 |
Ion Chambers |
|
Ion Chambers |
When ionizing radiation (ultra-violet light, x-rays, etc.) pass through a gas, collisions with the gas molecules produces ion pairs, typically charged molecules and free electrons. If an electric field is present, the ions will move apart, each moving in opposite directions along the electric field lines until they encounter the conductors that are producing the electric field.
An ion chamber is an extremely simple device that uses this principle to detect ionizing radiation. The basic chamber is simply a conducting can, usually metal, with a wire electrode at the center, well insulated from the chamber walls. The chamber is most commonly filled with ordinary dry air but other gasses like carbon dioxide or pressurized air can give greater sensitivity. A DC voltage is applied between the outer can and the center electrode to create an electric field that sweeps the ions to the oppositely charged electrodes. Typically, the outer can has most of the potential with respect to ground so that the circuitry is near ground potential. The center wire is held near zero volts and the resulting current in the center wire is measured.
This voltage required to sweep the ions apart and to the center wire or outer can before a significant number of them recombine or stick to a neutral molecule is usually under 100 volts and is often just a few volts. In fact, if the voltage is above a couple of hundred volts, the speeding electrons will produce additional ion pairs called "secondary emissions" giving an enhanced response. Geiger tubes operate at even higher voltages with a special mixture of gasses and exhibit a sudden and very large discharge for each ionizing particle. But below 100 volts only the ions produced by the radiation produce any current. The resulting current is extremely low in most situations and detecting individual x-rays is difficult, especially with ordinary air at atmospheric pressure. Usually the capacitance of the electronics connected to the center wire smoothes the individual pulses too much for detection even when feedback is used to greatly reduce the time constant. These room-pressure chambers therefore respond to the average level of ionizing radiation and do not provide "clicks" like a Geiger counter tube.
Sensitive homemade ion chambers for detecting nuclear radiation are fairly easy to build but the circuitry is tricky and should only be attempted by "seasoned" experimenters - the currents are likely to be well below 1 pA unless there is a serious nuclear war in progress! Special electronics is needed at the front end, typically called an "electrometer" circuit, which produces an output voltage in proportion to the input current. The electrometer must have a very low bias or leakage current to avoid masking the desired signal and the intrinsic impedance of the amplifier must be extremely high. The input impedance of the electrometer may be fairly low, however, using feedback to convert the tiny current into a usable voltage.
Older designs used special electrometer tubes like the 5886 which requires only 10 mA at 1.25 volts for the filament and about 10 volts for the plate. These tubes are great for the experimenter because they are relatively immune to static discharge and they consume about the same amount of power as a typical transistor stage. Some electrometers use vibrating capacitors or mechanical choppers to convert the tiny DC currents into AC before amplification to avoid DC bias and leakage problems. Newer circuits typically use MOSFETs or Electrometer grade JFETs in the front-end. MOSFET op-amps usually contain protection diodes which can be responsible for several picoamperes of leakage at room temperature and a fairly steep increase in leakage as the temperature increases but in some ion chamber applications this extra leakage is tolerable. Non-protected MOSFET front-ends are easily damaged by static electricity and special low-leakage protection diodes are usually added. Low current JFETs like the 2N4220 give respectable performance and the types intended for electrometer applications like the 2N4117A are quite impressive, exhibiting leakage well below 1 pA. They have the added benefit of being significantly less sensitive to static electricity than unprotected MOSFETs. Full ESD precautions must be observed with any of these approaches!
As mentioned earlier, most electrometer circuits use feedback to reduce the effective input impedance and to direct the tiny input current through a very large feedback resistor such that a reasonable voltage is produced at the output. The feedback resistor must be quite large, however. If the input current is 1 pA and the feedback resistor is 100 megohms, the output voltage will only be 100uV. Special resistors measuring in the millions of megohms are available but are usually difficult for the experimenter to obtain (see http://www.ohmite.com/catalog/v_rx1m.html , for example). Lower value resistors may be bootstrapped to increase the effective value by a large factor, perhaps 1000, but that factor would only bring a 10 megohm up to a mere 10,000 megohms. Actually, that is a workable value for most situations but the circuit can require careful adjustment since that amount of bootstrapping is pushing at the limits of practicality. See http://www.wenzel.com/pdffiles/cloud.pdf for an example of bootstrapping.
