SX100 EDS Adapter Aperture Selector Shaft Seal Leak

[zorch]

   UO's SX100's EDS aperture control developed a shaft seal leak Jan 2022. It was OK when left alone, but changing the setting caused vacuum pressure bursts. Early November 2024 the shaft seal failed and led to 5 Pa in the specimen chamber. The severity of the leak was initially masked by the fact that the vacuum control system thought the pressure was OK (perhaps to be expected, for rough pressure anyway), and opened the secondary valve to the DP. The penning reading immediately slammed to 2.3e-2 Pa, not a terrible reading, but the lack of a normal pressure curve slowly leading to that reading was the tip off. Installing a TC gauge (cleaned for HV/UHV use) teed in above the penning revealed the 5 Pa gross leak. I disabled the DP heater (turned the DP breaker off) for further investigation. That false good 2.3e-2Pa reading must be the result of the penning gauge going into some strange corona discharge state when subjected to 5 Pa actual pressure. Not a great failure mode for this gauge! At least we can now presume that an instantly reached 2.3e-2Pa reading is actually rough pressure and shut down the DP. By the way, the Santovac oil did not crack much at all, even after a couple of days of being in this high pressure state! Tough stuff! I changed the Santovac out anyway.

   Blank plugs were installed in place of the thin spectrometer windows. The spectro bubblers were bubbling normally.

  Both roughing pumps and all TC gauges were used with various valves opened and closed to confirm that a severe leak was present.

   The EDS adapter aperture selector shaft seal was the prime suspect since it had led to vacuum pressure bursts in 2022.

    Removing the OM lighting/Video camera assembly allowed removal of the EDS detector. Next I disassembled the EDS port adapter to access the EDS Aperture Control shaft seal. The shaft seal was found to be a black rubber top hat type. What is surprising is that one face of the seal, which would have to act as a seal, has raised lettering! The only reason this did not leak when new was that the entire seal was soft and compliant enough to conform to the seal space. Once it hardened with age, those raised letters acted as a gas path gross leak. As for the conical part of the seal, I don't see how it served any purpose because nothing was compressing it against the shaft.

   The original design of the shaft seal is roughly depicted on page 2 of the pdf. The new design is on page 4. The 4022... part number is an O-ring from Thermo Fisher instruments division. The part numbers in blue are Mcmaster.com parts. The aluminum spacer was made by our machine shop crew. If replicating this new design, be sure to pack a bit of Apiezon L or M grease on and between the two Viton O-rings. 

   With the new seal arrangement installed, the specimen chamber pressure has reached 6e-5 Pa, and changing the aperture selector no longer causes pressure bursts.

Note that the top hat seal is not visible in the photos because I neglected to extract it from the brass part with the two screw recesses.

[sem-geologist]

zorch,
Thank You!
while I never seen this leak on our two SX EPMA's It is quite scary to realize that there is potential of such a failure silently lurking there.
I checked the schematics and seems that SXFive FE still use that design for that part.
Although I do not switch between EDS apertures (there was a moment I did experimenting around, but then stopped), and use only smallest aperture (we don't have beryllium as option) - for most of probe work it keeps EDS at lowest dead time and has best spectral resolution (We have Bruker XFlash Nano). I guess It could start leaking randomly and it would be better to not find out what happens then, thus I am going to makes those proposed modifications on our two probes at next occasion.

placing marking letters on o-rings - That is a crime against humanity!

As for Penning gauge, Last time we had just bad readings with 3 from 3 units (no leak); I had enough and so I  replaced whole damn thing with modern gauge and custom designed interface PCB (https://smf.probesoftware.com/index.php?topic=1543.msg12133#msg12133). It is already one year it is working without hassle, and finally we have reliable vacuum state after every sample exchange requiring no knocking with wrench to the gauge (Seriously - all previously used penning gauges often needed gentle knock with wrench so it would wake up from 6E-5Pa and start showing real state of vacuum).

[zorch]

Thanks Sem-Geologist! By the way, this leak was coincident with a severe leak at the diffusion pumps fluid level check port! That O-ring had suddenly developed many cracks and was itself leading to a high pressure in that pump. The problems seem to come in pairs! As you can imagine, having two big leaks show up at the same time can lead to extra head scratching.

S.

[sem-geologist]

Do you have any details what kind of o-ring's fit there and how much santovac oil needs to be filled? (While we have no signs at this moment of a leak, but, taking into consideration the age of 25 years - it rather would be not stupid to prepare in-advance). What would be a procedure with venting the diffusion pump for oil replacement? The access of diff pump is quite limited (but which maintenance access is not hindered on that instrument ), Is it manageable to do that with pump in-place?

