Skip to main content

The room matters as much as the microscope: siting an electron microscope in Bangladesh

Vibration, stray magnetic fields, acoustic noise, temperature and monsoon humidity decide what performance you keep after installation. A planning and commissioning sequence for Bangladeshi campus and hospital buildings.

An instrument that met its specification on the factory floor can fail its acceptance test in a room three metres from a lift shaft, and there is nothing wrong with the instrument. Electron optics measure distances that are smaller than the movement a passing lorry puts into a concrete slab, and smaller than the beam deflection produced by an unremarkable magnetic field from a distribution panel on the floor below. In practice the room, not the column, sets the performance a Bangladeshi laboratory actually receives.

The costly part is that all of this is cheap to design in and expensive to correct later. A room chosen at the start of the project costs nothing. Shielding a bad room after the microscope has arrived costs a significant fraction of the microscope.

Do the site survey before the tender closes

The order of events that works is: identify candidate rooms, obtain the manufacturer environmental limits, measure the candidate rooms against those limits, then write the tender with the chosen room named in it. The order of events that goes wrong, and which is common, is: win the funding, order the instrument, then look for somewhere to put it. Make the site survey a condition of award. The bidder should state in writing what the room must deliver, and the measurement should be made in the actual proposed room with the building running normally.

Measure over a full working day, not for ten minutes on a quiet Friday. The lift must be running, the air handling plant on, the generator tested, and if there is construction on an adjacent plot, which in Dhaka there usually is, that must be in the record too.

Vibration

Vibration moves the column and the specimen relative to each other. Because the image is built up line by line over seconds, that relative movement appears as distorted or doubled features, and as a periodic waviness whose appearance changes when you change the scan speed. It is worst at high magnification and it is often blamed on the operator.

On a Bangladeshi campus or hospital building the sources are predictable: the lift and its machine room, roof and basement water pumps, standby generators, air handling units and chillers, air compressors, road traffic, and piling on the site next door. A ground floor slab cast on grade is the best available answer in most existing buildings, and an upper floor slab is the worst. If the building has not been built yet, get the vibration criterion to the structural engineer while the slab thickness and the column layout can still be changed, and ask for an isolated plinth for the instrument. Retrofitted isolation tables help with some frequencies and do nothing for others, so they are a correction, not a plan.

Stray magnetic fields

An electron beam is deflected by a magnetic field, so an alternating field at mains frequency sweeps the beam back and forth as the image is scanned. The result is a regular distortion, worse at slow scan speeds, worse at long working distances, and worse on field emission columns than on tungsten ones. Slowly changing fields, produced by large steel objects moving nearby, produce drift instead.

The sources are low tension distribution panels, busbars and cable runs in the slab or ceiling, transformers, lift motors and their traction cables, and unbalanced three phase distribution anywhere in the vicinity. Measure with a fluxgate magnetometer at the exact position the column will occupy, at column height, over a working day. Then treat the results in this order of preference: choose a different room, move the source by rerouting a cable or relocating a panel, and only then consider magnetic shielding or an active cancellation system. Shielding a room is a specialist job with a specialist price, and it constrains everything you do in that room afterwards.

Acoustic noise

Sound pressure couples directly into the column and chamber and shows up as streaking that appears when a door slams, when the air conditioner steps up, or when two people hold a conversation next to the instrument. The usual offenders are inside the room: the roughing pump, a wall mounted split air conditioning unit blowing across the column, an alarm sounder, and an open plan layout where the console shares the space with everything else.

Put the roughing pump outside the room on a flexible connection and an isolating mount. Use ducted air conditioning with diffusers positioned so that no airflow is directed at the column. Place the operator console away from the column rather than beside it. These are layout decisions, and they cost nothing if they are made before the room is fitted out.

Temperature and monsoon humidity

For thermal stability the rate of change matters more than the set point. A split air conditioning unit that cycles on and off produces exactly the wrong pattern: a saw tooth that the instrument follows as focus drift through every long acquisition. Specify continuous cooling, two units in lead and lag so one can be serviced without shutting the room down, and no direct airflow onto the instrument.

A recurring problem in shared university buildings is air conditioning that is switched off at night, at weekends and during holidays. An instrument allowed to soak in Dhaka ambient conditions every Friday will drift for hours on Sunday morning, and during the monsoon it will condense water on any surface below the dew point, including chilled water lines and the parts of the chamber that were last vented. Corrosion on stage components and on sealing surfaces follows, and hygroscopic samples are ruined before they reach the column. Argue for a dedicated, continuously running supply at the design stage, budget for the electricity, and add a dehumidifier plumbed to a drain rather than one with a tank that somebody has to remember to empty.

Keep the room slightly positively pressurised against the corridor. Dhaka corridor dust is a real contaminant, and it arrives every time the door opens.

Power quality, load shedding and earthing

Voltage fluctuation and switchover transients do more damage to sensitive instruments here than outright outages. The generator changeover is the dangerous moment, not the darkness. Specify an online double conversion uninterruptible power supply covering the microscope, the vacuum system and the workstation, sized for a controlled shutdown rather than for continuing a session, and add an isolation transformer where supply noise is a known problem in the building.

Services, layout, and what gets forgotten

Cooling water or air cooling has to appear in the room heat balance, and so does the heat rejected by the chiller if it is in the same space. Dry, oil free compressed air is needed where the instrument uses pneumatic valves. Dry nitrogen for venting is worth plumbing in from the start, because it keeps the chamber dry during the monsoon and shortens pump down all year. The roughing pump exhaust goes outside, not into the corridor.

Then the items that appear on no specification and cause real damage. No water pipe should run above the instrument, and no toilet or wet laboratory should sit on the floor directly above the room: leaking sanitary plumbing has ended more instruments than power failures. Cable trays should not cross above the column. The floor should be anti-static and cleanable. The door needs a seal and a viewing panel so nobody opens it to check whether the room is occupied. And check the delivery route before ordering: crate dimensions against lift capacity, door widths and corridor turns are a routine and avoidable reason for installation delay in older campus buildings.

Commissioning and the baseline record

  1. Repeat the vibration and magnetic field survey after the building is occupied, with lifts, air handling plant and generator running as they will run in service.
  2. Log room temperature and humidity continuously for at least one full week, including a weekend, before acceptance.
  3. Run the acceptance imaging test on a certified resolution standard at every accelerating voltage you specified, including the low voltage condition.
  4. Capture a slow scan image at long working distance and keep it. It is the most sensitive record you will ever have of the room, and it is what you compare against when performance is questioned in two years.
  5. Record pump down time from a standard vent condition, so a future slow pump down can be recognised as a change rather than argued about.
  6. File the earth pit test result and the uninterruptible supply transfer test alongside the acceptance report.
  7. Repeat the resolution test annually under the maintenance contract and keep the images together in one folder.
Symptom in the imageProbable environmental causeFirst thing to check
Regular waviness that changes with scan speedAlternating magnetic fieldMeasure at column height with a fluxgate while the lift runs
Doubled or blurred edges only at high magnificationFloor vibrationNote whether it correlates with pumps, air handling plant or road traffic
Streaks when a door closes or someone speaksAcoustic couplingMove the roughing pump out, redirect airflow, relocate the console
Focus drifting through a long acquisitionRoom temperature cycling or stage settlingLog room temperature against the acquisition time stamp
Condensation or corrosion on stage and sealsHumidity and surface temperature below dew pointCheck dehumidification and continuous air conditioning through weekends
Random resets and lost sessionsPower qualityCheck the uninterruptible supply mode, generator changeover and earth impedance

Back to all Insights