I know you were doing it for filming purposes, but just a heads up that it is best practice not to expose PMTs (or APDs for that matter) to room lights even without voltage applied. It won't necessarily damage the PMT, but it can cause trapped electrons in the photocathode to build up. This results in increased "dark counts" (thermionic emissions) for a period of time after light exposure as those trapped electrons work their way out. Source: I'm a quantum optical engineer and work with this kind of stuff daily
Thank you for making this comment. I should probably have mentioned this in the video, although a remark about this was displayed with the dark current specifications. The light in the room wasn't actually that bright, I just used 3200 ISO for filming and kept the PMT away from directly pointing at windows or lights.
@@kevinwhitfield3305 That damage is typically due to the light generating "large" (relatively) currents bombarding the dynodes and resulting in actual material breakdown. That said, any reasonable PMT will have current limiting resistors between dynode stages to prevent permanent damage. In that case, typically the only thing that happens is your count rate saturates and electrons start embedding in the dynodes. This will increase dark counts for quite some time, but can be fixed by tempering (gently heating the PMT to help the electrons wiggle out).
Many years ago I had a play around with a PMT and a red LED, and it could detect the LED with 4nA forward current at room temp, and 0.5nA when cooled with a peltier cooler. For a white LED, just visible to the eye at 50nA, it could detect with a LED current of about 20pA! It could also very easily detect triboluminescence from rubbing sugar cubes together, peeling clear adhesive tape and rubbing two pieces of quartz together.
@@HuygensOptics There is an interesting phenomenon related to this :- At cryogenic temperature, the dark rate in a photomultiplier is caused by single electrons, emitted spontaneously from the cathode surface. This “cryogenic” dark rate increases with decreasing temperature down to at least 4K. The average event rate is proportional to the area of the emitting surface and insensitive to the electric field at that surface. The electrons are emitted in bursts. The bursts are distributed randomly in time, but the events within a burst are highly correlated. The burst durations are distributed according to a power law. As the temperature decreases, the rate of bursts, as well as the number of events per burst, increase. The observed time distributions are indicative of a trap mechanism. So far, there is no physics explanation of the observed phenomenon.
@@primenumberbuster404 "The observed time distributions are indicative of a trap mechanism" - so I am right when I say "One go, a bunch go - and it takes time to re-load and repeat" and I have understood you correct? Very interesting - and awesome subject!
PMTs are indeed quite sensitive. I used to design downhole detectors for the oilfield and even tiny light leaks would cause the PMT to saturate to the point that the HV supply would sag. Some of them would even die an instant death if powered up in a well lit room. And as for what it could detect a physicist wanted to image a bare scintillator in operation in real time using a conventional CCD digital camera as a side experiment, but made the same mistake we all make in correlating numbers to our subjective reality. Even with the largest sources we were permitted to handle by hand and the most efficient scintillator we had (LaBr:Ce) they were still pitch black even after 20 minutes in a completely dark room to the naked eye, so no joy without a special setup and acquisition system to integrate which killed the real time part. (Edit: In hindsight the eye is quite insensitive to the peak emission, so it still might have worked with a camera) Speaking to some old timers, to satisfy their curiosity they got around this by using actual logging sources to “see” the scintillator in action. Works apparently but staring at something releasing many millicurie of radiation doesn’t strike me as very safe 😲
cone and rod cells also use a signal cascade to detect up to a single photon(in the case of rods). its an enzymatic chemical cascade but.. still cool😋😋. great dynamic range on the eyeballs
Hi there, great video! You summarized the topic really well. I work with PMTs and SiPMs daily, collaborating directly with Hamamatsu as a PhD student in particle physics, focusing on light detection in rare event liquid noble experiments. Just a few things I wanted to add: - Typically, dark pulses are similar in intensity, as the vast majority of thermal electrons originate from the photocathode. If you're seeing peaks with different heights, it’s likely due to ambient light. I totally understand the frustration-I've dealt with my share of light leakage issues. Even high-end, all-metal, ultra-high vacuum setups can have leaks. A quick calculation shows how significant diffuse light can be. For example, a 1W light, 10 meters away, shining through a 10μm x 10μm hole, produces an enormous flux of ~200301 photons per second (at 500 nm). This light is Poisson-distributed, so you can calculate the probability of two photons hitting within the same 10 ns to form a two-photon pulse. You probably have a much higher flux, and dimming the light might not make a noticeable difference to your eyes or oscilloscope. A Poisson analysis of your height distribution would likely be useful. - Don’t rely too much on the manual for specs like dark noise or quantum efficiency. These values are often idealized and may not reflect reality, especially for used devices. Who knows what the previous owner did to the tubes? The photocathode might be damaged, which could significantly lower both dark count and efficiency. - For photocounting, I’d recommend using the pulse's charge, as it compensates for slight differences in transit time and hit position on the PMT face. - Have you heard of SiPMs? They’re essentially solid-state PMTs-cheaper and more powerful when it comes to photocounting. SiPMs can distinguish between an N-photon hit and an N+1-photon hit, up to around 8 photons (depending on the model). With PMTs, it becomes tricky to distinguish beyond N=2. Skipper-CCDs, on the other hand, can detect even larger N-values, going beyond 300.
Thanks, that is some really good info and I'll use it for the next video. I'm aware that my measurements are not ideal and some remarks were kind of simplified. Also, I'm thinking of building a dedicated box to hermetically seal the setup and which fits in my freezer ;-). About the variation in peak heights: since I'm using 520nm light, you'd expect these also in the photon pulses since a single photon can release multiple electrons. This is actually illustrated on page 232 of the hamamatsu PMT handbook, showing individual peaks in the puls height histogram at 1, 2, 3, etc. electrons of current.
