Needing more than a spark test?

It is possible to model the output pulse, and it's driving impedance, in Spice, for a PMT, and settle the issue of pulse extraction design. Come to that, it is possible to model the entire dynode set in a PMT as a series of current-controlled current generators. I will know the truth of that aspect of the design.

I am not convinced about the thinner scintillator. They are glass-clear all the way through. An incoming event will produce a flash when it penetrates deep enough. I don't know much of the detail of the scintillation mechanism, but if a flash was caused, whether crystal was thick or thin, the now visible light will make it to the output surface.

The motivation to physically cut a thinner crystal out of a thicker chunk is not likely over saving money - unless the aim was to make several scintillators out of one.

The mind is also thinking on Si(PM). I don't have nearly as much information on circuits for these.
 
Last edited:
  • Like
Reactions: rwm
I believe the visible-light photons generated by an x-ray photon are emitted at random angles. Given the usual scintillator/detector geometry, it won't be possible to collect photons emitted away from the detector (if they don't exceed the angle needed for TIR). So you are inherently limited to the ones that either directly make it to the detector, or are reflected back into it (again, if their direction in the crystal results in TIR from the sides of the crystal). A thinner detector could have improved geometrical aspects, basically a greatly-reduced "side reflection" component. A specialized form of ray tracing program for optics might shed some light on this question. I have a copy of Goptical, which is a package for ray-tracing optical systems, but haven't used it enough (not any, in fact) to know if it could be used for this. Maybe a user-defined light source A.K.A. a "scintillator" is possible.

Speaking of optics, I had wondered if it might be possible to collect more light (which appears to improve energy resolution) by placing the scintillator at one of the foci of an elliptical mirror. Place the detector at the other focal point and (perhaps) enjoy better energy resolution.
 
homebrewed said:
I believe the visible-light photons generated by an x-ray photon are emitted at random angles. Given the usual scintillator/detector geometry, it won't be possible to collect photons emitted away from the detector (if they don't exceed the angle needed for TIR). So you are inherently limited to the ones that either directly make it to the detector, or are reflected back into it (again, if their direction in the crystal results in TIR from the sides of the crystal). A thinner detector could have improved geometrical aspects, basically a greatly-reduced "side reflection" component. A specialized form of ray tracing program for optics might shed some light on this question. I have a copy of Goptical, which is a package for ray-tracing optical systems, but haven't used it enough (not any, in fact) to know if it could be used for this. Maybe a user-defined light source A.K.A. a "scintillator" is possible.
Click to expand...
Hi @homebrewed Re: The first part.
Consider first a thin crystal. I don't know if one X-ray photon slamming into a crystal atom (molecule?) will beat it up enough to yield more than one photon of visible, but let us say for the moment it is only one, in a random direction.
The fraction that makes it to the photocathode are those from a wide solid angle, almost a hemisphere.

Now imagine add a small thickness. The photons from the layer next to the photocathode will still behave as before, but this time augmented by another layer's worth, able to collect a slightly smaller solid angle fraction of the slice volume, and those will be slightly attenuated by the light loss through the first slice.
.. and so on.
We might get up a calculus integration to provide an expression, and discover a maximum, if there is one.
Speaking of optics, I had wondered if it might be possible to collect more light (which appears to improve energy resolution) by placing the scintillator at one of the foci of an elliptical mirror. Place the detector at the other focal point and (perhaps) enjoy better energy resolution.
Click to expand...
Once we have visible light happening in a crystal volume, I can see value in using metal reflective sides, to re-direct light back into the crystal. Consider also the X-Rays coming at the crystal input. Even if they included X-rays that were made from a gamma slamming into aluminum in the test sample, the X-Rays will go right through a thin aluminum into the crystal, but the scintillator light going the "wrong" direction will get reflected back into through the crystal, to end up at the photocathode.

Making reflective optics of the kind you suggest would be worthwhile if the size of the photocathode was a point region focus, but instead, it is a large disc. Not so if it is a small Si(PM) diode. In that case yes, make one surface curved - or use a lens - or use a plastic light pipe

Covering the scintillator with metal (which they do) is an approximation to an "integrating sphere" where all that makes it into the inside can only come out one hole.
https://en.wikipedia.org/wiki/Integrating_sphere

Doing what you suggest could be a great idea to gather wayward photons heading out of the crystal in unfortunate directions, back onto a small area silicon at the focus. Goptical may be able to show (say) 8 rays, starting from a source inside a crystal volume, going in various directions, to test the reflector. Aluminium is, in this case, a frequency selective surface.

You happen upon what I do - i.e. "The dishes"!

