Home built mass spectrometer, continued

[Jan Ohlsson]

I have cleaned up my spectrograms by using a sawtooth sweep instead of the triangle sweep. I have also identified my peaks better by injecting known gasses into the system.

Please take a look at
https://diy-masspectrometer.se/contact
and please feel free to comment or be critical!

Kind regards,
Jan
Stockholm, Sweden

[ArkadiuszGibes]

Hi Jan!

This is great design, i was thinking about building something similar for leak detection in my vacuum system. It's good to see that someone already did it. I have question how did you mesure and adjust the magnets size/position ? Do you have gausmeter for that propose ? Or you used some different technique ?

Best Regards
Arek

[Jan Ohlsson]

Hi Arek,
I did measure the magnetic field with a sensor to 210 mT, but it really is not necessary to know the exact value. What is important is to keep the distance between the pole pieces as short as possible and to use as big and many stacked neodymium magnets as is practically possible. Use a 90 degree magnetic sector shape for good focus. The slits should be at about one radius from the edge of the pole piece, and the distance will be affected by the fringe field, so it must be adjustable for best beam focus. In my design the distance must be within 0,5 mm from optimum for good separation.

The larger the magnetic sector radius is, and the stronger the field, the larger the amu that can be detected is. Mine goes up to about amu 40, but for leak detection with helium at amu 4 you can get away with a much weaker field and smaller radius. But it is nice to be able to detect water and the air gasses, and those go up to amu 32.

[ChristofferBraestrup]

This really is an exellent project!

I think you should consider buying a channeltron as a detector. The resolution you'll gain is enormous - which means you can make your slit much narrower and get much better resolution.

They're as easy to hook up as a photomultiplier with built in dynode divider, basically you just need a few resistors to bias it at high voltage (1-4 kV typically), a decoupling capacitor and a current preamplifier out. I'm currently playing around with a few ex-soviet FEU-6 multipliers, but there are plenty of western models for sale on Ebay too.

[Jerry Biehler]

Bad thing about channeltrons is they don't like to sit at air for long periods, I have a spare one for my RGA and it is in a 6" section of conflat with ends on it, one end had a copper tabulation so they can seal it off under vacuum.

A faraday cup is plenty sensitive for most RGA work.

[Chuck Sherwood]

A similar project has been on my project list for a while. I acquired some parts from a friend that he salvaged from some very old equipment that could be quite useful if I can get some help figuring out how to use them.

Here are some pictures of what appears to be a single deflection 60 degree sector.

The source

The detector with electrometer tube.

Here are some pictures of what appears to be a double deflection unit.

The source

detector

The single deflection unit seems pretty complete and even has the magnets.
The double is missing something at the joint and there are no magnets.

I know nothing about the original machines and it would be great to get some general advice on what is needed to make something useful here. I have several books on mass specs but they just seem to put me to sleep and the info is so general. If someone recognizes these parts it might be possible to find some kind of manual that shows a block diagram of how they were used. Once again. All advice appreciated.
chuck

[Richard Hull]

The vacuum tube in that looks like the venerable 5886 electrometer tube. Used by Keithly for many years. 10e-15 amp easily measured and, when carefully used in a vibrating reed gold head by Keithly, can easily differentiate +/- 200 electron currents. Be careful 1.3 volt filament if you power it up, (some were rated .9 volt). 22 volts used on the plate is the norm. Old 5886 tubes found in older electrometers and helium leak detector, need to be removed and carefully cleaned with acetone and dried thoroughly. No dust or grease or any thing allowed on the glass especially around the isolated grid wire lead if it is to operate as an electrometer tube.

Richard Hull

[Chuck Sherwood]

I collected some data on the detectors. Both the single and double sector units have basically the same construction with minor physical differences associated with the construction.

The top 3 plates are grounded.
Plate #1 has a 17/32 by 1/2 inch square hole.
Plate #2 has a 3/8 by 1/2 inch rectangle hole
Plate #3 has a 1/8 by 1/2 inch slot.

Plate #4 is too close to #3 to measure but has a slot slightly bigger. I estimate 2/10 inch by 1/2 inch. This plate is connected to the plug via a small coax hardline. I do not understand the purpose of this disk and what voltage is applied to it.

