Litografia da fascio elettronico

SEM Imaging - System overview - Types of Electron Beam columns - Magnification and deflection system - Focussing (contamination dots) - Imaging

Technical setup of EBL tools Source: SPIE Handbook of Microlithography

Types of Electron Beam Columns Typical Electron Beam Column Zeiss Gemini™ column

Schottky field-emitter tip Courtesy of Leica Lithography Systems Ltd. (courtesy of Leo)

Electron Sources Yes ~ 1% / h 10 – 50 ~ 4000 2000 h (+) << 20nm Suitable for EBL Emission current drift Maximum probe current / nA Filament cost /€ excl labor Filament life time Spot diameter @ Iprobe @ Uacc Litho Resolution in PMMA resist Emitter Yes ~ 1% / h 1000 50 1000 h 100 h ~ 4 nm 20pA/ 30keV ~ 5 nm 5 pA / 30keV < 30 nm ~ 30 nm LaB6 Tungsten restricted ~20-30 % / h < 1 ? 2000 h (+) ~ 2 nm >20pA/20keV << 20 nm Cold FE Yes ~ 1% / h 10 – 50 ~ 4000 2000 h (+) ~ 2 nm >100pA/20keV << 20nm ThermalFE Current density (A/cm²) 25 160 > 650 > 3200 Focussing & Stigmation + ++ ++++

Comparison of stage specifications

Beam deflection (E-static/magnetic) Either magnetic or electrostatic fields can be used to focus electrons just as glass lenses are used to focus rays of light. Electro-static Electro-magnetic F = q · (E + v  B) EM-lenses more simply Fast deflection

Beam deflection (E-static/magnetic) Electron lenses have very poor performance (spherical and chromatic aberrations) compared to light lenses; thus electrons must be kept very close to the axis - small  Spherical aberration Chromatic aberration

Beam deflection E-static or B? Magnetic deflection - lower aberrations (as with electrostatic lenses (= condensor in gun)) - but max. frequency is limited by the coil resonances to ~10 MHz. - max. angle limited by off-axis aberrations to about 1 degree - Trade off between resolution and field size Increasing working distance: deflection angle and thus off-axis aberrations decrease on-axis aberrations increase   Compromise field size is much smaller than most chips!

Beam-blanking Beam-blanker off Beam-blanker on Filament Anode +250V GND Beam-blanker Aperture

to be good in SEM imaging! Motivation To be good in EBL means to be good in SEM imaging!

Focus/Stigmator correction

Focussing - contamination dots top view side view (Images taken during the acceptance test of Raith 200 at the University of Neuchâtel (Switzerland) 4/98)

Focussing = WD and stigmation (Image taken at KTH Stockholm (Sweden), see http://www.nanophys.kth.se/nanophys/facilities/nfl/sem-ebeam.htm)

Influence of acceleration voltage Low (short penetration depth) High (large penetration depth) + Clear surface structures + Less damage + Less charge up + Less edge effect – Lower resolution + Higher resolution – Unclear surface structures – More edge effects – More sample damage (heating) (A guide to Scanning Microscope Observation, Jeol web page 1999)

Influence of beam current - aperture Low (small aperture) High (large aperture) + Higher resolution + Less damage (heating) + Larger depth of focus – Grany image + Smooth image + Good Signal to noise – Deteriorated resolution – More damage (heating) – Lower resolution – Smaller depth of focus (A guide to Scanning Microscope Observation, Jeol web page 1999)

Influence of working distance Small Large + Higher resolution – Smaller depth of focus + Larger depth of focus – Lower resolution (A guide to Scanning Microscope Observation, Jeol web page 1999)

Important rule for SEM imaging Take always the low magnification images first!!!

E-Beam Lithography Methods E-beam strategies Motivation Stage movement strategies EBL writing strategies

Motivation Applications of EBL mask fabrication (e.g. chromium on glass) direct write (rapid prototyping) nano devices in R&D …. different writing strategies required Recommended Literature: SPIE HANDBOOK OF MICROLITHOGRAPHY, MICROMACHINING AND MICROFABRICATION Volume 1: Microlithography, Chapter 2.1

Stage Movement Strategies stationary stage moving stage versus stripes fields "write-on-the-fly“

EBL Writing Strategies round (Gaussian) beam shaped beam versus vector scan raster scan versus