@@HuygensOptics Thanks! I actually think your setup is really nice, a great start for sure. Nevertheless, a hermetically sealed freezer setup will work a lot better especially if you menage to tape the edges with aluminum tape -> you can get it to almost perfect darkness like this. Regarding one photon releasing more than one electron, I'm not too sure about this. In the photoelectric effect, the photon is absorbed by the electron, not being possible to excite another one. You are suggesting something like a Compton scattering? Or the variation in the multiplication factor at each dynode? (I looked at my copy of the handbook and page 232 is the chapter 13 cover page. Perhaps I have an old copy? (edition 3a) ) Anyway, I think your pulses during LED-on time look great. That's what I expected to see! The same for LED-off time, pretty consistent to what I usually see in a leaky setup -> dark noise + random high intensity pulses. A leaky setup is not necessarily bad. It just increases your noise floor. But if you want to measure large signals, that is usually fine :)
@brunopassarelligell1 about the figure in the PMT handbook. if you download the current version from the Hamamatsu website it's the figure 12-3 named: Output pulse height distribution in a multi-photoelectron event. Note that the pdf page and "physical" page numbers do not match. page 232 = page 245
Hello, another noble liquid experiment physicist here. Really cool video, I really enjoyed it! In the lab, we once made a dark box out of a big pelican case by drilling holes for cables into the bottom half, putting some coax ports in them, and sealing them (I think with black caulk? I don't remember). It was by far the easiest setup to use, because it is possible to quickly shut off power and open it when one needs to make tweaks, I highly recommend something along those lines if you needed a dark box setup that you can reopen frequently, and so you can work with the lights on :) 300 Hz sounds reasonable to me for room temperatures, though, so I don't think you're having a serious problem with light tightness; this is in line with what I would expect. When we run experiments, we often find that there are major variances between PMTs, even from the same batch, so they're rarely run at the exact manufacturer-specified voltage. Reducing the voltage can often reduce the dark current significantly, so if your circuit is capable of that that's an easier approach than cooling them. It seems like you are getting nice and clean waveforms well within what your oscilloscope can pick up, so there's probably quite a bit of margin to reduce the voltage (and hence gain). Coincidence triggers with multiple PMTs are also another way to get around dark currents, though that's a whole thing...if you are interested in that, Leo's Nuclear Methods and Techniques is a go-to for such techniques, and also gives an introduction of the NIM modules that one would use to implement these typically in a particle physics setting. They're often available cheap-ish (high tens, low hundreds) because some of these modules, such as various LeCroys, have been in production since the cold war era and there's tons of oldstock sitting in Physics department attics. I also recommend looking into SiPMs if you're interested, they are very easy to use and also much cheaper if you count in the fact that you don't need a HV supply; I personally haven't used them but I know they're much easier to use, and the only reason I haven't used them is because the experiments I work on really care about low dark current per area, which is where traditional PMTs shine (though SiPMs are making strides). Finally, here's a fun effect you might be interested in: did you know you can use PMTs to measure the frequency of light? Essentially, if the photon energy is greater than twice the workfunction of the photocathode metal, there's a non-zero probability for one photon to result in the release of two electrons. This essentially means that as long as the light is weak enough that pulses don't overlap too often, we can measure the frequency of light by looking at the mean pulse height and/or integrated area! The effect depends on the photocathode material, but in my experience starts to become visible
Small correction at the 9:11 mark: the current should be the charge divided by the transit time spread, not the transit time, because the current does not depend on the time delay it takes for the avalanche to hit the anode, but the interval within which the amplified charge is distributed.
Right. Although, to make good on that transit time spread, one needs a ~1GHz measurement bandwidth. That would probably require a difference socket with a 50 ohm coax output, and would definitely require a different preamp!
Throwback to my undergraduate work at university, calibrating photomultipliers for use in a neutrino experiment. The first thing we discovered was that our initial version of an isolation booth wasn't good enough! The temperature in the lab was about 1-1.5°C degrees higher by late afternoon than morning, and the difference in thermionic emission showed up in our testing.
At the South pole is the Ice Cube neutrino observatory. There is 6km of pure clear ice there. They made a 1x1 km observatory there by sinking (melting) strings of photo multiplier detectors there. Really amazing. They detect neutrinos that pass through the earth. They can also sometimes tell where they came from. Crazy. (The neutrinos cause a very small amount of light in the ice when they hit something, very very rarely, which the observatory detects. )
These are always one of my favorite types of tubes! I find it amazing that such a thing could be conceived of in the first place, let alone actually constructed. And it all flows from the work of some of the most brilliant physicists who ever lived. It's truly an amazing story of what's practically achievable by the physical sciences.
My friend and I made gamma spectrometer from scratch using PMT. We had to figure out low noise amp by ourslefs because PCBs with amp and PMT was way expensive back then. It was fun project. PMT are fun!
Great question Mike. The Hamamatsu PMT handbook lists 6 different reasons for this, among which are: leakage currents (I guess due to charge accumulation on the insulators), scintillation in the glass envelope caused by stray electrons on the glass surface, and ionization of residual gasses inside the tube. All these take some time to fade away and after 30 min. the performance approches the best attainable value again.
Well you could invert the negative voltages and maybe "purge" the electrons out, then invert back to negative, like a sort of electrostatic degaussing Much more complicated power supply though
@@Clancydaenlightenedyeah that seems like a good way to clear them out quickly! Who knows, could just be one of those things that can’t be rushed and ya gotta wait it out lol 🤷🏻♂️ I’m with you though, ideally ya just purge and get on with it.
This video is beyond what words can capture. I paused several times to think about things. Any questions that I had were answered in later stages of the video. I highly value the effort you put into this. Practically on the same level, as I would, if I were not too lazy to actually do this. Please take this comment as praise from my side. I am looking forward to see other videos. Thank you very much for doing this, even if only a handful of people acknowledge and recognize the effort you put into it. Respect. Best wishes, Erik.
I was product manager for a company that made photographic darkroom equipment. We had a color analyzer that utilized the PMT. When testing, we had a radioactive element that was contained in a steel, light tight container. The PMT was put in a fixture that was also light tight then a special valve was opened exposing the PMT to emissions from this radioactive target. The PMT was then adjusted to the constant we determined for our product.
Great video! Some possibly helpful hints: A cheaper choice for a photomultiplier tube is the 931A, and there are tons out there. They are just fine in normal light, and not going to get damaged. One example: The B&K 1076 used one in front of a CRT (video section) with louvers in the case exposing the multiplier tube to outside light while the device was in use, and it performed fine, in fact they still work today. Have fun, don't get zapped! :^)
That's just how chemists roll... When you have a fundamental understanding of how matter and energy interact, many other things just fall into place. (Physics knowledge is a prerequisite for understanding chemistry, however)
Many years ago, I took part in the TOXICHIP project. The goal was to measure single photons from genetacally modified Ecoli bacteria that started polluting light in the presence of toxins. We picked an avalange diods instead of photomultipliers because of high voltage. Plus, semiconductor design is more compact. Tyndall institute in Cork Ireland built for us such diods. I designed a platform that included a chassis, a chip, an array of avalange diods, and electronics. It was 20 years ago.
PMT is a fascinating technology starting with the 932A of WW2 vintage and a good example that even in these days, solid-state does not provide us with everything that we need. Good presentation! Keep up the good work!
Very well done! I've always wondered how those devices worked. For some reason I presumed they were single-stage devices (i.e. the initial electron being accelerated to a much higher velocity and detected by a single grid). Your description of the cascade made how they actually work very clear. Thanks!
I love the fact that today i had my photonics exam, also about single photon detectors, small word ahah Your presentation is so engaging and interesting that it is worth actively following also at 1am 🤩
Nice video. Reminded me of an experiment I did in high school. I got a surplus image intensifier from Edmond Scientific for $12 which is basically the first stage of the photomultiplier with a photocathode screen and electrostatic focusing onto a phosphor screen. I used a B&W TV that someone threw out (color TV was coming into fashion and people would just toss their B&W TVs on the curb) as the HV source, a string of resistors to create a variable voltage source for the focusing electrodes and a camera lens to focus the image on the photocathode. It worked. I tried to turn it into a spatial radiation detector by replacing the photocathode with a NaI scintillation crystal, but it was beyond my abilities as a high school student, but it was good enough to place me as a semifinalist in the Westinghouse Science Talent Search contest. Yippie!
This is a great resource to have for an introduction to PMTs. I have an old 90s high end film scanner that is based around three PMTs and I have to do some maintainance on it but was always a bit squeamish about the high voltage portion that drives the PMTs. I'll try calibrating it again in the future with this info.
