integrators improve uniformity to better than plus or minus 2 per cent with plus or minus 3 to plus or minus 5 degree collimation. The resist itself is prepared on the track (Spin Coater and Resist Track page). Reference: UCSB Nanofabrication Facility wiki, ‘Contact Aligner (SUSS MA-6)’: 350 W Hg arc lamp with UV400 optics (350 to 450 nm), 9 mW/cm2 at 365 nm and 15 mW/cm2 at 405 nm, vacuum contact below 0.8 um, hard contact below 1.5 um, soft contact below 2.5 um, proximity at 20 um below 3.0 um (1.2 um AZ 4110), gap 10 to 300 um, topside alignment down to 0.5 um and backside down to 1 um (fetch-tool summary); R. Voelkel, U. Vogler, A. Bramati and W. Noell, ‘Micro-optics and lithography simulation are key enabling technologies for shadow printing lithography in mask aligners’, Adv. Opt. Technol. 4, 63 (2015): contact printing sub-micron but needing frequent mask cleaning, proximity resolution about 3 to 5 um at typical 30 to 100 um gaps, resolution degrading with the square root of the gap, narrow window for 5 um CD at 30 um gap, collimation +/-3 to 5 degrees, microlens integrators with about 300000 channels giving better than +/-2 per cent uniformity, mercury lines 254, 365, 405, 436 nm and excimer 193, 248, 308 nm, alignment +/-0.5 um best and +/-2 um in production from mask heating, Talbot lithography for periodic patterns, OPC assist features and AAPSM for aligners (fetch-tool summary); Spin Coater and Resist Track page of this vault.

Ques: When is a mask aligner used instead of a projection scanner?

When features are micrometres and cost matters. Chen summarizes contact printing as cheap and high-resolution but damaging to masks, proximity as clean but low-resolution, and projection as expensive but high-resolution and high-throughput; projection resolution follows $k_1\lambda /NA$ with DOF $K_2\lambda /N{A}^2$, 0.60 $\mu \mathrm{m}$ for his example (computed). Voelkel et al. describe proximity lithography as very cost-efficient for high-volume production and show that OPC, phase-shift masks and Talbot printing extend it toward sub-micron periodic patterns. Critical layers of advanced chips use projection scanners instead (High-NA DUV Immersion Scanner page), while aligners serve MEMS, packaging and thick-resist layers (my reading). The UCSB tool also offers backside alignment, which wafer-level MEMS and bonding processes need. Reference: E. Chen, ‘Introduction to Nanofabrication’, Applied Physics 298r, Center for Imaging and Mesoscale Structure, Harvard University (12 April 2004): contact half-pitch b = k sqrt(lambda d) and proximity b = k sqrt(lambda (s + 0.5d)) with k about 3 (equations garbled), g-line example d = 0.5 um and s = 10 um giving about 0.5 um contact and 2.6 um proximity, projection R = 0.61 lambda/NA and b = k1 lambda/NA with k1 0.3 to 0.9, DOF = K2 lambda/NA^2, contact printing cheap with high resolution but mask contamination and damage, proximity low contamination but poor resolution (fetch-tool summary); R. Voelkel, U. Vogler, A. Bramati and W. Noell, ‘Micro-optics and lithography simulation are key enabling technologies for shadow printing lithography in mask aligners’, Adv. Opt. Technol. 4, 63 (2015): contact printing sub-micron but needing frequent mask cleaning, proximity resolution about 3 to 5 um at typical 30 to 100 um gaps, resolution degrading with the square root of the gap, narrow window for 5 um CD at 30 um gap, collimation +/-3 to 5 degrees, microlens integrators with about 300000 channels giving better than +/-2 per cent uniformity, mercury lines 254, 365, 405, 436 nm and excimer 193, 248, 308 nm, alignment +/-0.5 um best and +/-2 um in production from mask heating, Talbot lithography for periodic patterns, OPC assist features and AAPSM for aligners (fetch-tool summary); High-NA DUV Immersion Scanner page of this vault; UCSB Nanofabrication Facility wiki, ‘Contact Aligner (SUSS MA-6)’: 350 W Hg arc lamp with UV400 optics (350 to 450 nm), 9 mW/cm2 at 365 nm and 15 mW/cm2 at 405 nm, vacuum contact below 0.8 um, hard contact below 1.5 um, soft contact below 2.5 um, proximity at 20 um below 3.0 um (1.2 um AZ 4110), gap 10 to 300 um, topside alignment down to 0.5 um and backside down to 1 um (fetch-tool summary).

References

1. E. Chen, “Introduction to nanofabrication,” Applied Physics 298r, Harvard University, 2004. Available: https://www.mrsec.harvard.edu/education/ap298r2004/Erli%20chenFabrication%20I%20-%20Lithography-1.pdf
2. R. Voelkel, U. Vogler, A. Bramati and W. Noell, “Micro-optics and lithography simulation are key enabling technologies for shadow printing lithography in mask aligners,” Advanced Optical Technologies, vol. 4, no. 1, pp. 63-69, 2015. Available: https://www.frontiersin.org/articles/10.1515/aot-2014-0065/pdf
3. UCSB Nanofabrication Facility, “Contact Aligner (SUSS MA-6),” wiki page. Available: https://wiki.nanofab.ucsb.edu/wiki/Contact_Aligner_(SUSS_MA-6)
4. High-NA DUV Immersion Scanner page of this vault.
5. Spin Coater and Resist Track page of this vault.

Related pages in this vault: Spin Coater and Resist Track, High-NA DUV Immersion Scanner, Resolution Enhancement and Multiple Patterning, Photoresist and Process Window, Laser Direct-Write and Interference Lithography System.