Ion Beam Milling System

1. Theory

• Definition: An ion beam mill etches by sputtering with a broad, neutralized argon ion beam from a gridded (Kaufman) source, with no chemical reaction (Stanford; NASA; MIT).
• Rate: Set by sputter yield, current density and atomic volume, $z/t=MSj/\rho N_{A}e$; at 500 eV and 1 mA/cm${}^2$, gold mills at 1553 angstrom/min and silicon at 310 (SPECS).
• Extraction: Beam current rises as the 3/2 power of extraction voltage (Child’s law) (NASA).
• Neutralization: Without electrons a 500 eV, 1 mA/cm${}^2$ beam would reach more than 14 000 V of space-charge potential (KRI).
• Character: Anisotropic and universal, but with low selectivity, charging, trenching and redeposition (MIT).

2. Working Principle

• Source: Electron-bombardment discharge with multipole magnets, 200 to 1000 eV, 1 to 2 mA/cm${}^2$ (NASA).
• Neutralizer: Filament, plasma bridge or hollow cathode emitting about the beam current (KRI).
• Stage: Water-cooled, rotating and tilting, up to 100 mm (Stanford).
• Pressure: At or below 1 mTorr, where recombination is negligible (KRI).

3. Applications

• Metals and compound semiconductors: Patterning materials without volatile etch products, such as HgTe Hall bars (Stanford).
• Magnetic and noble-metal stacks: Any material sputters, as the yield table shows (SPECS; my reading).
• Low-damage thinning: GaAs with damage below about 100 angstrom (NASA).

Source and Milling Rate

Name	Formula	Related Formulae	Variables/Constants
Gridded ion source	$j=1\text{to}2mA/c{\mathrm{m}}^2,E=200\text{to}1000\mathrm{eV}$	$I_{beam}\propto V_{ext}^{3/2}$	$\begin{array}{r}NASA:30cmsourcefor1mA/c{\mathrm{m}}^{2}at500\mathrm{eV},\\ plusorminus5percentover20cm,10cm\\ sourceto2mA/c{\mathrm{m}}^2;Chil{d}^{\prime }slawfor\\ extraction;Stanford:energyandcurrent\\ setindependently\end{array}$ $\begin{array}{r}(computed:doublingtheextractionvoltage\\ raisesthebeamcurrent2.8times)\end{array}$
Magnetic confinement	$\int Bdl\approx 48\mathrm{G}\mathrm{cm}(50\mathrm{eV}\text{electrons})$	$\text{design margin}\approx 50\%$	$\begin{array}{r}NASA:multipolepermanent-magnetdischarge\\ chamber\end{array}$ $\begin{array}{r}(computed:thedesignintegralisabout72\\ \mathrm{G}\mathrm{cm})\end{array}$
Milling rate	$\frac{z}{t}=\frac{MSj}{\rho N_{A}e}$	$r_{500eV,1mA/c{m}^2}:Au1553,Cu818,Al630,Si{O}_{2}400,Si310\mathring{\mathrm{A}}/min$	$\begin{array}{r}SPECS:Ti336,W340\mathring{\mathrm{A}}/min;DCandRF\\ MagnetronSputterSystempageforyields\end{array}$ $\begin{array}{r}(computed:AutoSirateratio5.0,Auto\\ W4.6)\end{array}$

Neutralization and Energy Limits

Name	Formula	Related Formulae	Variables/Constants
Space charge without neutralizer	$\Delta V>14000\mathrm{V}$	$500\mathrm{eV},10\mathrm{cm},1mA/c{\mathrm{m}}^2$	$\begin{array}{r}KRI:unrealisticbeampotential,arcsfrom\\ about200V;NASA:neutralizersetuntil\\ nonetcurrenttoaFaradayprobe\end{array}$ $\begin{array}{r}(myreading:thebeamwouldblowitself\\ apartbeforereachingthewafer)\end{array}$
Neutralizer settings	$I_e=I_{beam}\text{(current neutralization)}$	$\pm 10\%\text{(grounded)},0\text{to}+5\%\text{(charge-sensitive)},\pm 5\mathrm{V}$	$\begin{array}{r}KRI:filamentinthebeam,plasmabridge\\ andhollowcathodelesslocation\\ sensitive,isolatedtargets0to10per\\ centexcess,dielectriccoatingswithin\\ plusorminus5percent\end{array}$ $\begin{array}{r}(myreading:devicewafersare\\ charge-sensitivetargets)\end{array}$
Low-energy limit	$E<200\mathrm{eV}\Rightarrow \text{beam divergence}$	$p\le 1\mathrm{mTorr}:\text{recombination negligible}$	$\begin{array}{r}KRI;NASA:200to1000\mathrm{eV}(appendixB)and\\ 200to800\mathrm{eV}(appendixD)\end{array}$ $\begin{array}{r}(myreading:Chil{d}^{\prime }slawmakeslow-energy\\ beamsweak,sotheyspread)\end{array}$