Quite a few years ago, Dave from EEVBlog demonstrated a photo-multiplier where he, using a micro-amp meter, tested the current at which an led emitted a tiny amount of photons. That was quite the demonstration. (20uA was enough to have some quantum-dice rolling the right way to, now and then, emit a photon in the direction of the multiplier that I remember)
Awesome video! (Especially because I just bought a similar pmt on eBay and haven’t had a chance to play yet!). A couple things that have been bothering me about PNTs though, do you know if it’s common for electrons to “skip” and hit the wrong plate, like going from the -800v to the -600v? I have to imagine the field DOES curve, but are the electrons just emitted directionally and going fast enough that they don’t miss? I would have expected a uniform hemisphere of ejected electrons. Maybe it doesn’t matter if some small fraction miss… For my yet-theoretical project, I’m hoping for much better temporal resolution than you got here, probably maxing out my oscilloscope’s frequency acceptance plugged up to the pmt directly. my fingers are crossed I can probe ahead of any op-amps that may be present! (And the attached RC network is optimized), but are there any other obvious sources of slowness? I’m really wanting less than 5ns rise time, and while the pmt specs say that’s easily clearable, I know there’s a lot more error elsewhere. Thanks for a great vid! Curious about your last animation with the splitter 🤔
Hi Brian, thanks for stopping by! You are right about the dynode "skipping" . The process that I showed is of course idealized and is in practice quite a bit messier than depicted. Some electrons will even end up on the glass envelope. So indeed, a fraction of the electrons will miss the next dynode in line but that does not matter so much, it will probably lead a broader spread in the electron transit time. But a 200V difference will also mean a 200eV kinetic energy and will thus generate more secondary electrons. In the end, the dynode configuration is optimized such that the internal field in combination with the kinetic energy gained by the electron with its inertia leads to the large gain and a short transit time. You can actually measure the signal directly on the anode output using a scope. Of course in this case it depends on which value you use for the grounding resistor. If you use 50 ohm, your signal will be very small but very fast. if you use 1k, it will be 20 times higher but not so fast. so I guess there is a trade-off for you to make here. And of course there are much faster OpAmps that you can use.
@@HuygensOptics ah yeah! Very cool. I want to do more research on the different dynode layouts too - I just picked a supposedly fast one but I was surprised there are multiple common ones! It’s amazing that it works at all. And yeah on speed I guess I’ll be comparing the input impedance of your op-amp to the input impedance of my cheapo scope lol 😂. I would have hoped that an onboard op amp was pretty darned optimized for this stuff but maybe it’s a setup specific for counting photons and that’s not great if you end up with 10-100 of them per nanosecond 🤪
Someone pointed out here that I actually made a mistake here: the transit time is the time it takes for the electrons to travel through the detector. But if you want to know the pulse width then you should look at the transit time spread (TTS), which is the spread at which the electrons arrive at the anode grid. The TTS has a value is 1.2ns for my photomultiplier. And therefore the shortest attainable pulse width would be in the order of the TTS, say 2 ns for the photomultiplier used in this video rather than the 22ns value I mentioned, which is good news for you!
Great video. PMT are also widely used for detection of x-ray or gamma radiation. A scintillator is used to convert the gamma photon into (visible) light which in order is amplified by a PMT. The PMT with electronics is a bargain. You have also quite an impressive collection of Philips lab equipment.
PMTs has fascinated me since mid-school, mostly from all of the physics going on. Even from to the point of studying statistical thermal dynamics, which you bring up in discussion of dark current. It was a very terse discussion of all aspects of the detectos. I really look forward to watch your next video.
Very helpful clear explanation of the photomultiplyer and the temperature effects on the dark current. I wish to see videos with other detectors like InGaAs for near infrared but I could probably try to do it myself seeing your experiment here. Also you are excellent with the electronics tinkering here.
This is a very useful video that might work well for QuarkNet students (high school) exploring the technology we use in our cosmic ray muon detectors. Thank you!
Awesome stuff as always. You have so much to teach. I hope to see collaboration with some of these channels setting out to make awesome things, even if just for spectacle.
Amazing content; I can only imagine what you could achieve in a dedicated facility and with a decent budget! The mechanics of the photo-multiplier was explained beautifully, in particular the amplification avalanche but I was wondering how we know there is a 1:1 relationship between an incident photon and a photo cathode electron in the first place? Is this postulated or demonstrable? I look forward to the follow-up video with anticipation.
Actually, this isn't the case, it's way more complicated than that. A photon with sufficient energy can release way more than one electron from the photocathode. I will talk about that in the follow-up video.
@@HuygensOptics it was usually explained to me that if a high energy photon gives enough energy to an electron, it can then knock out more electrons using its kinetic energy. So if that interpretation is correct, there is a sense in which there's a 1:1 relationship(a photon cannot be partially absorbed by multiple electrons). You do get multiple electrons out of the photocathode in that case, that's right
I'm wondering, have you worked through the math of the superposition of the photon? Have you landed on virtual photons as a force or are you advocating for something like irradiated electric field with probability amplitudes? And might I say, just wow. The presentation is outstanding. Miles beyond anything else on TH-cam. It's all here. This is the end of science. The technical jargon with the demo to boot really scratches that quest for knowledge itch I haven't been able to quite reach. thanks!
So interesting to see how commercial stuff is made. I made a device with photomultiplier but I had to make everything, including a custom socket to all the electronics. Making a suitable custom socket is always a pain.
Thank you for the excellent explanation of how these work! I use them at work in scanning electron microscopes where they amplify the faint flashes of light created when secondary electrons emitted from a sample strike a scintillator plate. This is amplified and used to form the image we see. Also used in scanning laser confocal microscopes to also amplify faint fluorescent light from a sample.
Some will in fact do that, depending on the local electric field vector and their kinetic energy. However, the construction is designed such that the electrostatic field gradient can create as many secondary electrons as possible during an avalanche.
PMT's are really interesting ! They were used in very high-end drum scanners to digitize film. The film was mounted to a transparent tube with a light inside, and it would spin very quickly and scan the frame pixel by pixel using three PMT's behind dichroic filters to separate R, G and B signal. These scanners were the best ever made in terms of dynamic range, which is unsurprising when PMT's can count photons, as well as in terms of resolution, with some reaching 11k pixels per inch ! The optics used in these is also quite interesting, as the enlarged/projected light is pretty much non-image-forming as it consists of a single pixel. I have been looking into SiPM's as well when exploring this topic, and while they work very differently, I feel like it may be interesting to see a proper comparison between regular PMT's and SiPM's !
I used to work at a place that manufactured PET scanner detectors, they were basically a multi-anode PMT with a scintillator crystal on top (all wrapped up in opaque material so only high energy photons could reach the assembly)
I'm a retired EE, and have played with PMTs, on the cheap. Mostly because my junk parts bin is better stocked than most electronics parts stores. When those stores still existed. I use them to detect very dim light sources.
I have 4 931s in the office that were used for research, biggest use were pre 2010, until the detectors counters decided to fail and out unit could no longer be repaired (chip was impossible to replace). Many a Grad paper and research paper used the tubes. I was even working on a new unit until we switched over to using CCD’s 100% of the time. Getting the tubes to sub-0C will help immensely. We used ours with either a thermoelectric cooler or dry-ice coolers.