Plate #5 is grounded and has a slot even smaller that #3.

The bottom plate is brass and has a large insulator in the center with a small disk suspended above it. The disk is the collector. IMO.

The small disk is connected to grid #1 on the electrometer tube and a glass resistor that measures 100 GigOhm.

The other end of the resistor and all electrometer tube pins are brought out via the connector.
Most of this is not mysterious. All the grounded disks catch ions that are not associated with the current acceleration voltage. BUT what is a mystery (to me) is the purpose of disk #4 and what voltage is applied to it.
Chuck

[Chuck Sherwood]

The source is constructed as follows and resembles sources described in some of my books. But there is one unique difference. Book discuss a magnetic field perpendicular to the ion beam. I don't see that here.

Top 3 plates are grounded just like the detector.
Plate #1 has a rectangular hole 15x13 mm.
Plate #2 has a 10x13mm hole
Plate #3 has a 1x10mm slot.

Plate #4 has a 3mm slot all the way across creating 2 halves. Each half is insulated and brought out to the connector. I suspect the acceleration voltage is applied to this plate. I also suspect that a slight voltage differential can be applied between the two halves to center the beam.

Plate #5 has a thin slot with the filament assembly attached to it. Some of the concepts associated with this design is covered in text books. There are two filaments which are thin metal ribbons about 1mm wide and 8mm long. The filaments are brought out to the connector with 3 wires. I assume only one is used at a time and providing two extends the life of the unit.

Under the filament assembly is a mica sheet with a square hole cut in it and a wire laced to form a grid. This wire appears to be a loop where both ends are brought out to the connector. It measures about 12 ohms.

Under the "grid" is a solid plate which I call the reflector. It is insulated and a connection is brought out to the connector.

The filament is explained in text books. what is not explained is the purpose of the wire grid. Perhaps I need a newer (or older) book. My books discuss a magnetic field applied to the filament assembly. Perhaps this loop of wire produces enough field ?

I love a mystery.
chuck

[Chuck Sherwood]

I found some info in the Leak Detection chapter of "Principles of Vacuum Engineering" by Pirani & Yarwood. 1963.

The odd grid structure in the filament assembly is called a Heater Repeller Grid. It is heated to 1000 deg C to prevent the formation of deleterious films caused by the back streaming of DP oils. I suspect I could just bias it the same as the reflector/repeller plate.

The ungrounded disk with a slit in the detector is called a "suppressor". I do not know the exact purpose but based on the name, perhaps it is similar to the suppressor grid in a Pentode vacuum tube? Its purpose would be to force ions that bounce off the collector back to the collector. Therefore I think it should carry a positive bias voltage to repel the ions back to the collector.

I am now at the limit of my references. Further analysis will require some testing.

chuck

[Alex Aitken]

Cool projects.

Chuck, a suppressor is normally about preventing electrons back streaming and it's made the most negative potential in the system.

[Chuck Sherwood]

You are correct that a "suppressor" grid in an electron tube is at cathode potential and much lower voltage than the high positive voltage applied to the "plate".

But there is a very significant difference in the two applications. Electrons are the moving particles in an electron tube where Positively charged ions are the moving particles in a Mass spec.

So I still believe, the suppressor needs to be positively charged compared to the collector plate to repel positive ions back to the collector that are not captured initially. Another supporting argument is that a negatively charged electrode near the collector plate would attract all the positive ions and prevent them from reaching the collector.

That said, I cannot find any references.

chuck

[Alex Aitken]

The electrons come from ions hitting the walls of the cup.
https://en.wikipedia.org/wiki/Faraday_cup suggests a voltage of -170V for the suppressor.
I found one reference talking about a positive voltage, but it also described electrons being repelled by the suppressor grid, which doesn't make sense if it's actually positive.
-90V to -200V seems to be the common range. Some are grids, some are rings.

