One fine point ref. RF Ion sources, that I have been unable to clearly define in my mind is "Push vs Pull" for the initial ion bottle, beam extraction? Clearly both methods will work, since there are working ion sources out there which use both methods. In Doug's microwave ion source, a positive HV is used to push the ions, similar in principal to the pin electrode at the end of the classic quartz RF ion source bottle. This is often in combination will a pull electrode downsteam. With this source the push electrode also serves the purpose of making it more simple to build, so that it makes perfect sense.
I have found quite a few good designs for ion sources, which do not push at all, and instead they only pull the ion beam with a negative HV electrode at the beam end of the bottle. I have posted an example below, of this design approach, that includes some good data.
My specific question is: Does it really matter, or does this issue simply boil down to preferance? With no big impact on the potential output beam current, etc. of a given ion source design. Also ref. my copy of the microwave source: Since I already have a nice vacuum assembly which already has a VCR gas inlet fitting built in, I do not need the end electrode for gas introduction, so I was thinking of essentially using an inverted quartz test tube as the ion source bottle, with a neg. HV pull extraction system.
Any thoughts on push vs pull? Thanks in adavance....
Push vs Pull Ion Source Extraction?
3 posts
• Page 1 of 1
Push vs Pull Ion Source Extraction?
by lutzhoffman » Fri Aug 06, 2010 3:24 pm
- Attachments
-
probelessrfionsrc..pdf- (672.25 KiB) Downloaded 371 times
- lutzhoffman
- Posts: 31
- Joined: Thu Jul 15, 2010 1:57 am
Re: Push vs Pull Ion Source Extraction?
by Doug Coulter » Sun Aug 08, 2010 9:37 am
The real issue is gradient, here. Ions can't tell if they are being pushed or pulled, just that "this way is downhill"..So they go that way. In a push or pull only situation, the other end the the field gradient is simply ground. Or in some designs, one simply allows the ions to drift thermally into a field a little away from where they were produced, such as one in the paper you put up in the parent post. (Lutz, there are two periods in the filename, making it hard to open with most software can you rename it so everyone else doesn't have to?). In other words, there is nothing magic about "ground" as far as charged particles go -- they don't know "where they are", just "which way is down". Depending on what other charges are around, ground could either be "down" or "up" so in some ways the issue of "push or pull" just isn't relevant; it's all slope, not altitude.
Ions, particularly hot ones, are among the most chemically reactive substances in existence. In general, that fact means most designs have a short life, as the electrodes and other materials tend to get into reactions with the ions. Not too many things easily resist this effect (even quartz can be reduced to Si by hot H ions), so most designs consider life and replace-ability of the parts.
A consideration is what you want the ion source output to look like to whatever you're putting them into downstream. You might want a focused beam, a diffuse flow, acceleration to some certain energy, a particular state of ionization -- only the end user can specify those things. Depending on the application, if you want one thing but get another, the job isn't done yet.
The glowing cloud in most ion sources is still net neutral -- the electrons are still there, but in general everything is so hot they can't re-stick onto the atoms -- but due to mutual attraction, they stay nearby. Let's assume we want positive ions, the normal case. In that case, we want to separate the ions from the electrons, which takes some work to overcome the attraction of unlike charges. So there has to be a field gradient to accomplish this. Now, many folks make the incorrect assumption when "putting on a field" that the resulting field is the one they calculated by using the volts and dimensions as the inputs.
This is not the case. As soon as one separates the electrons from the ions, they themselves create an opposite polarity field which tends to cancel the deliberately applied one. This also means that any oversimple assumptions about net ion velocity out of the source tend to be "optimistic" at best. In fact, in most designs, the resulting velocity is much lower than would be predicted as well as the amount delivered, due to this. It is all too easy to have a situation where the ion current is completely limited by this effect, particularly in an RF ion source, which tends to have plenty of ions available to be extracted -- many more than you can force out of there with a decently small applied field.
Once free of any applied fields, the ions tend to "Coulomb explode" due to the field they create themselves. This is directly analogous to space charge limits in focusing first worked over for electron devices -- only with ions, they are going a lot slower for the same net field, and the situation is more space-charge limited by a large factor (generally taken as the square root of the mass ratio between the ion and an electron, assuming charge is the same number of units).
Without some attention to this, the results of the Coulomb explosion brings the ions in contact with things that will easily donate electrons to them and render them neutral again. This is generally considered undesirable; why put in all that energy to strip electrons off, then just put them back?