The following circuits and experiments are not intended to be a hobby project but are for inspiration only. The circuitry is probably not optimal, is easily damaged, and requires patience and technical understanding. For those experimenters that understand how to protect circuits from ESD, understand the schematic without much explanation, and don't mind destroying a few good FETs, similar ion chamber experiments can be quite entertaining. I can think of dozens of science fair projects (see end of article). Working with radioactive materials requires great care. With those disclaimers, here is a narrative of my experiments:
An extremely sensitive circuit was desired that didn't require special resistors and that didn't fail every time a slight ESD mistake was made and the result is the experimental circuit shown below. It uses a 2N4117A as the input amp and another as the feedback resistor. If one studies the tiny curves supplied in the data books and uses a little "extrapolation" and imagination, the leakage of the 2N4117A with the drain and source connected together can be seen to have a slope equivalent to about 75 million megohms! There is unit-to-unit variation and it is necessary for the input JFET to have lower leakage than the feedback JFET so the circuit is not for everyone. (If the input JFET leakage is higher, the output voltage will be very low. Simply swap the JFETs.) The FETs are easily damaged, too, which can lead to frustration when the "best one" gets zapped. The circuit will not be particularly accurate since the actual feedback resistance is not known but the experimental ion chamber is not easily characterized anyway. Despite the circuit's shortcomings, it is extremely sensitive and surprisingly stable. A "good one" might drift only 0.1 fA in a day if ambient conditions are relatively constant. (That corresponds to about 10 mV drift on the output.) The short-term variation is below 1 mV which corresponds to 0.01fA! If the ion chambers really work, this circuit should be able to see the current!
If you have a well-stocked junk box, replacing the "resistor" FET with a very high value resistor will work well. Some sensitivity will be lost since the resistor will probably be below 10 million megohms but the circuit will be quite stable and a more sensitive meter or additional amplifier may be used.
The input FET (the one on the right) and the two transistors form an "error" amplifier that attempts to maintain the drain voltage at 10 volts (set by the resistor divider in the emitter of the NPN). If current flows OUT of the ion chamber causing the gate voltage to rise, the drain will begin to drop and the voltage on the collector of the NPN will go up. This rise will decrease the current in the PNP and thus lower the output voltage. The voltage drop across the first "resistor" FET will increase and more current will flow through it - nearly all of the ion chamber current, in fact. There is sufficient loop gain that the input voltage does not change very much and most of the ion chamber current flows through the feedback FET. The zener in the source of the input FET moves the gate voltage operating point up above ground so that dual polarity supplies are not needed. The output voltage should be a few volts, perhaps 3 or 4, depending upon the relative leakage of the FETs. If the voltage gets too near 6 volts, the sensitivity will drop and the response will become more logarithmic (which might be useful for some applications). If the voltage is too low, the circuit might "bottom out" and loose control. There isn't much that can be done to set the operating point expect swap out FETs! The glass around the FET leads must be VERY clean. Use a good solvent to remove any contaminants.
The first experimental ion chamber was made with a zinc can from a D-cell battery and an old 8-pin glass-to-metal header as seen in the photo below. The two FETs were mounted inside the chamber with the theory that this would eliminate the problem of connecting the extremely high impedance probe to the outside world without creating leakage paths to ground. The problem with the concept is that the transistor bodies and leads compete with the wire for the free ions! Carefully painting the transistor bodies and legs with conformal coating helped but the circuit will not tolerate the coating around the base of the transistor - it is too conductive! (In retrospect, the transistors should reside in their own can with the sense wire passing though a hole into the ion chamber which is what the schematic shows.) The pickup wire should be thin and near the center of the can to keep the capacitance low so the response time is as short as possible. When power is first applied, it can take a very long time, maybe 20 minutes, before the circuit settles out to a steady reading. At first, about 150 volts was applied to the can but it was soon discovered that only a few volts are adequate and a 9 volt battery was used instead. The 15 volt power supply voltage should be fine for most ion chamber sizes. If the voltage is too low, the readings will be low as the ions have time to recombine before being swept to the electrodes.
To test the chamber, a 1.3" diameter disk of radioactive material (a calibration disc from an old Geiger counter) was leaned against the can. The voltage changed a few 10s of millivolts but I quickly lost interest in this chamber when the lid slipped and zapped the FETs. I had already spotted an old mint can at the back of the workbench which I liked better for a chamber. (See pictures below.) An audio connector was added to the center of the 3" dia. tin can and the FETs were mounted directly to the pins. A ring of wire was used for the center electrode. The insides were washed well with a solvent and then dried with a hot air gun before the base was added and tack-soldered.
Little feet were added to the bottom of the can so that I could easily slip my radiation disk underneath without disturbing the chamber. This ion chamber gave gratifying result: the little radiation source gave an output voltage change of about 70 mV which was very large compared to the meter wander of about 2 mV.
