JEOL JSM-T330A

Our scanning electron microscope. Quick specs:

Maximum acceleration voltage: 30kV

Spot size: 4.5nm @30kV, SEI

Manufacturing year: 1988

The diffusion pump and diffusion pump throat coupler have a water cooling circuit. It came with an ancient hose (OD: 15.5mm, ID: 8.5mm). This should get replaced.

The vacuum control system operates on valves only - the diffusion and roughing pump start immediately when the device is switched to 'start'!

600W model by JEOL. Not much in terms of docs.

(To be) filled with Ravenol ISO VG 68.

Ours currently has surface rust on the o-ring mating surface between the body of the pump and the baffle.

During first DP disassembly it turned out that the core was not held together fully, with the nut tightening it together being loose. This was probably from 40 years of vibrations.

O-rings:

'LGSH' Large Sample Holder. This has a dovetail on which a 'specimen holder' is supposed to mount, on which then you can ether mount wafers or slugs with samples.

We don't have any 'specimen holder', just the 'Holder receiver'.

Aftermarket holders/adapters are available, notably the EM-Tec JV50 / JV50P / JV50J, retailing at around 175-195€. Judging by the description, we should be able to use any JEOL holder compatible with the iT210, iT200, iT100, 6510, 6390, 6380, 6360, 6010, 5700, 5600, 5500, 5410, 5400, 5300, 5200, T330, T300, T220, T200, T100 and T20.

O-Rings

Seemingly allows for beam control. Asserting pin 3 disables display and enables external control. Pins 1 and 7 then allow for X/Y scanning (+/- 5V?). Pin 4 should provide detector data.

JST XH connector on a three-colored (black, green, orange) wire harness. Cobbled together by q3k, will be used for image acquisition.

H/V sync needs to be figured out (by following the wires or with an oscilloscope).

H/V sync values are ramps +/- 5V (opamps are supplied +/- 8V).

The wires all go to U54 (opamp input) on PK00492 (the scanning board). This opamp is used to read back data from the EK9 connector. This connector allows for chaining scan/detector though the BCX x-ray unit. There might be signal integrity issues with tapping the sync data like this, we'll see.

Connector types observed in the machine:

1. 7556 series: 7.50-5.00mm-vertical-header-7556-series-header.pdf
2. DF1 series: df1-12s-2.5r26_05_cl0541-0068-0-05_catalog_d49650_en.pdf

1. Machine/acquire sample holders
2. Track down column leak(s)
3. Check/replace scintillator

1. Replace diffusion pump with turbomolecular pump
2. Get a better roughing pump (scroll) so it doesn't spew oil everywhere
3. Figure out what the 'PC' wire harness does
4. Digital control and acquisition
5. E-beam lithography

1. 2x M4x7.5mm screws for clamping down transformer wires in control module
2. Double-sided carbon adhesive tape ~24€
3. (Torx?) socket cap head M4x12 screws for vacuum fittings

Replaced the succbone relay boards with a MODBUS-based system. Added MODBUS-based temperature measurements for the diffusion pump.

Buzzed out 'PC' connector and added scan output connector.

Cleaned apertures in column a bit after building custom tool to extract the pole parts from the column.

Got some even better images.

Fixed alignment of the cathode. Spent some time dialing in optics.

Got some better images.

Attached samples to stage (QF105 die, stainless steel M3 screw).

Took first images, but they're very grainy.

Weird electron gun behaviour. Filament doesn't seem to saturate. Increasing current would increase emissions, but past some point the emissions would just stop. No second bump even with the filament knob all up to 11. Gun bias anything other than lowest setting would pretty much stop all emissions.

q3k says:

1. Maybe I reassembled the wehnelt wrong?
2. Maybe I damaged the cathode when cleaning the wehnelt?
3. Maybe I reconnected the HV cable to the electron gun wrong? I pulled it out when I thought I had to empty the transformer tank.

To set the electron gun bias according per the manual's process, filament current readout is needed. The one built into the scope broke because the comparator in the checker board got fried again. We should figure out what's happening there. We might also be able to just remove the comparator.

Tested diffusion pump with new oil. After outgassing, got down to the 6e-3 mbar that we also got with the other oil. Tested the ionization gauge which started showing lower values. Unsure whether we can still trust the pirani gauge values in this pressure range, but also uncertain about the state of the ionization gauge.

Turned on the electron beam for the first time.

Replaced the diffusion pump oil with a proper one. Unfortunately didn't get to test it yet.

Replaced the DP heater wiring. Did some work on the succbone control system.