To overcome this, ions are most usually accelerated somewhat, and focused into a beam of some kind. The normal electrostatic lenses are used, as well as magnetic fields in some cases. This all works out just like the (more available information) for electron devices, with appropriate scaling for the charge/mass ratio differences. While a focused beam of ions is still trying to Coulomb explode, the fact that they are now going fast means they might get to where you want before that happens too severely. That process can't be slowed unless you are making them heavier due to relativistic effects, but you can get them where you want them quicker to ameliorate the issue.
Nearly without exception, accelerating fields tend to act like a convex lens, and decelerating fields like a concave one. As with light, the lens dimensions and shape determine a focal point, which is the only place the photons or ions would come to a point -- after crossing through focus, they expand again. Therefore, just as with light, one needs successive lenses to either collimate the beam, or to keep refocusing it before it hits a wall or limiting aperture. Only photons don't repel one another, but charged particles do, so there are more issues, no focus is ever perfect, and there is a space charge limit on how many you can push through a pipe vs how fast they are going -- and for heavy, slow moving positive ions, the limits are much worse than for electrons, where for example, to get a decent sized spot on a CRT screen at 5 ma or so, it takes quite a lot of voltage (usually tens of kV), both to get them to the screen before the expansion has time to take place, and to put enough energy into such a low current to make the phosphor respond decently. These side effects are on the order of 42 times worse for protons than electrons, and about 60 times worse for deuterons. Thus, in general to do e-beam-like things with ions, one needs some pretty serious fields by comparison. All the same tricks work (accounting for polarity) but it just takes more to get an ion beam under control than an E beam. This implies that fairly high voltages are needed (to create a somewhat smaller NET field) for ions if you want any serious amount of current in a controlled beam. At some point, the higher voltages cause their own problems, so many people will simply split the supply to so voltage at any point off ground can be half as much. That is the basic main reason you'd ever do a push-pull design -- it's easier to handle the needed supply voltages.
There is another effect that applies in the case of fusors. In general, you don't want a beam of ions, but a more or less diffuse, uniform cloud of them (which is what they try to do themselves anyway due to repelling one another). To get this to happen more immediately, the push-pull design is superior. As the ions pass beyond the last, negative electrode and into the tank, the field they see is now a decelerating one back towards the ion source (assuming no other fields are present at the time, but all fields add linearly at any rate). As stated above, decelerating fields tend to act like concave lenses, and here we have the help of the ions themselves pushing on one another -- and the more we slow them down, the better that works as well in producing a diffuse cloud. So that is a possible advantage of a push-pull design, if that is what you wanted -- "it depends" is altogether too-common response to many questions, but often the only true one. Stinks, but there you are
So far, I have only found two half-decent sources on E field lensing, and really none on magnetic optics for charged particles.
The sources are Terman's Radio Engineer's Handbook from WWII, and "Building Scientific Apparatus" more recently. The only source I have seen for magnetic optics that even gives rough order of magnitude is an article on the old "amateur science" SciAm disk about making an electron microscope -- no numbers are given on field strengths or shapes there, but at least since they use an electromagnet, you can work the math and figure some things out from that. It seems mag lensing is a guild secret maintained by the electron microscope community.
Anybody who knows me knows how much I'd love to have and reveal those secrets. "He who would deny you knowledge considers himself your master" and I accept no master over me.
Here is some of Terman's stuff, I've not yet found that book scanned in anywhere on the 'net, but would love to have a full digital copy. Luckily they are still around used, often not too expensive, but most are also falling apart because they couldn't use good paper (war shortages) and because these things really got read.
Click pics to expand.
To learn a little of this at the gut level, we constructed a fairly simple E beam device, which will be the subject of another post. But the experience can be summarized in few words. One is that slow moving electrons are affected by even tiny stray magnetic fields, so you get them moving quickly and early if you want many to get downstream, rather than spiral around some field line and hit the wall. And it takes volts to get any current out of it due to that and the self repulsion effects.
The ion source here -- has some issues, it's far from perfect as is. So far, even with +/- 2.5kV or so on it, most of the ions are not extracted or are lost by hitting the sides of the extraction electrode. The reasons are the stray magnetic field from the cyclotron magnets, which is expanding going towards the tank, and the all too long distance the ions have to travel to get there. At the lengths and voltages I've got, the ion current is still increasing when the voltage control hits the top. At my current speeds and feeds, it seems about half the ions are hitting the walls of the ion source -- to get ~2.6 ma out, I'm putting in about 5 ma. That measurement is fairly indirect, however.