How can we ever be sure that it is indeed only single photons that are responsible for the observations of events? Just assuming that the minimum detection level of any setup equates to single photons seems unsafe. I've read many accounts of the double slit experiment that say they reduced the light level to single photons but never any justification or verification of single photons.
That is a good question and indeed this is not the case: many of the pulses observed are the result of more than 1 electron emitted simultaneously (either by a single photon or multiple). If you record the surface / height of many pulses and then put them in a histogram with the number of pulses having a specific surface area, you will find that there are several peaks in the histogram that correspond to the amplification of initially 1,2,3,4 etc. electrons. The effect is described on page 232 of the Hamamatsu PMT Handbook. I did not want to go there in the current video, because it complicates the discussion quite a bit. But it is actually the reason why we observe large height differences between individual pulses.
You can do it easily with a Poisson approach in light emission. Just do a small enough Poisson distribution so the chance of getting 2 photons is basically 0. For that you can pulse an LED with lower and lower voltages, until you see only a photopulse every 10 or so LED pulses. That gives you lambda=1/10, which has a change of 2 photons smaller than 10^-5. Now do the same with 1:100 or 1:10000, you will see the pulses remain the same height, just more sparse
As always you switched on my brain, love it, thank you again. The links to the other potentialshifters made my day😆and I'm not in any way jealous of your guys equipment parks, nooooooooooo. 😂 Was a more relaxing time with current than lights "preferences". Btw. did I mention I hate dust?
Dont forget, a photo multipler is very similar to a geiger counter tube. Some strikes of background radiation on the charged plates with knock-out and electron that then can rush down the pipe and create an event - despite the tube being in the dark.
Very interesting and well presented. I have several PMTs in my collection of electronics. However, zi have never attempted to use them. So I know now what not to do.
This is fascinating. Have you thought about setting up some compact split experiments to show effects of laser light (I remember from your older video, lasers at low energy are not stimulated emission right? I would love to see a series on all the quantum experiments done in depths, setup and explained, maybe get @SabineHossenfelder involved?
In the follow up video, I'm going to discuss the difference between detecting coherent and incoherent radiation quanta using multiple photomultipliers.
Sleep can wait. Whew, good video. I’d be excited about immersing the tube in liquid nitrogen, but I’m not sure how the hermetic seals would handle it. Another thing to test would be if the energy of the photon corresponds to the average size of each charge pulse. Don’t suppose you’ve got a low-amplitude tunable light-source, do you? It’s never too late to build yourself a visible light free electron laser!
For this type of photomultiplier (plastic base), you should not cool below -30C. At this temperature, the thermionic emission will be lower than 1/100th of the room temperature value. The avalanche gain of the PMT will also go up with decreasing temperature.
Excellent explanation. Just curious how is it detecting single photons, isn't the laser diode emitting lots of coherent photons? Only one photon makes it through the aperture of the photomultiplier? What happens if a large sum of photons goes in there?
Detecting single photon events in a thermal background of a few 100 counts per seconds will be a challenge. However, if the laser is pulsed and you gate the amplifier with the laser pulse the background may be reduced. The amplitude of the amplifier output then should vary with the number of primary photoelectrons. I am curious if this will show up. Eagerly awaiting your next video.
Really cool little units with all the needed parts integrated, and at a very reasonable price. What do you think they came from - a university undergrad optics lab being cleared out, or is there a commercial use?
Some kind of classified apparatus: the seller could not disclose the application because of a confidentiality agreement with the company that sold the apparatus to him that contained these items.
One supercool commercial application with PMT was the 1953 thru early 1959 Autronic Eye from General Motors. This device would automa6tic control your headlights (high beam/ low beam)! But the type PMT is unknown, it was an RCA or Delco tube.
If you are looking for an easy way to block light - use tin foil or foil tape. The metallization does a great job blocking light when compared to fabric or similar (cardboard), and the stiffness makes for it staying in place without too much extra effort. Worth mentioning that a grounding clip to earth ground could help too - when measuring such small currents, it's not so difficult for a floating metal piece to act as an antenna and couple in some stray EMF to your sensing lines, so having it grounded dramatically reduces any chance of electrical interference.
Does the photoelectric effect work the same way in a bulk material compared to like, a nanostructure/nanoparticle? Assuming the substance it's made of stays the same
I love how this old tech vacuum tube is superior to semiconductor sensors. A great example for electronic engineers that newer isn't always better. The amplifier circuit you mentioned is more a current to voltage converter. We don't realy speak of amplification here because we can't compare voltage with current even though the power gets amplified.
Curious, when a photon of enough energy lands on a material surface (with a given work function), does the "kick" to the photoelectron always send the electron flying out above the surface? Can the electron be lodged further into the material's volume instead? What would happen if a photon hits a super-thin sheet of photoelectric material? Could the electron come out on the other side?
I wonder if the 1.5pF input capacitance of the AD817 contributes to the long trailoff of the pulses, given the relatively high impedance path to ground on that side.
These are super videos, beautifully explained and diagrammed. A question I have is that these random pulses-described as thermal electrons with the photo multiplier might not be triggered also by random particles as detected by a Geiger-Müller tube detecting cosmic events?
A Gamma camera uses an array of the Hamamatsu miniature square PMTs with Cesium iodide crystals made into an X-Y array. They are used in CAT scan medical instruments. The company I worked for built a 16x16 gamma camera to be used in oilfield pipe inspection. This was back when the 3/8 diameter PMTs were just coming out.
In the past, to develop compact gamma cameras I used position-sensitive PMTs like Hamamatsu's H8500 or H9500, but recently, I've switched to SiPM arrays coupled to CsI(Tl).
challenge accep... uh actually no, my best semiconductor current meter only has 10 aA resolution
Dang, I need to design you some new stickers for that resolution level. PPMs seems like a bulldozer by comparison.
Well, that is still pretty damn impressive Marco, especially if these last digits are significant.
well, still a challenge, go ahead and build one.
Glad to see we watch the same stuff, marco
10 attoAmpere resolution?? 🤯 impossibru ! 🤯
I know you were doing it for filming purposes, but just a heads up that it is best practice not to expose PMTs (or APDs for that matter) to room lights even without voltage applied. It won't necessarily damage the PMT, but it can cause trapped electrons in the photocathode to build up. This results in increased "dark counts" (thermionic emissions) for a period of time after light exposure as those trapped electrons work their way out. Source: I'm a quantum optical engineer and work with this kind of stuff daily
Thank you for making this comment. I should probably have mentioned this in the video, although a remark about this was displayed with the dark current specifications. The light in the room wasn't actually that bright, I just used 3200 ISO for filming and kept the PMT away from directly pointing at windows or lights.
The specifications show measurements taken 30 minutes after startup. This allows time for almost all of the trapped electron buildup to dissipate.
Gratitude for this window ♥️
Interesting! Does this also apply to over-exposed PMTs with power applied? Or do those really become permanently damaged in that case, and if so, how?
@@kevinwhitfield3305 That damage is typically due to the light generating "large" (relatively) currents bombarding the dynodes and resulting in actual material breakdown.