Investigated the roughing pump in more detail. Connecting our pirani gauge to the pump directly, we can reach 1.5e-2 mbar. Not great, but significantly better than the results with the full system connected. So we're out to hunt for a few more leaks at least.

Serviced the “RP vent valve” as this hasn't been done before. No improvements, though.

Turned on the diffusion pump. Got to ~6e-3 mbar. Nice bump from outgassing of oil.

Will probably have to replace oil with proper diffusion pump oil/fluid…

Finished connection of fake-pirani evac-controller inputs. Added a few more software interlocks to succbone software to prevent unwanted system behaviors. Verified that all software and hardware interlocks are working correctly.

Installed second relay board for succbone. The additional relays are used to emulate pirani gauge inputs to the original evacuation controller.

Connected most equipment to the succbone control panel (roughing pump, diffusion pump, chiller alarm, SEM button panel intercept). Verified that roughing pump and pump-down/vent buttons can now be controlled by succbone remotely.

TODO

Rebuilt valve blocks. Reinstalled diffusion pump.

Mounted succbone into new control cabinet, preliminary support for RP/DP/vacuum button control (not yet wired in).

Rebuilt diffusion pump. New oil (Ravenol ISO VG 68) was added, but it's a crapshoot on whether it will work.

Connected new chiller to diffusion pump, made sure it runs.

Triple checked seller's info on whether the diffusion pump has been serviced - it wasn't, so we decided to fully service the diffusion pump.

Diffusion pump was removed from column. Surface (?) rust was found on the mating surface between the diffusion pump body and the diffusion pump baffle. This needs to be addressed.

Waiting for acetone to start the process of cleaning and rebuilding the pump.

Connected succbone, we now have pumpdown graphs at succbone.lab.fa-fo.de. Best result: 3.4 x 10^-2 mbar after ~30min of roughing.

Testing replacing broken inter-module vacuum hose with 25mm spiral PVC hose from Hornbach. It leaks even when clamped hard (~7 x 10^-2 mbar). Heating with heatgun before clamping down helped a bit, but still not great.. Will try 24mm hose from Hornbach and 22m hose from eBay.

Work on replacing the components of the evacuation board.

Board works again.

Accidental swap of FM1 and FM3 connectors on the evacuation control board caused circuitry to be damaged. Pretty much all components are dead, but no other boards seem to have been damaged. Replacements are being ordered.

Whoops.

Installed new pirani gauge and adapter on top of the column; removed the ion gauge for now. This allowed for a first reading of around 100mbar which is not in the ballpark of a real vacuum. Recalibrated the pirani gauge to ambient air level, no improvements. We suspect a large leak.

Found a leak on the one hose that was replaced. All '88 hoses, while crusty, seem to work okay-ish. “Fixed” the new hose with lots of duct tape, improved vacuum to 1mbar.

Built a somewhat sturdy holder-contraption for the roughing pump, so that we can mount it outside the SEM's housing. Then bypassed the new hose by plugging the roughing pump into the 3-way-valve directly using the old (but seemingly better) hose. Ran a new test and got the vacuum down to 0.1mbar.

q3k confirmed we would want to reach somewhere below 0.01mbar possibly.

Ran some tests on how the auto-venting behaves, it seems it enables automatically after 5min-6min if PRE EVAC was not reached until then. Thinking of faking that signal for the SEM just to see if it would stop venting automatically.

Replaced missing o-ring on column panel. Slightly oversize (at least 2-3mm?) but got it to fit. A warning label has been put on the panel to make sure no-one takes it apart.

Replaced comparator in checker, fixing spurious alerts.

Still waiting for pirani gauge.

Attempted to discover vacuum leaks by ear. Found a blinding panel on the SEM that's not only super dirty, but is just missing its o-ring. This is likely where the previous owner(s) had an X-Ray probe attached, and the panel was reinstalled without an o-ring during decommissioning.

Ordered o-ring for above, ordered a custom machined blinding panel for the broken pirani gauge (as a test piece).

Attempted to cool diffusion pump with garden hose and garden pump contraption proved unsuccessful - the heat from the garden pump is already too much, we would really need a radiator. Shelving the issue of cooling the diffusion pump for now.

Discovered that the pirani gauge on the SEM is physically broken.

Discovered that the aftermarket vacuum gauge is actually a cold cathode ion gauge (Balzers IKR IKR 020) and the readout is showing its maximum pressure rating, so it seems not broken, and we're probably having a vacuum leak in the column. This tracks with how weak the vacuum seems - we can easily pull out the pirani gauge when the column is being roughed.

Cleaned up some of the static vacuum seals. Ordered a Pfeiffer TPR280 pirani gauge to mount (for now) instead of the ion gauge, this will be used to figure out our rough vacuum situation and to troubleshoot the leak.