At this point in the festivities, I decided to try a crude calibration. Really crude. My calibration reference was a Heathkit Geiger counter which has a meter that reads counts per minute and mR/ hour. The scales are a little suspicious since the CPM scale is an exact power of 10 bigger than the mR/hr scale. (0.3 mR/hr = 300 CPM on the X1 scale, for example.) It is entirely possible that the Geiger tube dimensions were selected to achieve just this result. In the past I had compared this Geiger counter against another "bomb shelter" type and obtained surprisingly close readings - maybe within 10%. The radiation disk gives 1500 CPM when held directly against the Geiger tubes mylar window and 500 CPM when the lid of a mint can is placed in between (to simulate the ion chamber walls). The background radiation measured about 13 CPM. Now here is where the calibration gets a bit "iffy". The disk is large compared to the Geiger tube window but it is small compared to the diameter of the ion chamber. To make a long story short, the ion chamber will read low by some factor - maybe 4. What I think it all means is that the ion chamber gives about 6 mV for a radiation level that causes about 10 CPM in the Heathkit unit corresponding to 0.01 mR/hr. The background radiation should give a reading just above 2 mV which is about how much the readings wander from minute to minute. This calibration may be within one order of magnitude.
But now I am hooked. I want an ion chamber that can easily see the background radiation. To make a bigger chamber, I chose a 4.5" by 4" dia. peanut can (see below). I also decided to move the FETs into their own compartment. For this compartment, I used a steel wheel from the center of an electronic component reel. Any small can would work here, but this piece fits nicely on the bottom of the peanut can and it had a hole perfect for the 8-pin header that I happen to have in large quantities. A hole was drilled in the bottom of the peanut can for the electrode. No insulator was used - just the air gap. The end of the peanut can was sealed with aluminum foil to keep out air currents and electric fields but to allow less energetic or larger particles in. Actually, foil is really too thick and other choices might be better. The chamber is at room pressure so the membrane does not need strength.
The voltage on the outer can was increased to 22.5 volts by using an old 'B' battery on the theory that a larger chamber would need a larger field to sweep out the ions quickly enough.
After power was applied and sufficient settling time went by (about 15 minutes), the meter reading was seen to be significantly more jumpy. The FETs were the same ones from the previous chamber and great care was taken to keep everything clean so I immediately suspected that I was seeing individual ion trails. When I slipped the radiation disk under the aluminum window, the reading climbed to a whopping 1.5 volts! And what a coincidence! The Heathkit gives a count of 1500 CPM for this same source when covered with aluminum foil. (Actually, the foil hardly attenuates the radiation coming from the disk.) So now I have a direct readout of mRads/hr: 1 volt = 1 mR/hr. Unfortunately, I have not yet corrected for the much larger detector area of the new chamber but it works out to be nearly 10! So the sensitivity of the new chamber is 1 volt per 0.1 mR/hr which is pretty sensitive! The radioactive element from a smoke detector was held up to the Geiger counter and the count soared to about 22,000 CPM but placing a piece of aluminum foil in between dropped the count to 200 CPM. The ion chamber gave a reading of 200 mV which is in perfect agreement. But I didn't expect agreement since this source is small relative to the Geiger tube, also. The mylar window on the Geiger tube blocks the alpha particles some and this may account for the agreement. These calibrations are really coarse! By using the Geiger counter to measure the background radiation it was determined that the ion chamber should be indicating 13 mV but since the zero setting is arbitrary, it was hard to confirm this level. Reversing the polarity on the outer can caused a shift of about 30 mV (after several minutes of settling) which is about what is expected if the background is near 15 CPM (plus 15 to minus 15 is a total of 30). The experimental setup is shown below:
More Experiments:
I tried a long chamber made from a section of air duct with a thin wire stretched between two Styrofoam plugs:
|
The internal wire was soldered to both ends, the plugs taped in place and the tube stretched to tighten the wire. |
| The finished detector was pretty awful! First of all, the electrostatic shielding on the ends was inadequate, causing a huge 60 Hz signal. And, the wire would vibrate with the slightest bump causing wild swings in the output. Maybe this thing might be the start of a seismometer, but it stinks an an ion chamber! |
|
But then:
Another chamber was constructed with large cookie tins similar to the peanut can design above. The performance of this much larger chamber was excellent. A single Coleman lantern mantle nearly "pegged" the output (after amplification), with the relatively low value 100,000 megohm feedback resistor. The background radiation gives about 4 mV (400 mV after amplification) which corresponds to 40 fA current. (Many CMOS opamps have input currents well below 40 fA.) Even though the circuit was given a very low frequency response to reduce 60 Hz response, the meter jitters in response to individual ion trails. (The superior shielding of the cookie tins would probably allow for a faster response, if desired, but watch out for circuit instabilities.)