That said, any reasonable PMT will have current limiting resistors between dynode stages to prevent permanent damage. In that case, typically the only thing that happens is your count rate saturates and electrons start embedding in the dynodes. This will increase dark counts for quite some time, but can be fixed by tempering (gently heating the PMT to help the electrons wiggle out).
Many years ago I had a play around with a PMT and a red LED, and it could detect the LED with 4nA forward current at room temp, and 0.5nA when cooled with a peltier cooler.
For a white LED, just visible to the eye at 50nA, it could detect with a LED current of about 20pA!
It could also very easily detect triboluminescence from rubbing sugar cubes together, peeling clear adhesive tape and rubbing two pieces of quartz together.
They are indeed incredibly sensitive, which can be a disadvantage because you have to do all of that in absolut darkness.
@@HuygensOptics There is an interesting phenomenon related to this :-
At cryogenic temperature, the dark rate in a photomultiplier is caused by single electrons, emitted spontaneously from the cathode surface. This “cryogenic” dark rate increases
with decreasing temperature down to at least 4K. The average event rate is proportional to the
area of the emitting surface and insensitive to the electric field at that surface. The electrons are emitted in bursts. The bursts are distributed randomly in time, but the events within a burst are highly correlated. The burst durations are distributed according to a power law. As the temperature decreases, the rate of bursts, as well as the number of events per burst, increase.
The observed time distributions are indicative of a trap mechanism. So far, there is no physics explanation of the observed phenomenon.
@@primenumberbuster404 "The observed time distributions are indicative of a trap mechanism"
- so I am right when I say "One go, a bunch go - and it takes time to re-load and repeat" and I have understood you correct?
Very interesting - and awesome subject!
PMTs are indeed quite sensitive. I used to design downhole detectors for the oilfield and even tiny light leaks would cause the PMT to saturate to the point that the HV supply would sag. Some of them would even die an instant death if powered up in a well lit room. And as for what it could detect a physicist wanted to image a bare scintillator in operation in real time using a conventional CCD digital camera as a side experiment, but made the same mistake we all make in correlating numbers to our subjective reality. Even with the largest sources we were permitted to handle by hand and the most efficient scintillator we had (LaBr:Ce) they were still pitch black even after 20 minutes in a completely dark room to the naked eye, so no joy without a special setup and acquisition system to integrate which killed the real time part. (Edit: In hindsight the eye is quite insensitive to the peak emission, so it still might have worked with a camera) Speaking to some old timers, to satisfy their curiosity they got around this by using actual logging sources to “see” the scintillator in action. Works apparently but staring at something releasing many millicurie of radiation doesn’t strike me as very safe 😲
@@primenumberbuster404 Interesting, I have never heard about this! Thanks!
The cascade system to amplify the signal was one of the coolest things I learned in my sophomore years.
That's not that cool tbh
cone and rod cells also use a signal cascade to detect up to a single photon(in the case of rods). its an enzymatic chemical cascade but.. still cool😋😋. great dynamic range on the eyeballs
After a long time with no observations.... The signal has returned.
Checks notes: Wow! (Underlined)
Hi there, great video! You summarized the topic really well. I work with PMTs and SiPMs daily, collaborating directly with Hamamatsu as a PhD student in particle physics, focusing on light detection in rare event liquid noble experiments.
Just a few things I wanted to add:
- Typically, dark pulses are similar in intensity, as the vast majority of thermal electrons originate from the photocathode. If you're seeing peaks with different heights, it’s likely due to ambient light.
I totally understand the frustration-I've dealt with my share of light leakage issues. Even high-end, all-metal, ultra-high vacuum setups can have leaks. A quick calculation shows how significant diffuse light can be. For example, a 1W light, 10 meters away, shining through a 10μm x 10μm hole, produces an enormous flux of ~200301 photons per second (at 500 nm). This light is Poisson-distributed, so you can calculate the probability of two photons hitting within the same 10 ns to form a two-photon pulse. You probably have a much higher flux, and dimming the light might not make a noticeable difference to your eyes or oscilloscope. A Poisson analysis of your height distribution would likely be useful.
- Don’t rely too much on the manual for specs like dark noise or quantum efficiency. These values are often idealized and may not reflect reality, especially for used devices. Who knows what the previous owner did to the tubes? The photocathode might be damaged, which could significantly lower both dark count and efficiency.
- For photocounting, I’d recommend using the pulse's charge, as it compensates for slight differences in transit time and hit position on the PMT face.
- Have you heard of SiPMs? They’re essentially solid-state PMTs-cheaper and more powerful when it comes to photocounting. SiPMs can distinguish between an N-photon hit and an N+1-photon hit, up to around 8 photons (depending on the model). With PMTs, it becomes tricky to distinguish beyond N=2. Skipper-CCDs, on the other hand, can detect even larger N-values, going beyond 300.
Thanks, that is some really good info and I'll use it for the next video. I'm aware that my measurements are not ideal and some remarks were kind of simplified. Also, I'm thinking of building a dedicated box to hermetically seal the setup and which fits in my freezer ;-). About the variation in peak heights: since I'm using 520nm light, you'd expect these also in the photon pulses since a single photon can release multiple electrons. This is actually illustrated on page 232 of the hamamatsu PMT handbook, showing individual peaks in the puls height histogram at 1, 2, 3, etc. electrons of current.
@@HuygensOptics Thanks! I actually think your setup is really nice, a great start for sure. Nevertheless, a hermetically sealed freezer setup will work a lot better especially if you menage to tape the edges with aluminum tape -> you can get it to almost perfect darkness like this.
Regarding one photon releasing more than one electron, I'm not too sure about this. In the photoelectric effect, the photon is absorbed by the electron, not being possible to excite another one. You are suggesting something like a Compton scattering? Or the variation in the multiplication factor at each dynode? (I looked at my copy of the handbook and page 232 is the chapter 13 cover page. Perhaps I have an old copy? (edition 3a) )
Anyway, I think your pulses during LED-on time look great. That's what I expected to see! The same for LED-off time, pretty consistent to what I usually see in a leaky setup -> dark noise + random high intensity pulses.
A leaky setup is not necessarily bad. It just increases your noise floor. But if you want to measure large signals, that is usually fine :)
@@HuygensOpticsBe careful with cooling and HV! Condensation, high voltage and sensitive op amps do not mix well.
@brunopassarelligell1 about the figure in the PMT handbook. if you download the current version from the Hamamatsu website it's the figure 12-3 named: Output pulse height distribution in a multi-photoelectron event. Note that the pdf page and "physical" page numbers do not match. page 232 = page 245
@@HuygensOptics So you can finally answer the age old question if the light really goes out if you close the fridge!
Hello, another noble liquid experiment physicist here. Really cool video, I really enjoyed it!
In the lab, we once made a dark box out of a big pelican case by drilling holes for cables into the bottom half, putting some coax ports in them, and sealing them (I think with black caulk? I don't remember). It was by far the easiest setup to use, because it is possible to quickly shut off power and open it when one needs to make tweaks, I highly recommend something along those lines if you needed a dark box setup that you can reopen frequently, and so you can work with the lights on :) 300 Hz sounds reasonable to me for room temperatures, though, so I don't think you're having a serious problem with light tightness; this is in line with what I would expect.