IMAG0790a.jpg

Ignore the (expensive) little black random circles, which are there for photogrammetric surface measurement.
The feed at the prime focus is of a frequency that can pass right through the smaller dish. The feed down in the middle of the main dish at the Cassegrain focus is of a frequency that can only reflect off the smaller dish.
Inside the smaller optic dish is dichroic frequency selective surface tchnology - a microwave array. This arrangement is somewhat equivalent to X-ray photons going through aluminium, but visible light unable to.

We have one answer to the question about why thin crystals.
.. From Lucian's colleague Stanislov.
" - - - - - - -
[9/6, 11:47] Stanislav: To cut high energy background and minimize light losses
[9/6, 11:48] Stanislav: At low energies, crystal has no active volume but an active surface

And this is all! For XRF we need a thin and whide. 25x5mm for me Is a good compromise but I know that Advantech UK makes also 1mm thick cristalls with Be window.
" - - - - - - - -

My thought on that is the compromise can be shifted considerable by limiting the high background energy possibilities into the test sample region - say with a sheet of lead under the test sample.

I think Stan means that at low energies, the X-Rays don't penetrate the sintillator crystal, but only work at it's surface.
Huh?? I dunno about that!

Once the scintillator has made light, a very large fraction will travel on through it. They are glass-clear!
A pause now - while I scramble some hardware together.
 
" Consider first a thin crystal. I don't know if one X-ray photon slamming into a crystal atom (molecule?) will beat it up enough to yield more than one photon of visible, "
My understanding is that an xray absorption event will cause a chain of photoelectric interactions with a shower of light photons of various energies. The sum of the photon energy is equal to the energy of the incident xray.
The only problem I see with putting a reflector on the front of the crystal is that it may block some of the very low energies we are looking for. I suppose you could try very thin aluminum. It would be a good experiment to try with and without this thin reflector to see how the efficiency changes.
"I think Stan means that at low energies, the X-Rays don't penetrate the sintillator crystal, but only work at it's surface.
Huh?? I dunno about that! "
I think that is what he means also. That is a great website BTW! Lots of good stuff.
I can't wait to see some hardware!
Robert
 
RWM is correct in that a single x-ray photon interacts with multiple emitter atoms in a scintillator. The number of atoms varies in proportion with the energy of the x-ray. As a result every event results in an angularly-distributed shower of visible-light photons, VLPs (my acronym to reduce the amount of tyeping oops typing corrections I have to do :). If only a random portion of them wind up hitting the detector, the resultant pulse has a random variation. I think this is one of the reasons for the relatively low resolution of scintillators vs solid-state detectors. For solid state detectors, once the hole-electron pairs are generated they are swept up by the internal electric field, so the notion of "direction" affecting detector efficiency doesn't exist.

I like the idea of coating the scintillator with a thin Al layer. More elegant than the ellipsoid scheme, and the aluminum wouldn't need to be very thick so it probably won't absorb the low-energy photons to any great extent. Unfortunately, Al-coated scintillators don't seem to be all that available, maybe @graham-xrf has found a source? If not, there's a relatively inexpensive answer -- send your crystal to a telescope-mirror aluminizing service company. I found one that serves amateur astronomers and will coat one for $69 (USD). That is, if the scintillator is less than 6 inches in diameter.

Alternatively, awhile back I came up with a scheme to edit integrated circuits using electroless silver -- the old mirror-silvering process -- and it was pretty easy to do. You need silver nitrate, ammonium hydroxide, distilled water and Rochelle Salt. The silver film is pretty thin and is highly reflective. You just need to practice some basic safe handling methods -- PPE's, and be sure to discard any left-over silvering solution once you're done. If allowed to dry it may form silver fulminate, which is a sensitive explosive. OK, nothing's free :laughing: . The stock solutions -- silver nitrate in distilled water and rochelle salt in distilled water -- are stable. Silver nitrate will stain your hands so you will know soon enough how good your handling protocol is.
 
  • Like
Reactions: rwm
Biasing SiPMs.

The KETEK data sheet for their 3x3 device says the breakdown voltage, Vbd, has a temperature coefficient of +22mV/K. The gain tempco also varies with Vov (V over-voltage). Gain tempco varies from -0.6%/K at Vov = 2.5V to -0.3%/K at Vov=5.0V. Avalanche gain is about .9E6 at Vov=2.5V and around 1.75E6 at Vov=5.0V. The final parameter we need to know is the variation of dark current vs Vov: .2ua(max) @2.5V and 1uA(max) @5V (this is at 21C).