Profile and Damage

Name	Formula	Related Formulae	Variables/Constants
Physical etching	$\text{argon only, chemically inert}$	$\text{anisotropic, low selectivity}$	$\begin{array}{r}MIT:charging,trenchingandredeposition;\\ Stanford:argononly;Parallel-Plate\\ ReactiveIonEtcher(RIE)pageforthe\\ chemicalalternative\end{array}$ $\begin{array}{r}(myreading:ionmillingetchesanything,\\ includingmetalswithnovolatile\\ products)\end{array}$
Tilt and rotation	$\text{tilt sets sidewall angle},\text{removes fences}$	$\text{rotation for uniformity, cooled stage}$	$\begin{array}{r}Stanford:water-cooledrotatingtiltable\\ stage,drychuck,ratedependenton\\ surfacetemperature,HgTeHallbarswitha\\ Timask\end{array}$ $\begin{array}{r}(myreading:redepositedmetalon\\ sidewallsismilledoffatgrazingangles)\end{array}$
Damage depth	$d_{damage}<80\text{to}100\text{angstrom}(GaAs)$	$4000\text{angstrom/h}\text{(ion-beam sputter deposition)}$	$\begin{array}{r}NASA:GaAsthinningandsputterdeposition\\ at500\mathrm{eV}and1.5mA/c{\mathrm{m}}^2\end{array}$ $(computed:about67angstromperminute)$

Important Questions

Ques: How does an ion beam mill remove material?

By pure sputtering with a broad argon beam. Stanford describes argon ions accelerated out of a Kaufman source and mostly neutralized by a filament, and stresses that the tool uses only chemically inert argon. MIT classes ion milling as nearly purely physical, anisotropic and of low selectivity, with charging, trenching and redeposition as its problems. NASA’s report describes the source: an electron-bombardment discharge with a multipole permanent-magnet field, designed for 1 $mA/c{\mathrm{m}}^2$ of 500 $\mathrm{eV}$ argon with plus or minus 5 per cent uniformity over the central 20 $\mathrm{cm}$. Because nothing needs to be volatile, any material can be milled (my reading), unlike the chemical etch of the Parallel-Plate Reactive Ion Etcher (RIE) page. Reference: Stanford Nano Shared Facilities, ‘Ion mill’ page: argon ions from a Kaufman source neutralized by a filament, argon only and chemically inert, ion energy and current density independently adjustable, water-cooled rotating tiltable stage for up to 100 mm substrates with a silver-impregnated silicone dry chuck, sputter rate dependent on surface temperature, tilt to set sidewall angle and remove redeposited metal fences, 1 um HgTe quantum-well Hall bars with a Ti mask (fetch-tool summary); MIT 6.152J lecture 17 ‘Etching’ (2005): sputter etching and ion milling nearly purely physical, no reactive gas, anisotropic with low selectivity, problems of charging, trenching and redeposition (fetch-tool summary); H. R. Kaufman, R. S. Robinson and J. R. Sites (Colorado State University), ‘Industrial ion source technology’, NASA CR-135149, annual report (November 1976): 30 cm electron-bombardment source with multipole permanent-magnet field designed for 1 mA/cm2 of 500 eV argon with plus or minus 5 per cent uniformity over the central 20 cm, 10 cm Ion-Tech source at 200 to 1000 eV and up to 2 mA/cm2 (appendix B) or 200 to 800 eV (appendix D), single-strand neutralizer cathode about 5 cm from the accelerator adjusted until no net current reached a grounded Faraday probe, Child’s law with beam current as the 3/2 power of extraction voltage, about 48 G cm fringe-field integral for 50 eV electrons with about 50 per cent margin used, ion-beam sputter deposition of 4000 angstrom in about one hour at 500 eV and 1.5 mA/cm2, GaAs thinning damage below about 80 to 100 angstrom (fetch-tool summary); Parallel-Plate Reactive Ion Etcher (RIE) page of this vault.

Ques: How fast do different materials mill, and how selective is it?