When we run experiments, we often find that there are major variances between PMTs, even from the same batch, so they're rarely run at the exact manufacturer-specified voltage. Reducing the voltage can often reduce the dark current significantly, so if your circuit is capable of that that's an easier approach than cooling them. It seems like you are getting nice and clean waveforms well within what your oscilloscope can pick up, so there's probably quite a bit of margin to reduce the voltage (and hence gain). Coincidence triggers with multiple PMTs are also another way to get around dark currents, though that's a whole thing...if you are interested in that, Leo's Nuclear Methods and Techniques is a go-to for such techniques, and also gives an introduction of the NIM modules that one would use to implement these typically in a particle physics setting. They're often available cheap-ish (high tens, low hundreds) because some of these modules, such as various LeCroys, have been in production since the cold war era and there's tons of oldstock sitting in Physics department attics.
I also recommend looking into SiPMs if you're interested, they are very easy to use and also much cheaper if you count in the fact that you don't need a HV supply; I personally haven't used them but I know they're much easier to use, and the only reason I haven't used them is because the experiments I work on really care about low dark current per area, which is where traditional PMTs shine (though SiPMs are making strides).
Finally, here's a fun effect you might be interested in: did you know you can use PMTs to measure the frequency of light? Essentially, if the photon energy is greater than twice the workfunction of the photocathode metal, there's a non-zero probability for one photon to result in the release of two electrons. This essentially means that as long as the light is weak enough that pulses don't overlap too often, we can measure the frequency of light by looking at the mean pulse height and/or integrated area! The effect depends on the photocathode material, but in my experience starts to become visible
Small correction at the 9:11 mark: the current should be the charge divided by the transit time spread, not the transit time, because the current does not depend on the time delay it takes for the avalanche to hit the anode, but the interval within which the amplified charge is distributed.
Right. Although, to make good on that transit time spread, one needs a ~1GHz measurement bandwidth. That would probably require a difference socket with a 50 ohm coax output, and would definitely require a different preamp!
Yes you are correct, this is an error that I overlooked. I will set this straight in the follow-up!
очень правильное замечание. Вы полностью правы
Throwback to my undergraduate work at university, calibrating photomultipliers for use in a neutrino experiment.
The first thing we discovered was that our initial version of an isolation booth wasn't good enough! The temperature in the lab was about 1-1.5°C degrees higher by late afternoon than morning, and the difference in thermionic emission showed up in our testing.
Yes, I guess that 5% difference should be clearly detectable, especially if you test a lot of them.
the abundance of detail and explanations in the video is a breath of fresh air
At the South pole is the Ice Cube neutrino observatory. There is 6km of pure clear ice there. They made a 1x1 km observatory there by sinking (melting) strings of photo multiplier detectors there. Really amazing. They detect neutrinos that pass through the earth. They can also sometimes tell where they came from. Crazy. (The neutrinos cause a very small amount of light in the ice when they hit something, very very rarely, which the observatory detects. )
Placing them by sinking and melting is a certified "during a shower " idea.
These are always one of my favorite types of tubes! I find it amazing that such a thing could be conceived of in the first place, let alone actually constructed.
And it all flows from the work of some of the most brilliant physicists who ever lived. It's truly an amazing story of what's practically achievable by the physical sciences.
My friend and I made gamma spectrometer from scratch using PMT. We had to figure out low noise amp by ourslefs because PCBs with amp and PMT was way expensive back then. It was fun project. PMT are fun!
The datasheet mentions a 30 minute settle-down time after exposure to high light levels - what is the mechanism that causes this ?
Great question Mike. The Hamamatsu PMT handbook lists 6 different reasons for this, among which are: leakage currents (I guess due to charge accumulation on the insulators), scintillation in the glass envelope caused by stray electrons on the glass surface, and ionization of residual gasses inside the tube. All these take some time to fade away and after 30 min. the performance approches the best attainable value again.
Various trapped electrons can be released over time resulting in additional dark counts.
Well you could invert the negative voltages and maybe "purge" the electrons out, then invert back to negative, like a sort of electrostatic degaussing
Much more complicated power supply though
@@Clancydaenlightenedyeah that seems like a good way to clear them out quickly! Who knows, could just be one of those things that can’t be rushed and ya gotta wait it out lol 🤷🏻♂️ I’m with you though, ideally ya just purge and get on with it.
This video is beyond what words can capture. I paused several times to think about things. Any questions that I had were answered in later stages of the video. I highly value the effort you put into this. Practically on the same level, as I would, if I were not too lazy to actually do this. Please take this comment as praise from my side. I am looking forward to see other videos. Thank you very much for doing this, even if only a handful of people acknowledge and recognize the effort you put into it. Respect. Best wishes, Erik.
I was product manager for a company that made photographic darkroom equipment. We had a color analyzer that utilized the PMT. When testing, we had a radioactive element that was contained in a steel, light tight container. The PMT was put in a fixture that was also light tight then a special valve was opened exposing the PMT to emissions from this radioactive target. The PMT was then adjusted to the constant we determined for our product.
You are enriching the world, Mr.Huygens!
No need to wait to watch, instant thumbs up for any Huygens Optics vid.
Great video! Some possibly helpful hints: A cheaper choice for a photomultiplier tube is the 931A, and there are tons out there. They are just fine in normal light, and not going to get damaged. One example: The B&K 1076 used one in front of a CRT (video section) with louvers in the case exposing the multiplier tube to outside light while the device was in use, and it performed fine, in fact they still work today. Have fun, don't get zapped! :^)
Is very kind of you to try and detect single photons because photons deserve to be paired with someone else and not be lonely.
Freudian comment
For a chemist, you know more physics, electronics and SW than 90% of people working on those fiields
That's just how chemists roll... When you have a fundamental understanding of how matter and energy interact, many other things just fall into place. (Physics knowledge is a prerequisite for understanding chemistry, however)
Many years ago, I took part in the TOXICHIP project. The goal was to measure single photons from genetacally modified Ecoli bacteria that started polluting light in the presence of toxins. We picked an avalange diods instead of photomultipliers because of high voltage. Plus, semiconductor design is more compact. Tyndall institute in Cork Ireland built for us such diods. I designed a platform that included a chassis, a chip, an array of avalange diods, and electronics. It was 20 years ago.
True PMTs are old school, but they are still interesting and beautiful technology.
Can you recommend some materials, which would help build such decice? I would like to measure 100 photon/second flux for hobby astronomy project
@Woloszow try to search for Dr. Hadar Ben-Yoav. He published a few papers about this project.
@@HuygensOptics I think Hammamatsu would challenge that! They come out with new PMT designs each year.
@@glasslinger like the TO-8 packaged R7400U PMTs! Super neat: PMT the size of a fingertip.
PMT is a fascinating technology starting with the 932A of WW2 vintage and a good example that even in these days, solid-state does not provide us with everything that we need.
Good presentation! Keep up the good work!
New video - new portion of the greatest scientific information I probably never need in my life. But it is so interesting! Thank you
Very well done! I've always wondered how those devices worked. For some reason I presumed they were single-stage devices (i.e. the initial electron being accelerated to a much higher velocity and detected by a single grid). Your description of the cascade made how they actually work very clear. Thanks!