[zorch]

For the O-ring, I used a Thermo Fisher Scientific part from their EM division, 4022 199 20111 Viton O-ring. You can buy a viton O-ring from mcmaster.com that matches the dimensions in the attached pdf. If the old o-ring has fused to the stainless steal port or plug, you must scrape the o-ring material off with a non-metalic tool. I used one of those hard plastic I-phone cracking tools used to replace the batteries. If you want, you can try this out only on the oil level check port, since its the one that will degrade the most. The oil drain port is more likely to be good since its immersed on oil. Once the new O-ring is in place, tighten the nut fitting onto the O-ring reasonably tight, but not superman tight. The O-ring will degrade sooner if massive compressive force is applied. Here's a procedure for the entire operation:

- Record all pressures while still operating normally, to be sure things are back to normal once your work is done
- Vent the specimen chamber
- Put the vacuum system in Halt state
- Turn off the DP heater breaker (below the load lock) so that your first run up is done with the DP cold
- Allow the DP to cool for 10 minutes, but no longer since we want to avoid the Santovac oil from thickening too much
- Turn off the water chiller to avoid any more cooling of the DP
- With a 250 ml beaker at the ready, open the drain fitting and drain the Santovac 5 into beaker
    Note that Santovac 5, when new, is yellow in appearance. You should get around 100 ml into the beaker
- Clean old O-ring material off of this drain fitting with a plastic (to avoid scratching metal) tool, then fit a new
    O-ring here, and reinstall this fitting, tighten with significant torque, but not superman force.
- Remove the oil level check nut/dip stick. This is the higher of the two nut/fittings. Replace this O-ring
- If the Santovac 5 oil looks brown and or has charred debris in it, you might want to remove the DP and clean the
  jetter, but this is best done by someone with experience cleaning these pumps
- Poor 105 ml of new Santovac into a clean beaker (gets 100 ml into a Balzers PDI100‐W. If you have another pump
      model, you'll need to check its specs for the correct oil volume)
- Heat this Santovac to 45C to allow it to reduce its viscosity
- Draw this warm Santovac into a 150 CC plastic syringe
- Force this Santovac into the oil level check fitting
- Install the oil level check plug/nut
- Restart the water chiller
- Go from Halt state to Pump state (the DP heater is still disabled by the breaker = OFF)
- Watch the DP exhaust pressure, it should drop to the value you recorded earlier. If it does not fall to ~5Pa,
    tighten the two fittings by 1/10th turn CW. Repeat until you see around 5 Pa or lower after the backing valve
    has opened.
- Go to Halt state, turn the DP heater breaker ON
- Go to Pump state, everything should ramp up in the usual way

Note that cleaning up any stuck-on O-ring material from the oil drain port will likely require a work light and an inspection mirror to see what you're doing since that port may be turned away from rear of the instrument. If you're not very experienced with the above procedures, I recommend you call in the pros! But, if you make sure the DP heater is disabled, you can avoid the biggest kind of trouble, backstreaming a hot DP! Also, you don't want to have the RP roughing things for very long since backstreaming from that pump can contaminate chambers.

Zorch

[sem-geologist]

Thanks Zorch,

This diffusion vacuum pump guide part is also very helpful. Albeit, I see some mechanical differences on our much older SX100 equipped with Alcatel Crystal 102 diffusion pump.

It has different dimentions and different fill and drain o-rings than those presented for Balzers PDI100W.

After preventive machine shutdown for 16h of electric installation maintenance event in the building, our SX100 could not recover roughting (backing) vacuum for the diffusion pump on cold start. I ruled out all other sources of leaks and possible problems (possible reduced roughting pump power, leak in piping, wrong gauge reading...) and all clues points to diffusion pump. I still cant find any source or piece of information on these particular diffusion pumps.

My question again to all (in particularly those who work/-ed with EPMAs manufactured from 1980 to 1999) if anyone has any paper manual on those pumps (Alcatel Crystal 102) or a note on oil capacity?

My primary question is the amount of oil supposed to be in the pump (My wild guess is from 100 mL (maybe thats why it is model Crystal "102" (102mL ?)) to 125mL (the ammount for Edwards Diffstack 100/300 -similarly shapped and sized pump). (Agilent's HS-2 - 100mL - similar sized pump).

Another difference I found out is the oil. I think in our case it is filled not with Santovac 5. The viscosity of it more resembles that of water, than of honey (Santovac viscosity at 35C is known to be about 2000 - it should behave like dense honey at 20C to my understanding). It is flowing and dripping from the fill port dip-stick with too much ease. The filling port dip-stick has no oil measure grating, oil level wetting gets about to the middle of the dip stick (without a user manual or min-max grating marks it is hard to tell what does that mean). I think it could be silicon oil (which one? 702? 704?)  as it has much lower viscosity than santovac 5. It also is vaporizing at atmospherec pressure! I temporarily left fill-port covered with powder-free wipe, and forgot to cover back for night back with plug. During night it whole soaked in oil    (the wipe had not touched the oil and could not soak itself like capilary - makes me believe it happened only from vapor). I thought one of crucial physical properties of diffusion pump oil is very low vapor pressure... Just imagine contamination in vacuum it creates (we dont have cold trap). I think I miss something crucial in this case (can fluid vaporise more easily in atmosphere pressure than at vacuum at room temperature?).