Rates follow the sputter yield, so selectivity is low. SPECS gives $z/t=MSj/(\rho N_{A}e)$ and measured rates for 500 $\mathrm{eV}$ argon at 1 $mA/c{\mathrm{m}}^2$: gold 1553, copper 818, aluminium 630, SiO${}_2$ 400, tungsten 340, titanium 336 and silicon 310 $\mathring{\mathrm{A}}/min$. Gold therefore mills 5.0 times faster than silicon and 4.6 times faster than tungsten (computed), ratios far below those of chemical etches. MIT notes that low selectivity is inherent to physical etching. Stanford used a titanium mask for HgTe Hall bars, consistent with titanium’s low rate (my reading). Yields are discussed further on the DC and RF Magnetron Sputter System page, and endpoint is therefore usually set by time or by detecting the underlying layer rather than by selectivity (my reading). Reference: SPECS GmbH, ‘Sputter info’ technical note: rate z/t = M S j_p/(rho N_A e), Ar+ yields at 0.5 and 1 keV, measured rates for 500 eV Ar+ at 1 mA/cm2 of Al 630, Si 310, Ti 336, W 340, Cu 818, Au 1553 and SiO2 400 angstrom/min (fetch-tool summary); MIT 6.152J lecture 17 ‘Etching’ (2005): sputter etching and ion milling nearly purely physical, no reactive gas, anisotropic with low selectivity, problems of charging, trenching and redeposition (fetch-tool summary); Stanford Nano Shared Facilities, ‘Ion mill’ page: argon ions from a Kaufman source neutralized by a filament, argon only and chemically inert, ion energy and current density independently adjustable, water-cooled rotating tiltable stage for up to 100 mm substrates with a silver-impregnated silicone dry chuck, sputter rate dependent on surface temperature, tilt to set sidewall angle and remove redeposited metal fences, 1 um HgTe quantum-well Hall bars with a Ti mask (fetch-tool summary); DC and RF Magnetron Sputter System page of this vault.

Ques: Why must the ion beam be neutralized, and how much electron current is needed?

Because an unneutralized beam would charge to absurd potentials. KRI calculates that a 500 $\mathrm{eV}$, 10 $\mathrm{cm}$ beam at 1 $mA/c{\mathrm{m}}^2$ would develop more than 14 000 $\mathrm{V}$ between its axis and edge, and that electron shortfalls raise the beam potential until arcs strike, typically from about 200 $\mathrm{V}$. Neutralization means electron emission equal to beam current. KRI recommends staying within plus or minus 10 per cent for grounded targets and 0 to 10 per cent excess for isolated ones. For charge-sensitive targets it recommends 0 to 5 per cent excess, keeping the target within plus or minus 5 $\mathrm{V}$. NASA set its filament until no net current reached a grounded Faraday probe. Device wafers fall in the charge-sensitive class (my reading). Reference: Kaufman and Robinson, Inc., Technical Note KRI TN-02 ‘Ion-beam neutralization’ (2003, 2006): broad beams must be neutralized, an unneutralized 500 eV, 10 cm, 1 mA/cm2 beam would have over 14 000 V between axis and edge, too few electrons raising beam potential and causing arcs, hot-filament neutralizers needing to sit in the beam while plasma-bridge and hollow-cathode types are less location sensitive, current neutralization when emission equals beam current, grounded targets within plus or minus 10 per cent, isolated targets 0 to 10 per cent excess, charge-sensitive targets 0 to 5 per cent excess with potential within plus or minus 5 V, dielectric deposition within plus or minus 5 per cent, recombination insignificant at or below 1 mTorr, arcs starting near 200 V, beams diverging below about 200 eV (fetch-tool summary); H. R. Kaufman, R. S. Robinson and J. R. Sites (Colorado State University), ‘Industrial ion source technology’, NASA CR-135149, annual report (November 1976): 30 cm electron-bombardment source with multipole permanent-magnet field designed for 1 mA/cm2 of 500 eV argon with plus or minus 5 per cent uniformity over the central 20 cm, 10 cm Ion-Tech source at 200 to 1000 eV and up to 2 mA/cm2 (appendix B) or 200 to 800 eV (appendix D), single-strand neutralizer cathode about 5 cm from the accelerator adjusted until no net current reached a grounded Faraday probe, Child’s law with beam current as the 3/2 power of extraction voltage, about 48 G cm fringe-field integral for 50 eV electrons with about 50 per cent margin used, ion-beam sputter deposition of 4000 angstrom in about one hour at 500 eV and 1.5 mA/cm2, GaAs thinning damage below about 80 to 100 angstrom (fetch-tool summary).

Ques: What beam energy range is usable, and which figure is more likely?

