Uncle SEM was equipped with a diffusion pump (DP) which had the task to pump the chamber down to high vacuum around E-6 Torr.
Oil Diffusion Pumps
Oil diffusion pumps have some nice pros, like being completely free from mechanical parts making them vibration free, easy maintenance (cleaning and oil refill every 2-3 years in case of continuous use) and they can reach good vacuum levels.
On the other hand there are some drawbacks: a lot of power is needed to boil the oil (and that’s a problem when your home only has 3kW available), they need water cooling for condensing the vapours on their walls and traps (see below), they take a long time before being operative from cold (at least 20-30 minutes) and vaporized oil molecules tend to rise in the upper vacuum hoses and chamber, contaminating the surfaces. And this is something really unwanted in a SEM, expecially if you are doing EDX analysis. Vapours can cover surfaces and apertures causing many problems in the long period.
To avoid this, some “vapour traps” have been developed. The simplest are placed right over the pump and use a series of water-cooled lamellae in order to condense the oil vapour on them. The temperature in these devices can’t although reach low levels for obvious reasons, making them not much effective.
A second type, which was installed on my SEM, is the liquid nitrogen trap. Much more effective, but it requires handling liquid nitrogen, which is something I wouldn’t really like to do at home.
But how do they work?
It’s quite simple. A heater on the bottom boils the oil and the steam comes out from some compression stages, bringing with itself and concentrating the gas molecules that enter the pump. Those molecules are then compressed in the bottom of the pump, where the backing pump (usually a rotary pump) sucks them away.
Replacing a DP with a turbomolecular pump (TMP)
Already after the first attempts to turn on my microscope I was pissed off enough of the high power consumpion, waste of cooling water (2 l/min) and long start up times of the diffusion pump.
At some point I have found out the existence of a good alternative to DPs: turbomolecular pumps. Those are just turbines which spin at 60000-90000 rpm and pump gas molecules down to the backing pump drain. They don’t require water cooling most of the times, or in most cases a closed-circuit cooler would do the job. And their power consumption is around 120 Watts for a 210 l/s pumping speed.
They can also reach a very good high vacuum, expecially if the whole system is being baked out in order to force water evaporation from surfaces.
So I decided it was time to replace the DP with a more handy TMP. This is going to require machining new adapting flanges and a lot of work!
In my case, I could find a couple of good TMPs on Ebay: one Edwards EXT70 (60 l/s, 60 Watt, very small controller attached to the pump itself) and one Pfeiffer TMU260 (210 l/s, 120 Watts, bigger stand-alone controller).
First tests with Edwards EXT70
First of all, I had to machine a “cap” which could allow me to mount a vacuum gauge on the pump. In the first attempt, I used a pirani gauge.
The vacuum while just the Alcatel rotary pump on couldn’t even reach the same level of that measured on the exit of the RP itself, so that a leak was most likely to be present. I had to disassemble the gauge and noticed it was so old and faulty that there was no wonder why it couldn’t work!
Second try was with a penning gauge. This time I could finally measure a good vacuum, but had to struggle a bit with my old penning controller as it seemed to be out of calibration. For this reason, I had to buy a new one, and with that I could finally read a reasonable pressure. Infact, the old one, was reading a value of 4E-4 Torr when just the RP was running, which is an impossible vacuum for a rotary as they usually can’t go lower than E-3 Torr in the best scenario.
The pump took about 35 minutes to create a vacuum of about 4E-6 Torr, which is pretty good for any SEM!
This time I was also testing the anti-vibration bellows and dampers assembly I wanted to add to the system, as every SEM usually has one. Even though vibrations and noise from the running TMP are almost impossible to feel, they can be an issue when it comes to high magnifications.
In order to attach the Edwards EXT70 to my system I had to machine a big new adapting flange. The TMP has a ISO63 flange, while my scope has a…… not-standard 160mm flange. So I had to machine the whole piece from a 160mm block of anticorodal.
Little problem at this point: the chuck of my lathe can only openup to 140mm. So I had to make a 50mm hole in the middle of the piece and machine a prefectly matching iron bar, make 3 holes there, thread them, saw them from the outer edge and tighten the whole thing with 3 big screws. At this point I could just tighten the 50mm iron bar inside the chuck and go for a “spine machining”! Time consuming, but it worked!
After many hours (days) of work, this was the result:
I always polish all the parts where o-rings are supposed to be with abrasive paste to make them shine as mirrors!