This is a lot to swallow at one go. But there are some obvious things to consider. First, do we really care if gain varies a little bit over temperature? I think so, if we are planning on using some kind of reference library to help extract concentration information -- day-to-day temperature variations will cause shifts in the position of peaks from the MCA. Since many of the elements we're interested in will have overlaps, we need all the stability we can get. The alternative is to implement a quick & easy way to swap in your elemental standards without disturbing the temperature of your detector.....but it still would make it difficult for Graham and me to trade data sets (for instance). And to make this thingie generally useful to others, it would be great to have a data library of elements and alloys that everyone could use.

If we use a fixed bias voltage to give us Vov = 5V at 21C with a simple resistive load, we can expect the quiescent voltage to increase as temperature increases (because Vbd increases). This should increase Vov and therefore the gain should increase, but this is offset some by the negative tempco of gain vs temperature. If we put the detector inside a DC feedback loop to maintain a constant dark current, this would stabilize Vov and eliminate gain variation due to changes in Vov, but it would not address gain vs temperature. However, now we have a single temperature dependency that could be addressed in software -- place a temperature sensor near the detector and massage the raw data accordingly.

Before I go to all that trouble I need to write a system of equations to see just how bad a simple fixed-bias + resistive load approach really is. Wouldn't it be nice if there's a sweet spot with a very low overall gain tempco.
 
Last edited:
Would it be hard to control the temperature at a fixed value? Perhaps heat it slightly above room temp and maintain it there with a heater and control circuit? Although perhaps it would be simpler/easier to correct for temp in the software with a thermocouple as you suggest. Either route requires a probe to measure the temp.
Robert
 
The SiPM's breakdown voltage has a tempco, +22mv/C. It should be possible to use it for the temperature-measurement function. Small Peltier heat pumps are cheap and could be attached to the back of the SiPM board. I'd opt for cooling, just to reduce the background count rate. If you can reduce the temperature to 0C the thermally-generated background will go down by about an order of magnitude.

But then you'd have to worry about condensation, especially if you are using a CsI(Tl) crystal. While it's called "slightly" hygroscopic, it probably would NOT like liquid water on it. But halving the background count rate should be feasible, at least in the temperate parts of the world.
 
  • Like
Reactions: rwm
I've been thinking about Graham's comment about an integrating sphere as a more efficient light collection system. It has some attractive aspects to it. One issue I've wondered about all along is concerning the reflections off the crystal/coupler/detector interface. Scintillator crystals typically have a high index of refraction and that makes it difficult to avoid reflections -- yet another loss mechanism. Light emitted at an angle that is less than the angle of TIR relative to the sides of the crystal will be lost. And, of course, a portion of light emitted away from the detector will be lost (there still would be some internal reflections that would re-direct the light toward the detector).

While this might appear to be less optimal than an aluminized (or silvered) scintillator, it so happens it's not too difficult to make your own DIY integrating sphere. And, since it isn't an optical solution (like the ellipsoid idea), it should be more compatible with a 2D detector. Finally, concerns regarding the absorption of incoming x-rays by an aluminum or silver film would not be an issue.

You can buy aluminum hemispherical "baking pans" for very little money, then coat the interiors with highly reflective paint. For one example, see here. I just bought a pair of aluminum baking hemispheres and barium sulfate powder for that very purpose. Thankew Amazon -- the hemispheres cost less than $10. It shouldn't be too difficult to make some clips to clamp the two halves together. We ARE machinists after all (a brief justification for this crazy mostly-physics thread).

Personally, I'm not too concerned about absorption due to a thin aluminum layer. So I am thinking of an integrating sphere with a hole in it for the entering x-rays and a hole on the other side for the detector. The scintillator would be placed behind the entrance hole, which is covered by a piece of aluminized Mylar to reduce loss back out toward the sample being bombarded by the 59Kev gammas. Internal reflections in the scintillator crystal will eventually find their way outside the crystal and then (eventually) hit the detector. Worried about how to fix the scintillator in place? Glue one side to an acrylic rod. If the glue and rod are transparent, the light will exit THEM, too.

One issue regarding internal reflections in the scintillator is the possibility of a so-called whispering mode, where the photons stay trapped inside the crystal. This can happen with spheres, and has been used to make very efficient whispering-mode lasers. Rectangular hunks of glass may not exhiibit whispering modes; but if they do, modifying them so they have slightly tapered sides would eliminate that. Frosting their surfaces would likely work, since it would introduce random variations in the crystal's surface normal.
 
Sounds complex. Are you overthinking it? A crystal of the right geometry may be efficient enough. Is light collecting efficiency that important if we don't care much about acquisition time?
Robert
 
You must log in or register to reply here.
Back
Top