| This tin measures about 10.5" across and about 6" tall (a "regular" height tin should work as well). The center portion of the lid was cut out with scissors to make a frame to hold the aluminum foil window. The circuit is housed in a smaller cookie tin tack-soldered to the bottom. Connections are made via a 5-pin audio connector. |  |
 |
 |
The electrode is a 5" dia. wire ring mounted to a Teflon standoff and a short piece of stiff telephone wire connects the electrode to the circuitry. The wire passes through a large hole to reduce the chances for leakage currents. The circuitry is a modified version of the first schematic featuring a resistor for the feedback and an op-amp for boosting the output signal. The transistor circuit was also modified to increase the loop gain and improve the stability (see ckt. desc. below). | |
| A word of caution: the metal sure looks like ground and a person (um, like me) might start soldering components that go to ground to it. The can will actually be connected to +45 volts or more so the "ground" connections are made above the metal. The only components that connect to the can in the photo are a large yellow cap and a couple of white caps used to support my elevated ground buss. |  |
|
The new circuit includes several improvements. The feedback FET is replaced with a Victoreen 100,000 megohm resistor which is the long glass tube in the photo. A zener diode was added to the emitter of the 2N4401 to increase the loop gain and a .01 uF Miller capacitor was added to reduce the amplifier frequency response (for stability and to reduce 60 Hz gain). An op-amp (OP-07) was added to boost the output by a factor of 100. The "zero" pot is used to set the output to a few volts since the OP-07 cannot swing below 1 or 2 volts out without a negative supply. This pot must be able to be adjusted to the gate voltage and with some FETs the voltage may not go low enough. The symptom will be a high op-amp output voltage. If so, just lower the 10k resistor or add a 1k above the pot. An additional zero pot for the meter could be added as in the first schematic to get a near-zero reading for the background radiation, if desired.
Notice that the drain resistor was reduced to 125k. This value was experimentally determined by finding the drain current that gives the 2N4117A a near-zero temperature coefficient. The test circuit is simple: connect a sensitive current meter from +10 volts to the drain, ground the gate, and connect the source to ground through a 500k pot. The current is observed at room temperature then the FET is warmed and the current change is noted. The pot is adjusted until little or no change occurs. I heated the FET by touching a warm PTC to the can - probably reaching about 65 degrees Celsius and the final current change was below the current change caused by a 100 uV gate voltage change. (Corresponds to less than 1 fA ion chamber current for 40 degrees.) Room temperature may vary by +-4 degrees which would correspond to a wander of 0.1 fA which is well below the 40 fA background current from the chamber. The bias current that gave this wonderful temp-co was 40uA and since the drain resistor will have 5 volts across it, the desired resistor value is 5/40 uA = 125k. "Your results may vary." Actually, the FETs are surprisingly stable at all currents and the whole procedure may be unnecessary.
A 22.5 volt battery was insufficient to capture all of the ions but two batteries (45 volts) seemed to do the job - in other words, higher voltage did not result in a higher reading. Higher voltages may be desired for observing individual events, however, since the ions will be swept to the electrode faster.
Parts Notes:
Send me an email if you cannot find some of these parts easily. I probably can't help much with the high value resistor but I have most of the other parts in plentiful supply.
Additional Ideas I May Try:
A commercial geiger tube is pretty hard to beat for general radiation monitoring but the simplicity and versatility of a homemade ion chamber that requires no special gasses or pressures makes it an attractive alternative for many experiments.