EBL Writing Strategies strategy beam scan mode stage 1 (Raith) gaussian vector fixed 2 (Etec) raster moving 3 (Leica) shaped

gaussian beam, vector scan, fixed stage 1st Strategy (Raith) gaussian beam, vector scan, fixed stage chip write field wafer shapes beam path

gaussian beam, vector scan, fixed stage 1st Strategy (Raith) gaussian beam, vector scan, fixed stage meander mode line mode

gaussian beam, vector scan, fixed stage 1st Strategy (Raith) gaussian beam, vector scan, fixed stage + fast writing of sparse patterns (unwritten areas are skipped) + easy dose variation from shape to shape – settling time and hysteresis have to be calibrated – overhead time caused by increased stage settling time Applications: nano lithography, R&D, …

gaussian beam, raster scan, moving stage 2nd Strategy (Etec) gaussian beam, raster scan, moving stage stage motion beam motion (e.g. used by MEBES (Etec Systems Inc.))

gaussian beam, raster scan, moving stage 2nd Strategy (Etec) gaussian beam, raster scan, moving stage + very simple + very repeatable calibration possible – sparse patterns take as long as dense patterns – difficult to adjust dose during writing Applications: mask making, R&D, …

shaped beam, moving stage 3rd Strategy (Leica) shaped beam, moving stage electron beam first shaping aperture second shaping aperture beam deflectors two fields wafer

3rd Strategy (Leica) shaped beam, moving stage +  10 x faster than equivalent gaussian beam machines – extremely complex electron optical column – complicated calibration routines – resolution and focus varies with shape size Applications: mask making, advanced chip development - Similar to Gaussian vector scan but exposes an entire rectangle (up to typically 2µm x 2µm) in a single "flash" - Some machines combine vector scan with "write-on-the-fly“ Extension: Cell projection (one of the square apertures is replaced by a more complex shape such as a DRAM cell.

Summary: 3 strategies strategy beam scan mode stage 1 (Raith) gaussian vector fixed 2 (Etec) raster moving 3 (Leica) shaped

Summary: Strategy used by Raith vector scan meander mode stationary stage fields vector scan line mode gaussian beam Applications: Nano device fabrication, R&D

Basic resist theory Motivation transparency: Dose test in PMMA

Contents Electron scattering in resist and substrate Proximity effect Resist interactions (positive /negative/chemically amplified resists, resist contrast) Dose definition Influence of beam energy (penetration depth) Resolution limits

Forward scattering Forward scattering events Properties very often scattering under small angles small-angle hence very inelastic (color symbolizes kinetic energy) generation of secondary electrons with a few eV kinetic Energy

Backscattering Backscattered electrons Properties occasionally scattering under large angles large angle hence mainly elastic (color symbolizes kinetic energy) high kinetic energy, range of the primary electrons

What leads to an exposure? Secondary electrons with few eV kinetic energy are responsible for most of the resist exposure Hence forward scattering within the resist is responsible for exposure And backscattering is responsible for exposure apart from incidence

Scattering range versus energy

Proximity effect Test pattern in resist purposely overexposed to enhance the proximity effect

Proximity effect - simulation Simulation of electron trajectories: 1.5 µm resist thickness on a silicon wafer 50 trajectories, at 25 keV beam energy. (Mark A. McCord, Introduction to Electron-Beam Lithography, Short Course Notes Microlithography 1999, SPIE's International Symposium on Microlithography 14-19 March, 1999; p. 22)

Proximity effect – energy dependence (D. F. Kyser and N. S. Viswanathan, "Monte Carlo simulation of spatially distributed beams in electron-beam lithography", J. Vac. Sci. Technol. 12(6), 1305-1308 (1975))

Resist interactions Positive resists usually work by polymer chain scission; exposed polymer becomes more soluble Negative resists usually work by cross linking between polymer chains; exposed polymer becomes insoluble

Induced chain scission (PMMA)

Chemically amplified resist Chemically amplified resists, are modified during exposure. However the actual exposure takes place during the post exposure bake, when the acids are activated.