About 200 to 1000 $\mathrm{eV}$. NASA’s report lists the 10 $\mathrm{cm}$ Ion-Tech source as 200 to 1000 $\mathrm{eV}$ with up to 2 $mA/c{\mathrm{m}}^2$ in one appendix, but as 200 to 800 $\mathrm{eV}$ in another. KRI explains the lower limit: below about 200 $\mathrm{eV}$ the beam tends to diverge or the source stops. Child’s law, also cited by NASA, makes beam current rise as $V^{3/2}$, so doubling the voltage gives 2.8 times the current (computed). The 200 to 1000 $\mathrm{eV}$ range is more likely the source’s capability, while 200 to 800 $\mathrm{eV}$ is probably the range used in that particular work (my reading). The 500 $\mathrm{eV}$ design point lies inside both. Reference: H. R. Kaufman, R. S. Robinson and J. R. Sites (Colorado State University), ‘Industrial ion source technology’, NASA CR-135149, annual report (November 1976): 30 cm electron-bombardment source with multipole permanent-magnet field designed for 1 mA/cm2 of 500 eV argon with plus or minus 5 per cent uniformity over the central 20 cm, 10 cm Ion-Tech source at 200 to 1000 eV and up to 2 mA/cm2 (appendix B) or 200 to 800 eV (appendix D), single-strand neutralizer cathode about 5 cm from the accelerator adjusted until no net current reached a grounded Faraday probe, Child’s law with beam current as the 3/2 power of extraction voltage, about 48 G cm fringe-field integral for 50 eV electrons with about 50 per cent margin used, ion-beam sputter deposition of 4000 angstrom in about one hour at 500 eV and 1.5 mA/cm2, GaAs thinning damage below about 80 to 100 angstrom (fetch-tool summary); Kaufman and Robinson, Inc., Technical Note KRI TN-02 ‘Ion-beam neutralization’ (2003, 2006): broad beams must be neutralized, an unneutralized 500 eV, 10 cm, 1 mA/cm2 beam would have over 14 000 V between axis and edge, too few electrons raising beam potential and causing arcs, hot-filament neutralizers needing to sit in the beam while plasma-bridge and hollow-cathode types are less location sensitive, current neutralization when emission equals beam current, grounded targets within plus or minus 10 per cent, isolated targets 0 to 10 per cent excess, charge-sensitive targets 0 to 5 per cent excess with potential within plus or minus 5 V, dielectric deposition within plus or minus 5 per cent, recombination insignificant at or below 1 mTorr, arcs starting near 200 V, beams diverging below about 200 eV (fetch-tool summary).

Ques: How are sidewall angle, redeposition and heating controlled?

By stage tilt, rotation and cooling. Stanford’s mill has a water-cooled, rotating, tiltable stage for substrates up to 100 $\mathrm{mm}$. Tilt sets the sidewall angle and removes redeposited metal fences, and rotation keeps the etch uniform. A silver-impregnated silicone dry chuck conducts heat away because sputter rate depends on surface temperature. Small chips are mounted on a carrier wafer with silver paste. MIT lists redeposition and trenching among the problems of physical etching. In my reading, material sputtered from the floor lands on the sidewalls and forms fences, so a final step at a grazing angle cleans them. KRI notes that recombination is negligible at or below 1 $\mathrm{mTorr}$, which keeps the beam intact on its way to the stage. Reference: Stanford Nano Shared Facilities, ‘Ion mill’ page: argon ions from a Kaufman source neutralized by a filament, argon only and chemically inert, ion energy and current density independently adjustable, water-cooled rotating tiltable stage for up to 100 mm substrates with a silver-impregnated silicone dry chuck, sputter rate dependent on surface temperature, tilt to set sidewall angle and remove redeposited metal fences, 1 um HgTe quantum-well Hall bars with a Ti mask (fetch-tool summary); MIT 6.152J lecture 17 ‘Etching’ (2005): sputter etching and ion milling nearly purely physical, no reactive gas, anisotropic with low selectivity, problems of charging, trenching and redeposition (fetch-tool summary); Kaufman and Robinson, Inc., Technical Note KRI TN-02 ‘Ion-beam neutralization’ (2003, 2006): broad beams must be neutralized, an unneutralized 500 eV, 10 cm, 1 mA/cm2 beam would have over 14 000 V between axis and edge, too few electrons raising beam potential and causing arcs, hot-filament neutralizers needing to sit in the beam while plasma-bridge and hollow-cathode types are less location sensitive, current neutralization when emission equals beam current, grounded targets within plus or minus 10 per cent, isolated targets 0 to 10 per cent excess, charge-sensitive targets 0 to 5 per cent excess with potential within plus or minus 5 V, dielectric deposition within plus or minus 5 per cent, recombination insignificant at or below 1 mTorr, arcs starting near 200 V, beams diverging below about 200 eV (fetch-tool summary).