Here is an interesting email from a reader:
Hello Charles,
I really enjoyed you piece on ion chambers. I built an AC output chamber from a soft drink can a few years back and got an excellent signal from alpha particles. It made a great radon detector and gave several counts per minute. I cut up a second can to to make a piece of sheet aluminum and used it for the charged electrode by lining the inside of the ion chamber with a thin layer of poly foam for insulation and then placing the sheet aluminum on the inside of the foam. This allowed me to run the outside of the chamber at ground (safety feature) but also allowed any leakage current to flow to ground. That allowed me to use air and PCB material (no copper) as the other insulators, since the center electrode was near ground potential. I used a piece of small gauge nichrome wire as the center electrode and suspended it from a piece of monofilament line from three holes punched in the open end on the chamber (electronics at the other end). Therefore, I could open the chamber and insert test samples in it easily. The monofilament didn't cause much obstruction. Of course, you have to discharge the 400V polarizing supply first! The wire would vibrate and the chamber made a very sensitive capacitance mic. vibration detector. I could test the chamber by gently tapping the table and watching the damped sine wave! Not a practical portable tool. I used a 100,000 meg resistor from my junk box, which, along with a 2pF feedback resistor, gave me a time constant of about 200 ms. Since the time to sweep the ions out is about 10-50ms (depends on chamber voltage and geometry), that worked out great. I used a 2N4416 jfet as the amp, since an AC chamber is much less sensitive to leakage (the second amp is AC coupled to block the DC offset). Your comment about the difficulty in obtaining high value resistors stuck with me. I was recently working on another project requiring 100+ meg resistors and considering the problem again. I was thinking of using a current source in a feedback loop to simulate a resistor at low frequencies and ran a series of experiments, both thought and physical. A pin photodiode controlled by an LED works but tends to be noisy and very temperature sensitive. Also, even good photodiodes have a substantial capacitance and a leakage current floor of 0.5 nA or so. Very good photodiodes are expensive and we are back to the original problem. I tried a high frequency (low collector capacitance) bipolar transistor, but the noise was much too high. Then I started considering a miniature ionization chamber built using an alpha source from a smoke detector to supply the ion pairs. You could control the "resistance" by shielding the source with alum. foil with a pin hole in it to let through only a small percentage of the alpha particles. But that struck me as cheating, although I am sure it would work. If you don't apply voltage, the ion pairs would all recombine and there would be no current. The number of ion pairs that are separated and reach the electrodes would be dependent on the applied voltage. Therefore, a "resistor". But was there another, cheaper and easier way? Finally, it occurred to me that an old style vacuum phototube might work. Light strikes a low work function material and knocks off electrons, which are swept to the other electrode by the applied voltage and become current. It would only work as a unipolar device and would tend to be a constant current device at higher voltages, but I liked the idea. Unfortunately, those devices are collectors items these days and would be very expensive. It finally occurred to me that since I am interested in the low nanoamp to picoamp region, it wouldn't take much in the way of electrode surface to do the job. So I found an NE2 neon lamp in the junk box and connected it in in series with my DVM set to the 200mV scale and two 9V batteries. ( I don't know if you have used the DVM trick, but if you push 1nA through a 10meg resistor you get a 10 mV drop, which shows up on the display. So if the meter reads down to 0.1mV, that is 10pA. Instant auto-zero picoamp meter with a 10meg shunt.) And lo, current flowed when the bench fluorescent light was shining on the bulb! At first everything was very drifty and unstable. Then I washed the bulb in isopropyl and distilled water and heat dried it. No more noise and I had a controllable current source from about 500pA down to 10pA (measurement limit). For a constant illumination, the current varied with the square of the voltage applied across the tube wires. Handy if you want high dynamic range. I tried a flashlight with a krypton bulb and it worked as the light source. I tried a yellow led and it worked. I tried a blue led in one of those keychain lights and it worked well. ALL RIGHT! I haven't designed his gadget into anything yet, but it has to work given what I saw on the bench. The neatest part is that you can vary the "resistance" by changing the light striking the tube. Or use a photodiode feedback loop to keep it very constant. Probably a constant current drive on the led would work for most cases. Keeping the tube above room temperature keeps surface moisture from being a problem, but that comes free with the heat generated by the light source. The voltage isolation is a good as the glass that the tube is made of. Just don't get any gunk between the wires where they come out of the tube. And it is CHEAP! Maybe a buck if you buy he parts new in singles. BTW, infrared from a IR remote handheld did NOT work. I think that is because the photon energy isn't high enough(950nm) to exceed the work function of whatever material the tube electrodes are made from, but I haven't proved that. Anyway, there it is. I plan to try it out as the current return element in an ion chamber as soon as I can get around to it. It is just too cute to resist playing with. Oh, I hooked up the NE2 in place of a photodiode (10V bias) in a very sensitive amplifier to look for high frequency noise. When I illuminated it I saw DC current but NO increase in the (band limited) AC noise. Cool. Also, think of the possibilities of using a light pulse to remove charge and thereby converting the ion chamber into an "ion to frequency" converter. Should work. Just apply a fixed width, constant current, pulse the led whenever the amplifier voltage exceeds some set point and suck out a hunk of charge. Calibration should be interesting. Maybe a step voltage across a small value capacitor? If you use the thing before I do, let me know how it works out.
Craig Taylor
Nerd Experimenter