Resist contrast = Slope in resist Positive Negative remaining thickness remaining thickness log(Dose) log(Dose) - D0 D1 D0 D1 Contrast  = [log10(D1)-log10(D0)] -1 (Mark A. McCord, Introduction to Electron-Beam Lithography, Short Course Notes Microlithography 1999, SPIE's International Symposium on Microlithography 14-19 March, 1999; p. 22)

Resist contrast High contrast: Low contrast: + Steeper side walls + Greater process latitude + Better resolution (not always) + Less sensitivity to proximity effects Low contrast: + 3d lithography

Calculation of dose Ibeam = beam current Tdwell = dwell time s = step size [µAs/cm²]

Dose table for PMMA (950k) 10 kV 20 kV 30 kV Areas 100 µC/cm² 200 µC/cm² 300 µC/cm² SPLs 300 pC/cm 600 pC/cm 900 pC/cm Dots 0.1 pC 0.2 pC 0.3 pC (developer: MIBK + IPA, 1:3) Doses are values which resist manufactures will give you such that the resist is "cleared or fully developed" for a certain dose of current. With the measured beam current, pixel size, and dose value, the dwell time per pixel can be calculated. If the doses are wrong, then the resist can be over or under exposed. The sensitivity of the resist increases (dose value decreases) as you go down in kV. This means you can do faster exposes at lower kV - but may be with loss in terms of resolution. The above values are good starting points. The best way to get optimum results is to perform a dose scaling: SPLs 0.5 – 5, Dots 0.1 – 10

Dose versus voltage Increase of dose with voltage for all resists Graph gives general behavior

Design must be adapted to dose Line width in design Dependence of line width versus dose for different line width in design Johannes Kretz, Infineon, Munich

Influence of beam energy 100 keV + Small scattering in resist + Small proximity effect – High beam damage – Strong sample heating 20 keV + Small beam damage + Small sample heating + Best electron-optical performance (classical columns) – Scattering in thick resist – Strong proximity effect 2 keV + No beam damage + No proximity effect + High throughput (high resist sensitivity) – High scattering in resist – Needs very thin resists

Penetration depth versus energy At small kVs you should keep an eye on the penetration depth Y. Lee, W. Lee, and K. Chun 1998/9, A new 3 D simulator for low energy (~1keV) Electron-Beam Systems

Resolution limits Beam resolution Resist limits Thick resists (forward scattering) Thin resists (~0.5nm by diffraction, de Brogli wavelength) Resist limits Polymer size (~5-10nm) Chemically amplified resists (acid diffusion ~50nm) (Mark A. McCord, Introduction to Electron-Beam Lithography, Short Course Notes Microlithography 1999, SPIE's International Symposium on Microlithography 14-19 March, 1999; p.63)

Risoluzione dg : riduzione della sorgente dg=dv/M-1 ds : aberrazioni sferiche ds=1/2 Cs a3 dc : aberrazioni cromatiche dc= Cc aDV/Vbeam ddiff : diffrazione ddiff=1.2l/a Il limite di diffrazione non è così importante (elettroni accelerati a 15 keV è l=0.1Å. se a è 10 mrad): ddiff=1.2l/a = 1 nm La somma in quadratura di questi contributi dà una dimensione dello spot nei più moderni sistemi di ~2nm.

Resolution limits Secondary electron range (~5-10nm) In practice, the best achievable resolution in polymer resists is about 20nm, with inorganic resists (currently impractical for most applications) 5nm. (Mark A. McCord, Introduction to Electron-Beam Lithography, Short Course Notes Microlithography 1999, SPIE's International Symposium on Microlithography 14-19 March, 1999; p.63)

Ultra high resolution in PMMA (45nm thickness): What is possible ? Ultra high resolution in PMMA (45nm thickness): 16nm line width in resist

Process Technology Pattern Transfer substrate resist Spin coating Exposure Developing Wafer Coating or stripping step After x process steps Remover Lift-Off Etching metal Pattern Transfer

Process Technology Pattern Transfer substrate resist Spin coating Exposure Developing Raith support Special Knowledge, e.g. EBL resist database Remover Lift-Off Etching metal Pattern Transfer General Knowledge

Raith Application Lab: Typical Instrumentation substrate Spin coating Exposure Developing Cleaning Instrumentation Film thickness measurement Cleaning Spin coating Exposure Developing Inspection wet bench (eye-) shower for accidents with acids storage for chemicals stove / hotplate refrigerator spin coater Film Thickness Probe Raith 150 / R50 FE optical microscope sputtering machine Process step Spin coating Film thickness measurement Exposure Developing Inspection

Lift-Off Tips & Tricks obtain an undercut resist profile by using a double layer resist using low beam energy overexposure or overdeveloping use an aspect ratio of resist: metal as large as possible if possible use evaporation, not sputtering