O-ring of fill port.
I had to do no scrapping, O-ring could detach from the port without leaving any residuals. It looked even kind OK (no cracking) on the first look, albeit a bit flat as for O-ring. Fortunately accidentally I got over very handful handbook on vacuum systems (Handbook of Vacuum Science and Technology by Hoffman et al, 1997. Elsevere Science & Technology Books), it has whole section on O-rings. It allowed me to comprehend and understood what exactly did I observed - the O-ring probably was intialy round, but got flattened due to setting of compression with time and heat applied. Its set-compression reached 100% (that is it does not return to its original form any bit no more then metal- elastomer seal pressure is relieved) and due to thermal cycling of shutdown, and followed cold startup, it started to leak. The diameter of ring in not compressed direction is about 2.8mm, where in compressed direction it is only 2.2mm. The compression which was initialy applied would be at least 21%. It is within recommended range of 20%-25% given in the mentioned book.
replacement O-ring should be ~9.1mm inner and ~14.7mm outer diameter. (what is proper O-ring dimention nomenclature for easier search up?)

Not applying a "Giant" force is good general approach for tighting the Fill/drain ports. I think I will try to be a bit more precise and will try to hit that 20%-25% of elastomer compression range. That is possible to achieve by measuring thread size - it is then clear how much turn it requires to tighten it to that particular O-ring compression range (it is better to aim for 20% than 25%). In my case thread step is ~1.56mm. To compress the elastomer of 2.8mm thickness by 20% would require:
2.8mm * 0.2 / 1.56 * 360 ~= 130 degrees of tightening. In case of 25% compression it is about 160 degrees.
So more practicaly in my case - after feeling the obstacle of threaded plug, it should be tighted twice by 60 degrees (60 degrees of hexagon head for easy orientation * 2) and just a little bit more.

I am very tempted to go an extra step and drain all oil, clean the pump dry and refill with santovac 5 as: 1) contact of drain port o-ring with oil rather wont help to mitigate the "Set compression" problem, and It very highly possible it already has 100% set compression at this moment; 2) the currently used oil vaporises at room temperature - looks for me not very good choise - contaminates vacuum inner surfaces and stuff such as vacuum detectors, electron detectors etc... Seeing it makes me believing, and so I finally understand how our previous BSE got so badly contaminatted. Better oil will payback with less maintenance needed for other parts. Any ideas?

[sem-geologist]

Additon considerations.
2.8x9.1 (mm) seems to be not standard size for O-ring.

Also 2.8mm probably is not right O-ring crosssection dia, as O-ring in long term elevated temperature could experience mass transport from compressed part into sideways parts. Recalculating observed ring into non-deformed form shows the crosssection of ring should have ~2.75mm dia.

Additionally considering hot temperature conditions FFKM (perfluorelastomer, i.e. Kalrez(R)) instead of FKM (fluorelastomer, i.e. Viton(R)) O-ring would better fit the function. Idealy I think Kalrez(R) 7075 would fit best as its set-compression is one of lowest in the market (10% at 200C, compared to Viton(R) of 25% at 200C). That could extend time between it loosing elasticity in sealing direction. Kalrez 6k types seems to have worse elasticity retention for ISO short term standard measurements (72h experimental measurement; worse than Viton), however their long term experiments for high temperature show that Viton would set-in compression after a year and Karlez material still would have enough elasticity able to keep seal OK. (even Kalrez 6k types in half a year at elevated temperatures nearing 200C will overtake Viton elacisity specs). Taking into consideration replacing FKM ring with FFKM type of ring, 50% larger thermal expansion compared to FKM needs to be addressed when deciding on dimensions of new ring.

(2.75 * 2 + 9.1) * 105% = 15.3 mm  – That stays withing 16mm at ~100C.
Thus crossection of FFKM ring where thermal expansion at 100C is about 8%, should have lower ring crossection to fit within 16mm at high temperature.

I think FFKM  with AS110 O-ring size (American Standard, 9.19mmx2.62mm) can replace the FKM 9mmx2.75mm in this case within limits of clearence of fill/drain ports of Diffusion Pump Alcatel Crystal 102.

[zorch]

More info on the oil fill port O-ring for the Balzers PDI100‐W. See attached PDF for drawing of that O-ring.