Testing the Pfeiffer TMU260, 210 l/s turbomolecular pump
At this point, I wanted to test my other TMP pump, to see which one was working best. Of course, the answer was obvious, but in case of a defect of any sort, the bigger one could have miserably failed, so it was worth to test before deciding which one to install on my SEM.
As long as my anti vibration bellow has a ISO63 flange and my “testing cap” for gauges also has a ISO63 flange, I had to machine a new flange in order to fit the gauge over the pump which, being bigger, has a big conflat flange of 150mm with 16x 8mm screw holes. This kind of flange is supposed to work with copper gaskets which are used in Ultra High Vacuum systems in order to allow the baking of the whole system, but a standard ISO100 centering gasket can also fit in, so I went for this second option.
This time I also had to machine a 150mm piece and had to face the same chuck problem I had with the other flange. But this time, I tried a new solution for this.
I placed the 150mm x 65mm anticorodal block in my rotating platform under my milling machine and machined three grooves at 120° so that just the jaws could fit and thus hold the block.
The testing was smooth and this pump could reach a good 8E-6 Torr vacuum in just 15 minutes and 4E-6 torr in 30 minutes. No water, low power consumpion, no water hoses all across the room, no boiling oils!
Long story short, it was just a matter of a few more issues to solve and I then proceeded to dismantling the old high vacuum system of my SEM.
Removing the DP and mounting the new TMP.
First step was to remove the old DP and the big nitrogen trap over it. It was really unespectedly easy and fast. A total of just 8 screws kept everything together, so it was a matter of a bit of care and little time to remove all the parts.
After removing both the old rubber hoses which were kind of ruined and the DP, I proceeded to clean all the big hoses with acetone to remove the diffusion pump oil from surfaces as much as possible.
Once removed the nitrogen trap, I could finally measure the holes on the upper flange and the gasket size, so I could machine a proper groove for the o-ring and holes on the previously machined flange. Those extra holes in the picture are due to a mistake in the process, but that wasn’t MY fault: never let your old half-blind man feel useful on your own work EVER!
The groove was also polished with paste.
Now time to assembly the whole thing, step by step.
Testing the new pumping system
Now time for a test!
The magnification LED on the SEM turned up after just 2-3 minutes of pump running at 100% speed. But that’s not a big deal, as my SEM gauge only reads a relative vacuum and it’s also probably a bit faulty, so I’ve waited for 30 minutes before turning the machine on.
The first pictures taken showed some weird stripes.
I had similar ones when my RP hose was accidantally touching the machine basement, so I first thought of a vibration problem. That wasn’t unexpected, as long as I knew that all SEMs with TMP also mount an anti vibration bellow, so I decided to use the ISO63 bellow and dampers assembly I already had and tested in the past. I’ve also covered the cables and hoses from/to TMP with a lead foil and an insulating foam. It took a while, but it was required.
This time, before running the machine again, I decided to install a penning gauge straight into the second, and last, free side port of my SEM with a newly machined part in order to have a real vacuum read inside the chamber.
I was a bit disappointed now about how slowly the vacuum was going down. But I think the fact that all surfaces were exposed to atmosphere and thus full of water molecules should be taken into consideration. That would slow down the process of course!
So it took about 1 hour and 25 minutes to reach a vacuum of 6E-5 Torr inside the chamber. Which is not the best, but still ok for operating.
Then I had to go out and shut down the pumping. When I came back, the system took only 25 minutes to reach 5E-5 Torr in the chamber, which is ok for operation! I’d have liked to reach E-6 Torr, but probably the pump is now so far from the chamber that it would require many many hours of pumping.
Maybe in the future I will install a proper, shorter, anti vibration bellow, thus making the distance smaller and hopefully the pumping time shorter.
Images this time were super nice, even better than in the past! No vibration issues, even at 15000x.
Maybe the old diffusion pump oil was exhaust and vacuum wasn’t as good as it is now.
Little notice: I could STILL see those stripes on the specimens on ONE stub (Diatoms), but not on other stubs, even at high mag…… I have realized then that they might be due to specimen charging issues for a poor specimen preparation (maybe too thin gold layer and no silver paste was used prior to putting the Diatoms on the stub).
So far, I can say I’m very satisfied with the new TMP but I guess there could still be an improvement if I could replace the bellow with a proper one (shorter and wider).
I will soon detatch the whole pumping system from my SEM (I’m waiting to receive the 220V/110V, 1000W converter to remove the Jeol RP from the system -the scope is 110V-, so I can power it separately), so that I will be able to just turn the pumps on and leave for one hour or so, without the need of turning on the circuitry of the SEM, hopefully saving some of its left lifetime for actual operation.