Etching Tips & Tricks Resist Wafer Wafer obtain cross-sections without undercut or overcut by using high beam energy avoiding overexposure and overdeveloping Resist apply postbake to improve resist stability during etching for organic resists avoid etching processes with oxygen Wafer

Process Steps substrate Spin coating Wafer resist substrate Exposure inspection e.g. resist thickness measurement e.g. electron beam lithography substrate Developing substrate inspection optical microscope SEM metrology

How to avoid charging effects On isolating samples an additional metal layer between resist an substrate or on top of the resist is required, e.g. use additional Al layer on top of the resist. thickness should be 10 nm to 20 nm remove with KOH With an additional isolating interlayer, e.g. SiO layer use a higher acceleration voltage, e.g. 25 kV for 1 µm SiO.

Resist characterization I: Dependence of thickness on spin speed open cover closed cover

Resist characterization II: Contrast curves

Positive EBL Resists application dose resolution comments µC/cm2 @30keV nm High resolution: PMMA ~200 <10 dry etch (--), low sensitivity or low contrast ZEP xx ~ 50 <20 dry etch(3x better than PMMA), adhesion (-) Mask Making: EBR-9 ~20 >200 long shelf life, large process latitude PBS ~2 >250 low contrast, difficult process (--) AZ5206 ~20 ~250 high contrast, good etch resistance Double Layer with PMMA (e.g. Lift-Off): P(MMA-MAA) ~ 70 <200 PMGI ~ 40 <100 more info at: http://cnf.cornell.edu/ebeam/resist.html www.microresist.de http://snf.stanford.edu/process/lithography/ebeamres.html www.allresist.de www.microchem.com

Negative EBL Resists Application dose resolution comments µC/cm2 @30keV nm High resolution Calixarene ~4000 <10 very low sensitivity, high resolution (+) HSQ ~300 <20 dry etch(+) ma-N2400 ~100 <70 DUV & E-beam sensitive NEB xx ~ 20 <40 dry etch(+), adhesion (-) SAL ~10 <100 high contrast, dry etch (+) SU8 ~2 <100 3D EBL applications COP ~1 >1000 difficult process (--) Mask Making more info at: http://cnf.cornell.edu/ebeam/resist.html www.microresist.de http://snf.stanford.edu/process/lithography/ebeamres.html www.allresist.de www.microchem.com

Which resist for which application? decide about positive or negative resist with respect to the minimum area that will be exposed check literature and resist suppliers for resist performance with respect to e.g. resolution sensitivity etching stability check suitability for your lab, e.g. required baking steps and chemicals

EBL Applications

PMMA 950K for High Resolution resist thickness: 45 nm closed cover developing 15 s without moving sample linewidth < 20 nm reproducibel! HEMT 16 nm line width 100 nm gate 50 nm lines

Calixaren for High Resolution 18 nm linewidth

Resists - 3Dimensional Lithography Gradation Curve Contrast Conventional lithography Third Dimension! Positive resist Negative resist Dirk Brüggemann: MEMS application DOEs Microlenses Phase holograms Optical mirrors

SU8 and PMMA50K for 3D EBL SU8 PMMA50K Fresnel lens Deutsches Museum München

Long Lines in PMMA line width = 20 nm pitch = 50 nm length = 10 mm stitching error < 20 nm 50 nm PMMA 950K

Large Gratings → exposure time < 16 h line width = 40 nm pitch = 100 nm size = 5 x 5 mm2 write field = 100 µm → 2500 WF Optimizing exposure time small aperture (depth of focus!) but: fast resist (e.g. ZEP) Layout: SPLs in addition: SPLs in „area mode“ exposed → exposure time < 16 h

Exposure of Circles Circular mode (V. 3.0) Circle arrays Fresnel lens / Zone plate

Proximity Effect Each pattern element needs a different dose - here shown by colors -

Large Dot Arrays Application: 2 x 2 mm2 array with 200 nm wide circles and 250 nm pitch Idea 1: Exposure of single dots (1x settling time, 1x transmission time per circle) Problem: Upper limit of diameter, because positive resists behave like negative when overexposed by a factor of about 1000 Solution: Exposure with defocussed beam resist residues Defokussiert

Large Dot Arrays Idea 2: Problem Settling Time: 64,000,000 cirles x 5 ms  90 h Solution: 0,5 ms settling time and layout in meander form Set BeamParkPosition on first element  Overall Settling Time < 9 h Meander shaped sorting → constant distances for the beam movement