I love the fact that today i had my photonics exam, also about single photon detectors, small word ahah
Your presentation is so engaging and interesting that it is worth actively following also at 1am 🤩
2:40 19.5 Digit Multimeter. Impressive!
Now let's see Paul Allen's Multimeter
"This bad boy can keep it's cal for a solid 3 femtoseconds"
Marco Reps eat your heart out😅
Nice video. Reminded me of an experiment I did in high school. I got a surplus image intensifier from Edmond Scientific for $12 which is basically the first stage of the photomultiplier with a photocathode screen and electrostatic focusing onto a phosphor screen. I used a B&W TV that someone threw out (color TV was coming into fashion and people would just toss their B&W TVs on the curb) as the HV source, a string of resistors to create a variable voltage source for the focusing electrodes and a camera lens to focus the image on the photocathode. It worked. I tried to turn it into a spatial radiation detector by replacing the photocathode with a NaI scintillation crystal, but it was beyond my abilities as a high school student, but it was good enough to place me as a semifinalist in the Westinghouse Science Talent Search contest. Yippie!
You left us with a "cliff hangar"! I cant wait for the next episode.
Always so excited to see a new video from this channel!!
thank u man
now i understand how a photomultiplier works
i studied about them in the past, but your visuals and explanation was much better
Great video - very informative. Thanks for taking the time to put all of this together.
Ooh, I designed some bases for some Hamamatsu PMTs a few years ago. Very cool stuff!
This was a very well done high quality video. Very nicely done illustrations/graphics and diagrams. You get 10 points of 10 points available.
"Even yours, Marco" made me laugh more than I should have.
That was awesome. Loved the tie-in.
This is a great resource to have for an introduction to PMTs. I have an old 90s high end film scanner that is based around three PMTs and I have to do some maintainance on it but was always a bit squeamish about the high voltage portion that drives the PMTs. I'll try calibrating it again in the future with this info.
The detail in your videos is absolutely amazing
Quite a few years ago, Dave from EEVBlog demonstrated a photo-multiplier where he, using a micro-amp meter, tested the current at which an led emitted a tiny amount of photons. That was quite the demonstration. (20uA was enough to have some quantum-dice rolling the right way to, now and then, emit a photon in the direction of the multiplier that I remember)
th-cam.com/video/TJcMgoJ4DOE/w-d-xo.html
Excellent example of using science to study more science. I love that kind of advancement. Every device is an experiment in this perspective.
Flow Cytometry specialist here: Without this we wouldn't be doing ANYTHING>.
Nice video. A step by step guide about how to make a Photomultiplier Tube work and how it works. Thank you.
Awesome video! (Especially because I just bought a similar pmt on eBay and haven’t had a chance to play yet!).
A couple things that have been bothering me about PNTs though, do you know if it’s common for electrons to “skip” and hit the wrong plate, like going from the -800v to the -600v? I have to imagine the field DOES curve, but are the electrons just emitted directionally and going fast enough that they don’t miss? I would have expected a uniform hemisphere of ejected electrons. Maybe it doesn’t matter if some small fraction miss…
For my yet-theoretical project, I’m hoping for much better temporal resolution than you got here, probably maxing out my oscilloscope’s frequency acceptance plugged up to the pmt directly. my fingers are crossed I can probe ahead of any op-amps that may be present! (And the attached RC network is optimized), but are there any other obvious sources of slowness? I’m really wanting less than 5ns rise time, and while the pmt specs say that’s easily clearable, I know there’s a lot more error elsewhere.
Thanks for a great vid! Curious about your last animation with the splitter 🤔
Hi Brian, thanks for stopping by! You are right about the dynode "skipping" . The process that I showed is of course idealized and is in practice quite a bit messier than depicted. Some electrons will even end up on the glass envelope. So indeed, a fraction of the electrons will miss the next dynode in line but that does not matter so much, it will probably lead a broader spread in the electron transit time. But a 200V difference will also mean a 200eV kinetic energy and will thus generate more secondary electrons. In the end, the dynode configuration is optimized such that the internal field in combination with the kinetic energy gained by the electron with its inertia leads to the large gain and a short transit time.
You can actually measure the signal directly on the anode output using a scope. Of course in this case it depends on which value you use for the grounding resistor. If you use 50 ohm, your signal will be very small but very fast. if you use 1k, it will be 20 times higher but not so fast. so I guess there is a trade-off for you to make here. And of course there are much faster OpAmps that you can use.
@@HuygensOptics ah yeah! Very cool. I want to do more research on the different dynode layouts too - I just picked a supposedly fast one but I was surprised there are multiple common ones! It’s amazing that it works at all.
And yeah on speed I guess I’ll be comparing the input impedance of your op-amp to the input impedance of my cheapo scope lol 😂. I would have hoped that an onboard op amp was pretty darned optimized for this stuff but maybe it’s a setup specific for counting photons and that’s not great if you end up with 10-100 of them per nanosecond 🤪
Someone pointed out here that I actually made a mistake here: the transit time is the time it takes for the electrons to travel through the detector. But if you want to know the pulse width then you should look at the transit time spread (TTS), which is the spread at which the electrons arrive at the anode grid. The TTS has a value is 1.2ns for my photomultiplier. And therefore the shortest attainable pulse width would be in the order of the TTS, say 2 ns for the photomultiplier used in this video rather than the 22ns value I mentioned, which is good news for you!
Oooo! Greatly looking forward to part 2.
Great video. PMT are also widely used for detection of x-ray or gamma radiation. A scintillator is used to convert the gamma photon into (visible) light which in order is amplified by a PMT.
The PMT with electronics is a bargain. You have also quite an impressive collection of Philips lab equipment.
18:30 so you could also just use a simple LED instead of a laser? Would there be any difference?
Have you figured out what the measurement output labeled "video" is for? Thanks for the very detailed explanation on this interesting topic!
PMTs has fascinated me since mid-school, mostly from all of the physics going on. Even from to the point of studying statistical thermal dynamics, which you bring up in discussion of dark current. It was a very terse discussion of all aspects of the detectos. I really look forward to watch your next video.
it's already published a few weeks ago: th-cam.com/video/NVqT2Gbrvxs/w-d-xo.html
I've been looking for a video like this for the past two months
Very helpful clear explanation of the photomultiplyer and the temperature effects on the dark current. I wish to see videos with other detectors like InGaAs for near infrared but I could probably try to do it myself seeing your experiment here. Also you are excellent with the electronics tinkering here.
Another great material to watch! Thanks! 😃
another remarkable video
the detail and clarity are just fantastic!
This is a very useful video that might work well for QuarkNet students (high school) exploring the technology we use in our cosmic ray muon detectors. Thank you!
Awesome stuff as always.
You have so much to teach. I hope to see collaboration with some of these channels setting out to make awesome things, even if just for spectacle.
Amazing content; I can only imagine what you could achieve in a dedicated facility and with a decent budget!
The mechanics of the photo-multiplier was explained beautifully, in particular the amplification avalanche but I was wondering how we know there is a 1:1 relationship between an incident photon and a photo cathode electron in the first place? Is this postulated or demonstrable?
I look forward to the follow-up video with anticipation.
Actually, this isn't the case, it's way more complicated than that. A photon with sufficient energy can release way more than one electron from the photocathode. I will talk about that in the follow-up video.