[John Futter]

suppressor grid potential is directly related to the energy of the ions hitting.
the - 170 volts sounds about right for a mass spec we use up to 700 volts on our ion implanters but a 30kV implanter will only need -400 volts to fully suppress secondary electrons out of a Faraday cup for ion current.
You can make it self biasing by using a transil diode to clip the potential the secondary electrons do the work to generate the potential ie 1.5KE440A.
you will have to find out more from my patent from a few years ago

[Chuck Sherwood]

Wow. Thank you guys for the explanations. It would have taken me a long time to figure that out.

Both sensors I have use an electrometer tube inside the chamber. I don't think the insulators are good enough to pass the signal outside the chamber. Therefore I need to figure out how to use the tube inside.

I have not been able to find very much info on how to make an electrometer with tubes. I have studied the Keithley 601 electrometer but the schematic is incredibly difficult to follow. I resorted to trying this out and it basically works but is hyper sensitive to the Grid #2 bias. I would appreciate any advice available.
Thanks

Chuck

[Chuck Sherwood]

I was hoping to find a circuit that allowed me to use the electrometer detector as it was originally. I came across this circuit from an article titled "A HIGH PERFORMANCE ELECTROMETER AMPLIFIER OF HYBRID DESIGN"
by N. V. Rao »nd C. K. Naiare Technical Physics Division

It eventually mutated to this design. The only significant change I made was improving the FET follower connected to the plate of the electrometer tube. This design worked pretty well when the feedback resistor was 1 gig or less. With larger values, the bandwidth and response time was very slow, like less than 1/10 Hz. So I abandoned a single tube design and started work improving a dual tube design.

[Chuck Sherwood]

After lots of experimenting I came up with this circuit and integrated it into the housing. I had 30 electrometer tubes to choose from and devised a way to match them.

A number of comments on this circuit. The feedback resistor is 400 Gig, so 1 volt output corresponds to 2.5 pA. The bias is set to about 1/2 volt less than cutoff values for the tubes. Screen 1 is adjusted without feedback connected so its plate is about 6 volts measured with a high impedance electrometer (Keithley 610C in my case). Screen 2 is used to zero the output with the feedback connected. This circuit works well, but the bandwidth is still a bit lower than I hoped for. The filaments are lit with a small 1.5V battery that should last for 150 hours.

The tubes and feedback resistor are mounted inside the housing as shown here.

I made a brass enclosure to hold the hand wired circuit board as shown here.

I stimulated the amp by connecting a 470 gig resistor to the "collector" and connecting it to a function generator. This trace show a very low frequency about 1/10 Hz and cranking it up to 1 Hz half way through. As you can see, the amplitude comes down a bit.

The circuit seems quite stable and the zero doesn't seem to move much. That trace was vertical scale at 50 mV which is 1/8 pA per major division. I'm happy with that.

Now I need to work on the emitter end. Probably take a couple more months. Wish me luck.
chuck

[Chuck Sherwood]

I've made a lot of progress on the single sector unit. I believe it is from an old VeeCo MS9. I found a manual for the dual sector unit and it appears to be an MS17. Lucky for me, the ion source and detector are similar so I could figure out how to bias all the elements to make it work.

This is the MS9 sector connected to my Pfeiffer 170 turbo for all my testing. The balloon has some helium in it and is leaked into the chamber by the two brass valves. One has been modified to make it very restrictive. With the turbo at 60% speed, I can leak into the chamber and keep the pressure at 1x-5 Torr.

This is a sweep from 10-300 volts. The large peak at the end is helium centered at 277 volts. This is important because it allows me to calculate the strength of the magnet in the sector. Principles of Vacuum Engineering by Pirani and Yarwood uses this formula.

p=144/H times the square root(VM) where p is the sector radius, H is magnetic field in gauss, V is volts and M is atomic mass.

Helium at M=4, the sector radius is 5cm according the manual. Solving for H=959 gauss. Using this value for H and now solving the same equation for V with M= various values produces

Nitrogen 14, Voltage =79
0xygen m=16 voltage=69
H2O m=18 voltage =61
N2 m=28 voltage =39.6
O2 m=32 voltage = 34.6

This sweep is from 25V to 55 volts
The peak at 41 volts is N2. But you can see O2 on the left side which should occur at 35V. It is unabie to isolate the two and they blur together.

