Beam Current Control of Ion Beam System Power Supplies in Focused Ion Beam and Nano Polishing Work
A focused ion beam column accelerates gallium ions to thirty kilovolts and focuses them into a spot a few nanometres across for imaging and milling, while ion beam polishing uses a broad, low-energy beam to remove material from a surface without introducing damage. Both depend on a supply whose beam current is stable to a fraction of a percent, because the milling rate and the polishing depth are integrals of current over time, and an error in current becomes an error in depth that cannot be corrected afterwards.
Ion sources behave differently from electron sources, and the supply must match. A liquid metal ion source ignites at a threshold and then operates in a regime where emission depends steeply on the extraction voltage; a small change in voltage produces a large change in current. A supply in voltage mode therefore produces an unstable current, and the usual design closes the loop on current with the voltage as the actuator.
The loop's stability is limited by the source's own dynamics. Emission responds to the extraction voltage with a delay of microseconds to milliseconds, depending on the source's thermal state, and a current loop tuned for a fast source will oscillate with a slow one. Setting the loop bandwidth below the source's delay, and adding a slow integrator for the drift, produces a controller that works across source types without retuning.
Extraction voltage and beam energy are separate quantities that share one supply in simple designs. Splitting them, with a stable energy supply and a separate extraction supply referenced to it, allows the current to be controlled without changing the landing energy. This separation is what makes depth control possible in polishing, where the energy determines the damage layer and the current determines the rate.
Beam blanking and the supply's transient response interact. Blanking deflects the beam away from the work piece in a few nanoseconds, and during the transition the beam passes through intermediate positions that can mill unwanted features. A supply that holds current constant during blanking, with a fast current loop, keeps the unwanted milling to a minimum; a supply that lets current surge during the transition produces a visible artifact.
Current measurement at the low end is the weak point. A beam current of one picoamp corresponds to a few thousand ions per second, and measuring that with a sense resistor is impossible; the measurement is done with a Faraday cup and an electrometer, and the supply's loop uses a lower-resolution signal plus calibration. Separating the fast control loop from the accurate measurement, and periodic calibration between them, is the practical approach.
Detector and column supplies share a ground reference with the beam supply, and the shared reference determines the landing position. A beam supply whose return is not at the same potential as the column's apertures produces a shift that varies with current, and the shift appears as a focus error. Star-point grounding, with a single reference for the column, is a requirement rather than a refinement.
Vacuum quality determines both the beam's transmission and the supply's stability. At poor vacuum the beam is scattered and the current at the work piece falls; the supply raises the extraction voltage to compensate, which changes the source's emission characteristics and can shorten its life. A vacuum interlock that reduces the beam before the compensation runs away protects both.
Polishing applications run for hours at low current, and drift over hours matters more than transient response. Drift is dominated by the source's ageing and by the supply's divider and reference. Separating the two contributions requires logging the voltage required to hold current: a rising voltage with constant current indicates source ageing, while a rising current at constant voltage indicates a supply drift.
Thermal management of the source and the supply are coupled through the extraction electrode. The electrode heats from beam interception, and its temperature changes the field distribution, which changes the emission. A supply with a slow current loop absorbs the change; a supply with a fast loop may amplify it into an oscillation. Matching the loop time constant to the electrode's thermal time constant is part of commissioning.
Charging of insulating samples is a process problem that the supply can mitigate. A sample that charges positive repels ions, and the beam deflects; a supply with a programmable beam energy can be operated at a voltage where the charge balance is neutral. Sweeping the energy to find that point is a standard procedure, and it requires a supply with fine energy resolution and repeatable settings.
Maintenance intervals follow the source's life and the supply's drift. Source replacement is triggered by the extraction voltage trend; supply recalibration is triggered by the annual comparison with an external divider. Recording both trends in the same log makes it possible to attribute a change in beam behaviour to the right component.
Electromagnetic compatibility is less demanding than in welding but still relevant because of the sensitive detectors nearby. Fibre links for control signals, shielded cables with a single reference, and filtering on the supplies' inputs keep the secondary electron detector's baseline quiet. A noisy supply shows up as noise in the image before it shows up in the current reading.
Cost trade-offs concentrate on the measurement chain and the loop flexibility. A supply with a current loop that adapts to the source's delay, an independent energy supply, and a calibrated measurement path costs more than a fixed-gain unit, and the difference is justified by the ability to change sources without requalifying the process.
Limits should be stated. The beam's spot size is set by the column's optics, not by the supply; the milling rate is set by the current, and the achievable current is limited by the source. A supply can provide a stable current within the source's range, and cannot extend that range without changing the source.
An acceptance test that fits the application uses a Faraday cup and a known sample. Measure current at three settings against an electrometer, then mill a test structure at a fixed dose and measure its depth by profilometry. The first test checks the supply and the measurement chain, the second checks the dose integral, and together they show whether the depth control is as good as the specification claims.
Source ignition is a transient that the supply must survive. A liquid metal source ignites at a voltage slightly above its operating point, drawing a current surge for a few milliseconds, and a supply with a slow current limit lets that surge stress the source's emitter. A current limit that acts within a hundred microseconds, set just above the operating current, shortens the ignition and extends the emitter's life.
Image quality is the process-level indicator of supply stability. A beam current that varies by a fraction of a percent produces brightness variation in the image that is visible as a slow flicker, and a supply with a current loop that interacts with the column's deflection can produce a periodic pattern. When image quality complaints arrive, comparing the current log with the image artifacts is faster than any bench test.
Records of extraction voltage, emission current and vacuum per session support source life prediction. A trend of rising extraction voltage at constant current points to emitter wear, and the point at which the voltage reaches the supply's limit is the practical end of the source's life. Predicting that point allows replacement to be scheduled rather than forced by a failure.
Spares should follow the same logic: sources, extraction electrodes and the current sense board account for most outages. The board matters because its failure produces symptoms that look like source problems, and swapping a board is a ten-minute test that prevents an unnecessary source replacement.
Limits of the supply's role should be stated. Spot size, depth resolution and material contrast are set by the column and the detector, not by the supply. A perfectly stable current cannot improve a column that is out of alignment, and the effort spent tuning the supply in that situation would be better spent on the optics.
The column's apertures and the supply share a thermal environment that is easy to overlook. Apertures heat from beam interception and their expansion changes the beam's path; the supply's current loop compensates by adjusting extraction, which shifts the beam's energy. Monitoring aperture temperature and compensating the energy setpoint is a refinement that matters in long polishing runs.
Sample charging and the supply's energy range interact in a way that rewards a wide adjustment. Charging neutralisation is found by sweeping energy and observing image brightness; a supply with coarse energy steps cannot find the neutral point. Fine resolution in the energy supply, rather than in the current supply, is what makes the sweep usable.
A closing note on the measurement path: the Faraday cup's amplifier and the supply's internal sense should be compared quarterly, and any divergence corrected in the control loop rather than in the process recipe. Correcting the recipe hides a supply problem and moves the error into every dataset collected afterwards.