Ques: How much damage does ion milling leave?

A thin damaged layer, of order 100 angstrom at 500 $\mathrm{eV}$. NASA reports GaAs thinning with damage below about 80 angstrom, the resolution limit of its measurement, and a surface layer under 100 angstrom. The same report uses the beam for ion-beam sputter deposition, growing 4000 angstrom in about one hour at 500 $\mathrm{eV}$ and 1.5 $mA/c{\mathrm{m}}^2$, about 67 angstrom per minute (computed). MIT notes charging as a further problem, and KRI warns that poor neutralization can arc and deposit particulates on work pieces. In my reading, damage scales with ion energy, so final passes are best run near the 200 $\mathrm{eV}$ lower limit KRI describes. Reference: H. R. Kaufman, R. S. Robinson and J. R. Sites (Colorado State University), ‘Industrial ion source technology’, NASA CR-135149, annual report (November 1976): 30 cm electron-bombardment source with multipole permanent-magnet field designed for 1 mA/cm2 of 500 eV argon with plus or minus 5 per cent uniformity over the central 20 cm, 10 cm Ion-Tech source at 200 to 1000 eV and up to 2 mA/cm2 (appendix B) or 200 to 800 eV (appendix D), single-strand neutralizer cathode about 5 cm from the accelerator adjusted until no net current reached a grounded Faraday probe, Child’s law with beam current as the 3/2 power of extraction voltage, about 48 G cm fringe-field integral for 50 eV electrons with about 50 per cent margin used, ion-beam sputter deposition of 4000 angstrom in about one hour at 500 eV and 1.5 mA/cm2, GaAs thinning damage below about 80 to 100 angstrom (fetch-tool summary); MIT 6.152J lecture 17 ‘Etching’ (2005): sputter etching and ion milling nearly purely physical, no reactive gas, anisotropic with low selectivity, problems of charging, trenching and redeposition (fetch-tool summary); Kaufman and Robinson, Inc., Technical Note KRI TN-02 ‘Ion-beam neutralization’ (2003, 2006): broad beams must be neutralized, an unneutralized 500 eV, 10 cm, 1 mA/cm2 beam would have over 14 000 V between axis and edge, too few electrons raising beam potential and causing arcs, hot-filament neutralizers needing to sit in the beam while plasma-bridge and hollow-cathode types are less location sensitive, current neutralization when emission equals beam current, grounded targets within plus or minus 10 per cent, isolated targets 0 to 10 per cent excess, charge-sensitive targets 0 to 5 per cent excess with potential within plus or minus 5 V, dielectric deposition within plus or minus 5 per cent, recombination insignificant at or below 1 mTorr, arcs starting near 200 V, beams diverging below about 200 eV (fetch-tool summary).

References

1. H. R. Kaufman, R. S. Robinson and J. R. Sites, “Industrial ion source technology,” NASA CR-135149, Colorado State University, 1976. Available: https://ntrs.nasa.gov/api/citations/19770009305/downloads/19770009305.pdf
2. Kaufman and Robinson, Inc., “Ion-beam neutralization,” Technical Note KRI TN-02. Available: https://ionsources.com/wp-content/uploads/2024/01/TN-02_Ion-Beam-Neutralization-Resource.pdf
3. Stanford Nano Shared Facilities, “Ion mill.” Available: https://snsf.stanford.edu/facilities/fab/fcr/ionmill
4. SPECS GmbH, “Sputter info,” technical note. Available: https://www.specs-group.com/fileadmin/user_upload/products/technical-note/sputter-info.pdf
5. MIT 6.152J Micro/Nano Processing Technology, “Lecture 17: etching,” MIT OpenCourseWare, 2005. Available: https://ocw.mit.edu/courses/6-152j-micro-nano-processing-technology-fall-2005/59b9f31d836355a3380aa156925508bc_lecture17.pdf
6. DC and RF Magnetron Sputter System page of this vault.
7. Parallel-Plate Reactive Ion Etcher (RIE) page of this vault.

Related pages in this vault: DC and RF Magnetron Sputter System, Parallel-Plate Reactive Ion Etcher (RIE), Secondary Ion Mass Spectrometer (SIMS and ToF-SIMS), Focused Ion Beam (FIB).