The electrometer is setup with a 100 gig feedback resistor so 1 volt is 10 pA. In these traces that equates to 10 pA per division. The ion current varies with emission current and it is quite easy to saturate the electrometer. I didn't record the emission current but I find 50-100 uA useful.

I can post much more if there is interest.
chuck

[Chuck Sherwood]

I did a couple slow scans to see if that would improve resolution. Short answer is "a little".

This photo(3381) shows a sweep from 25 to 57 volts at 1 second per division. There is a very good peak at 40 volts for N2.
THe O2 peak should be 1 division before the N2 peak. Its there, but blurs in with the large N2 peak.
In this case the ion current is 20 pA per division and the emission current is 100uA.

This picture (3386) shows H2O at 62 volts and N1 at 80 volts.
Oxygen (O1) should be at 70 volts and you can see some slight distortion at that location.

I think this is the limit of a single sector with 5cm radius.
cs

[Chuck Sherwood]

Here are the schematics in case people are interested.

source ps.pdf

(267.33 KiB) Downloaded 18 times

This is the power supply to bias the elements in the ion source. A series of 25V zenor diodes provide various voltages. Guidelines for biasing were determined by studying the VeeCo manual for the MS17.

Six diodes in series (opps schematic shows 5!) provide 150 volts for the anode bias. Anode current is set by the filament temperature.

Another zenor diode on top provides 25 volts to bias the repeller grid positive relative to the anode. This forces the ions out of the ion chamber. This setting is not critical. I found 12-13 volts peaks the ion current.

If you study the pictures of the filament you will see there is a loop around each filament. This forms a Wehnelt cylinder which needs to be biased negative with respect to the filament. The VeeCo manual calls this plate the "object plate". Two more zenor diodes provide an adjustable 0-50 volt negative bias with respect to the filament. Peak ion current occurs at 22 volts but its a soft peak.

The filament is heated with a 2.5 volt transformer. The manual did not use a CT but standard practice is to put the bias on the CT. I found the variac adjustment a bit coarse so I added a small rheostat for a fine tune. 0-50 ohms will vary the anode current (aka emission current) 0-100 uA when the variac is set to 70%.

Another important detail is the repeller grid is heated with DC. The magnetic field from this coil also forces the electrons in the source to spiral and enhances ionization. Removing power to the repeller heater will reduce ion current by half. There are two filaments and this magnetic field needs to be reversed for filament #2.

And the last but very important detail is this whole assembly is isolated from ground the voltage with respect to ground is the acceleration voltage. The acceleration voltage will vary from 0-300 volts so the parts of this circuit will see 500 volts with respect to ground.
cs

[Chuck Sherwood]

The next schematic is the ramp generator.

The timing is a simple relaxation oscillator.
This circuit creates a 0-10 volt ramp that is fed to the next stage.
The dynamic component can be 0-10 volts for a full 0-300 volt swing in the next stage.
An offset is provided so the output can start at any desired voltage. The swing needs to be reduce to avoid saturation.

This allow any desired voltage swing from any starting voltage. Note previous scans started at 25 volts and ended at 50.
Also note that the time period is fixed so reducing the swing in affect also slows down the scan when you think of volts per unit time. For example it is possible to scan 0-300 volts or 0-20 volts in the same time period.

The "trig" output is fed to the scope and the ramp output goes to the next stage.

This stage produces the sweep voltage 0-300 volts by amplifying the ramp input signal by 31. The input power is from a bench PS at this time. Note this circuit is based on a design in "The Art of Electronics"
The purpose of this circuit is to change the voltage on the ion source anode with respect to ground. This requires it to change the voltage on the entire anode assembly with respect to ground. The anode voltage is tapped with a resistor divider network to provide voltages to the focus plates in the ion source. The VeeCo manual set these plates about 20 volts below the anode which is 250 and 270 respectively. I chose a similar voltage at max. There is an adjustment to change the bias on one focus plate, but it does not have much affect. I would add this circuit is MOST temperamental and it is not capable of going to zero. Lowest output voltage is limited by saturation voltage of Q2 and as shown will not go below 8 or 9 volts.

I am tempted to come up with another design but it does seem to work reasonably well.