I am going by how much the main fusor grid will draw when the vacuum is good enough that it doesn't draw any current with the ion source off.
A better tank entry design would help a lot, just get the thing closer. Also, putting the gas inlet tube (which in this case is also the push electrode) farther in towards the ion cloud, as far as it can go without messing up the RF/magnetic fields, would help increase the gradient. I am considering a new design for this, but it's not top priority at the moment - while it doesn't work as well as it might, it basically works well enough for my current needs (pun intended
) . Sometimes the good enough is the enemy of the perfect.
Ions, particularly hot ones, are among the most chemically reactive substances in existence. In general, that fact means most designs have a short life, as the electrodes and other materials tend to get into reactions with the ions. Not too many things easily resist this effect (even quartz can be reduced to Si by hot H ions), so most designs consider life and replace-ability of the parts.
A consideration is what you want the ion source output to look like to whatever you're putting them into downstream. You might want a focused beam, a diffuse flow, acceleration to some certain energy, a particular state of ionization -- only the end user can specify those things. Depending on the application, if you want one thing but get another, the job isn't done yet.
The glowing cloud in most ion sources is still net neutral -- the electrons are still there, but in general everything is so hot they can't re-stick onto the atoms -- but due to mutual attraction, they stay nearby. Let's assume we want positive ions, the normal case. In that case, we want to separate the ions from the electrons, which takes some work to overcome the attraction of unlike charges. So there has to be a field gradient to accomplish this. Now, many folks make the incorrect assumption when "putting on a field" that the resulting field is the one they calculated by using the volts and dimensions as the inputs.
This is not the case. As soon as one separates the electrons from the ions, they themselves create an opposite polarity field which tends to cancel the deliberately applied one. This also means that any oversimple assumptions about net ion velocity out of the source tend to be "optimistic" at best. In fact, in most designs, the resulting velocity is much lower than would be predicted as well as the amount delivered, due to this. It is all too easy to have a situation where the ion current is completely limited by this effect, particularly in an RF ion source, which tends to have plenty of ions available to be extracted -- many more than you can force out of there with a decently small applied field.
Once free of any applied fields, the ions tend to "Coulomb explode" due to the field they create themselves. This is directly analogous to space charge limits in focusing first worked over for electron devices -- only with ions, they are going a lot slower for the same net field, and the situation is more space-charge limited by a large factor (generally taken as the square root of the mass ratio between the ion and an electron, assuming charge is the same number of units).
Without some attention to this, the results of the Coulomb explosion brings the ions in contact with things that will easily donate electrons to them and render them neutral again. This is generally considered undesirable; why put in all that energy to strip electrons off, then just put them back?
To overcome this, ions are most usually accelerated somewhat, and focused into a beam of some kind. The normal electrostatic lenses are used, as well as magnetic fields in some cases. This all works out just like the (more available information) for electron devices, with appropriate scaling for the charge/mass ratio differences. While a focused beam of ions is still trying to Coulomb explode, the fact that they are now going fast means they might get to where you want before that happens too severely. That process can't be slowed unless you are making them heavier due to relativistic effects, but you can get them where you want them quicker to ameliorate the issue.
Nearly without exception, accelerating fields tend to act like a convex lens, and decelerating fields like a concave one. As with light, the lens dimensions and shape determine a focal point, which is the only place the photons or ions would come to a point -- after crossing through focus, they expand again. Therefore, just as with light, one needs successive lenses to either collimate the beam, or to keep refocusing it before it hits a wall or limiting aperture. Only photons don't repel one another, but charged particles do, so there are more issues, no focus is ever perfect, and there is a space charge limit on how many you can push through a pipe vs how fast they are going -- and for heavy, slow moving positive ions, the limits are much worse than for electrons, where for example, to get a decent sized spot on a CRT screen at 5 ma or so, it takes quite a lot of voltage (usually tens of kV), both to get them to the screen before the expansion has time to take place, and to put enough energy into such a low current to make the phosphor respond decently. These side effects are on the order of 42 times worse for protons than electrons, and about 60 times worse for deuterons. Thus, in general to do e-beam-like things with ions, one needs some pretty serious fields by comparison. All the same tricks work (accounting for polarity) but it just takes more to get an ion beam under control than an E beam. This implies that fairly high voltages are needed (to create a somewhat smaller NET field) for ions if you want any serious amount of current in a controlled beam. At some point, the higher voltages cause their own problems, so many people will simply split the supply to so voltage at any point off ground can be half as much. That is the basic main reason you'd ever do a push-pull design -- it's easier to handle the needed supply voltages.