Broke the KF DN 40 clamp on the external vacuum gauge. Ordered a replacement.

Realized the endstops preventing over-rotating the gun bias knob are loose. Possibly broke the underlying potentiometer in the process (VR1 in PK00334(BB), inside oil tank). Current hypothesis is that the pot is not soldered to a PCB, but was secured with a nut to the tank lid, and turning it past the stop caused the nut to loosen. Decided to figure it out later, if we find more reasons to open the oil tank we'll batch them.

Analysis of the evacuation controller board showed that it was waiting for the diffusion pump to heat up. Bypassing that by disconnecting the lower diffusion pump sensor (TH2). The scope now actually attempts to evacuate the column. However, the internal vacuum gauge is still pinned.

Took apart the two valve blocks in order to understand them. Diagrams will follow up soon.

Verified that the checker comparator (IC6/U6 IR2339) is likely toast: output is at -15V while it should be Hi-z / pulled up to 15V.

Reconnected the roughing vacuum pump and attempted to pull a vacuum. There is a vacuum forming at the line from the RP, and a vacuum further up against the valve block away from the column. However, there is no vacuum at the column. Inspecting the valve blocks / motors, no movement happens at the valve block closer to the column. This seems suspicious, as looking at the vacuum system sequencing diagrams the column should be at vacuum pretty much from the start - other than '(DP BACKING)', which should only run on a timer, as there is no sensor that would detect that the DP is been evacuated. The scope has been left in this early state for a couple of minutes, with no resulting valve movement.

Further troubleshooting requires understanding how the vacuum evacuation controller board works, It is fully responsible for the vacuum sequencing logic - it gets momentary signals from the evacuation buttons, and lights the LEDs next to the buttons. Taking apart the valve blocks will likely also be necessary for better understanding, but it's better to avoid that as long as possible.

The board in the scope seems to be a newer, more integrated, microcontroller-driven (Z80) version than the schematics we have (which show a multi-board system, and are also incomplete). Thus, the board will have to be reverse-engineered from scratch to understand how the vacuum system is supposed to work. It has been removed from the scope (after labeling all wire harnesses by the connectors they attached to), and it is now being analyzed.

Investigated the high voltage circuitry to determine how to disable the acceleration voltage during bringup. Switch EOS-HT (S1 on board PK00018, above transformer) has been set to 'OFF'. Verified that the base of transistor TR2 is now grounded - this should disable generation of acceleration voltage while allowing us to check the functionality of the HT switch and indicators on the operator console. Leaving CRT voltages enabled as they're less scary and we want to check the scan of the CRTs.

Disconnected RP and DP. Performed first start up of device. 'LOAD CURRENT' and 'DC POWER for HV' alerts are firing. Re-enabled EOS-HT switch. Alerts are still firing.

Checker shows 0.35/70µA, That should not be firing the alarm according to the manual (specified to fire at >0.75/150µA). Investigating the meter board (PK00514), inputs to comparator (U6/IC6) are -1.4V (reference, pin 4) and -0.58V (control voltage, pin 5). Reference resistor divider values measured in circuit are different from expected (R54 should be 20k, is around 1.42k, R55 should be 270, is around 1.35k). Resistor color band readout is unclear, but seems inconsistent with values.

Expected behaviour given above: non-inverting voltage is higher than inverting voltage, so output should be high impedance. R52 should pull output high. TR4 should conduct. Reset line of timer should be pulled low, thus alarm should not fire. In reality, alarm is firing.

Current conjecture: IC6/U6 is toast. 'DC POWER for HV' is also on U6, which would explain it firing, too.

Column raised on springs, all panels installed.

Vacuum hose between DP and RP has been shortened to allow both modules to sit flush against each other. Modules were then put up on their feet.

Figured out the stuckness of the stage tilt: more force needs to be applied. Likely an old gasket around the chamber which causes high friction/stickiness.

Checked 230/100V transformer: no breakthrough between windings, primary and secondary both at ~0.2ohm. Checked grounding of transformer.

Reinstalled high voltage connectors into high voltage transformer, including ground.

Measured resistance of diffusion pump heater: 16 ohms. This means that the diffusion pump alone eats 625W of power! We might want to use the ginormous heat exchanger we have after all, as that's a lot of power to dissipate.

Finished connecting two modules, started attaching side panels.

Repaired frayed wires. Moved column module on to its casters. Started connecting two modules together.

First inspection. Moved control module onto its casters. Roughing pump put back on isolation springs. Wires from roughing pump to PJ25 connector frayed.

Transported to lab.