There is another effect that applies in the case of fusors. In general, you don't want a beam of ions, but a more or less diffuse, uniform cloud of them (which is what they try to do themselves anyway due to repelling one another). To get this to happen more immediately, the push-pull design is superior. As the ions pass beyond the last, negative electrode and into the tank, the field they see is now a decelerating one back towards the ion source (assuming no other fields are present at the time, but all fields add linearly at any rate). As stated above, decelerating fields tend to act like concave lenses, and here we have the help of the ions themselves pushing on one another -- and the more we slow them down, the better that works as well in producing a diffuse cloud. So that is a possible advantage of a push-pull design, if that is what you wanted -- "it depends" is altogether too-common response to many questions, but often the only true one. Stinks, but there you are
So far, I have only found two half-decent sources on E field lensing, and really none on magnetic optics for charged particles.
The sources are Terman's Radio Engineer's Handbook from WWII, and "Building Scientific Apparatus" more recently. The only source I have seen for magnetic optics that even gives rough order of magnitude is an article on the old "amateur science" SciAm disk about making an electron microscope -- no numbers are given on field strengths or shapes there, but at least since they use an electromagnet, you can work the math and figure some things out from that. It seems mag lensing is a guild secret maintained by the electron microscope community.
Anybody who knows me knows how much I'd love to have and reveal those secrets. "He who would deny you knowledge considers himself your master" and I accept no master over me.
Here is some of Terman's stuff, I've not yet found that book scanned in anywhere on the 'net, but would love to have a full digital copy. Luckily they are still around used, often not too expensive, but most are also falling apart because they couldn't use good paper (war shortages) and because these things really got read.
Click pics to expand.
- space charge limits and effects-1
- Space charge limit - 2
To learn a little of this at the gut level, we constructed a fairly simple E beam device, which will be the subject of another post. But the experience can be summarized in few words. One is that slow moving electrons are affected by even tiny stray magnetic fields, so you get them moving quickly and early if you want many to get downstream, rather than spiral around some field line and hit the wall. And it takes volts to get any current out of it due to that and the self repulsion effects.
The ion source here -- has some issues, it's far from perfect as is. So far, even with +/- 2.5kV or so on it, most of the ions are not extracted or are lost by hitting the sides of the extraction electrode. The reasons are the stray magnetic field from the cyclotron magnets, which is expanding going towards the tank, and the all too long distance the ions have to travel to get there. At the lengths and voltages I've got, the ion current is still increasing when the voltage control hits the top. At my current speeds and feeds, it seems about half the ions are hitting the walls of the ion source -- to get ~2.6 ma out, I'm putting in about 5 ma. That measurement is fairly indirect, however.
I am going by how much the main fusor grid will draw when the vacuum is good enough that it doesn't draw any current with the ion source off.
A better tank entry design would help a lot, just get the thing closer. Also, putting the gas inlet tube (which in this case is also the push electrode) farther in towards the ion cloud, as far as it can go without messing up the RF/magnetic fields, would help increase the gradient. I am considering a new design for this, but it's not top priority at the moment - while it doesn't work as well as it might, it basically works well enough for my current needs (pun intended
) . Sometimes the good enough is the enemy of the perfect.Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
-
Doug Coulter - Posts: 3515
- Joined: Wed Jul 14, 2010 7:05 pm
- Location: Floyd county, VA, USA
Re: Push vs Pull Ion Source Extraction?
by lutzhoffman » Mon Aug 09, 2010 3:14 pm
Hello:
I have some serious information to digest now, and in this case I am sure that I will be "chewing my cud", several times. Getting over 2ma out, or even 1 ma for that matter, (which is my top figure needed), is a hell of an achievment. I realize that its always different from the inside looking out, but from the outside looking in, you did a hell of a job on the microwave source, thanks for the additional info : )
I have some serious information to digest now, and in this case I am sure that I will be "chewing my cud", several times. Getting over 2ma out, or even 1 ma for that matter, (which is my top figure needed), is a hell of an achievment. I realize that its always different from the inside looking out, but from the outside looking in, you did a hell of a job on the microwave source, thanks for the additional info : )
- lutzhoffman
- Posts: 31
- Joined: Thu Jul 15, 2010 1:57 am
3 posts
• Page 1 of 1
Who is online
Users browsing this forum: No registered users and 4 guests