@@HuygensOptics it was usually explained to me that if a high energy photon gives enough energy to an electron, it can then knock out more electrons using its kinetic energy. So if that interpretation is correct, there is a sense in which there's a 1:1 relationship(a photon cannot be partially absorbed by multiple electrons). You do get multiple electrons out of the photocathode in that case, that's right
Love this video and look forward to the next one!
I'm wondering, have you worked through the math of the superposition of the photon? Have you landed on virtual photons as a force or are you advocating for something like irradiated electric field with probability amplitudes?
And might I say, just wow. The presentation is outstanding. Miles beyond anything else on TH-cam. It's all here. This is the end of science. The technical jargon with the demo to boot really scratches that quest for knowledge itch I haven't been able to quite reach. thanks!
Thats incredible, fascinating videos I am looking while being sure I will never need such information. Thanks!
So interesting to see how commercial stuff is made. I made a device with photomultiplier but I had to make everything, including a custom socket to all the electronics. Making a suitable custom socket is always a pain.
amazing video 🎉🎉
can't wait for the second video
Thank you for the excellent explanation of how these work! I use them at work in scanning electron microscopes where they amplify the faint flashes of light created when secondary electrons emitted from a sample strike a scintillator plate. This is amplified and used to form the image we see. Also used in scanning laser confocal microscopes to also amplify faint fluorescent light from a sample.
Why do the electrons from the -900V dynode go to the -800V dynode across from it rather than the -700V dynode right next to it?
Some will in fact do that, depending on the local electric field vector and their kinetic energy. However, the construction is designed such that the electrostatic field gradient can create as many secondary electrons as possible during an avalanche.
PMT's are really interesting ! They were used in very high-end drum scanners to digitize film. The film was mounted to a transparent tube with a light inside, and it would spin very quickly and scan the frame pixel by pixel using three PMT's behind dichroic filters to separate R, G and B signal. These scanners were the best ever made in terms of dynamic range, which is unsurprising when PMT's can count photons, as well as in terms of resolution, with some reaching 11k pixels per inch ! The optics used in these is also quite interesting, as the enlarged/projected light is pretty much non-image-forming as it consists of a single pixel.
I have been looking into SiPM's as well when exploring this topic, and while they work very differently, I feel like it may be interesting to see a proper comparison between regular PMT's and SiPM's !
I used to work at a place that manufactured PET scanner detectors, they were basically a multi-anode PMT with a scintillator crystal on top (all wrapped up in opaque material so only high energy photons could reach the assembly)
Insane!! Thank you for making such cool stuff... Looking forward to part II!
What a cliffhanger! Can't wait for part 2 :)
I'm a retired EE, and have played with PMTs, on the cheap. Mostly because my junk parts bin is better stocked than most electronics parts stores. When those stores still existed. I use them to detect very dim light sources.
I have 4 931s in the office that were used for research, biggest use were pre 2010, until the detectors counters decided to fail and out unit could no longer be repaired (chip was impossible to replace). Many a Grad paper and research paper used the tubes. I was even working on a new unit until we switched over to using CCD’s 100% of the time.
Getting the tubes to sub-0C will help immensely. We used ours with either a thermoelectric cooler or dry-ice coolers.
This is awesome... the video, the content and way of expressing 👍👍:))
Nice! There are semiconductor equivalents that I am looking into, but I love to see glass 🙂 looking forward to seeing the follow up!
How can we ever be sure that it is indeed only single photons that are responsible for the observations of events? Just assuming that the minimum detection level of any setup equates to single photons seems unsafe. I've read many accounts of the double slit experiment that say they reduced the light level to single photons but never any justification or verification of single photons.
That is a good question and indeed this is not the case: many of the pulses observed are the result of more than 1 electron emitted simultaneously (either by a single photon or multiple). If you record the surface / height of many pulses and then put them in a histogram with the number of pulses having a specific surface area, you will find that there are several peaks in the histogram that correspond to the amplification of initially 1,2,3,4 etc. electrons. The effect is described on page 232 of the Hamamatsu PMT Handbook. I did not want to go there in the current video, because it complicates the discussion quite a bit. But it is actually the reason why we observe large height differences between individual pulses.
You can do it easily with a Poisson approach in light emission. Just do a small enough Poisson distribution so the chance of getting 2 photons is basically 0.
For that you can pulse an LED with lower and lower voltages, until you see only a photopulse every 10 or so LED pulses. That gives you lambda=1/10, which has a change of 2 photons smaller than 10^-5. Now do the same with 1:100 or 1:10000, you will see the pulses remain the same height, just more sparse
@@brunopassarelligell1 if you could elaborate on this procedure, I'd be very grateful 🙏
He elaborated a similar procedure using nd filters in one of his videos@@wizardatmath
It's fascinating to see how well that works on such low voltage.
As always you switched on my brain, love it, thank you again. The links to the other potentialshifters made my day😆and I'm not in any way jealous of your guys equipment parks, nooooooooooo. 😂
Was a more relaxing time with current than lights "preferences". Btw. did I mention I hate dust?
Dont forget, a photo multipler is very similar to a geiger counter tube. Some strikes of background radiation on the charged plates with knock-out and electron that then can rush down the pipe and create an event - despite the tube being in the dark.
gold like this really sets you apart. amazing
Shout out to Marco reps!!!! Sir, my hat is off.
Always a pleasure to watch!
Very interesting and well presented. I have several PMTs in my collection of electronics. However, zi have never attempted to use them. So I know now what not to do.
This is fascinating. Have you thought about setting up some compact split experiments to show effects of laser light (I remember from your older video, lasers at low energy are not stimulated emission right?
I would love to see a series on all the quantum experiments done in depths, setup and explained, maybe get @SabineHossenfelder involved?
In the follow up video, I'm going to discuss the difference between detecting coherent and incoherent radiation quanta using multiple photomultipliers.
@@HuygensOptics This is going to be fantastic, this is exactly what we're waiting for, this is some amazing stuff, thank you!
Very nice. I worked once with a gamma ray log for oil rock samples that used a large photomultiplier, very expensive machine and very sensitive.
Excellent video as usual, thank you.
Sleep can wait.
Whew, good video. I’d be excited about immersing the tube in liquid nitrogen, but I’m not sure how the hermetic seals would handle it. Another thing to test would be if the energy of the photon corresponds to the average size of each charge pulse. Don’t suppose you’ve got a low-amplitude tunable light-source, do you? It’s never too late to build yourself a visible light free electron laser!
For this type of photomultiplier (plastic base), you should not cool below -30C. At this temperature, the thermionic emission will be lower than 1/100th of the room temperature value. The avalanche gain of the PMT will also go up with decreasing temperature.
Awesome explanation. Thanks.
Awesome video! Feels like direct response to my question
this channel is so freaking cool
Looking at spontaneous emission with two photo multipliers will be cool, haven't heard of the term photon bunching but i bet it does what it says
Thank you! Very interesting, but how do we know that it is just a single photon that ejects the electron?
Excellent explanation. Just curious how is it detecting single photons, isn't the laser diode emitting lots of coherent photons? Only one photon makes it through the aperture of the photomultiplier? What happens if a large sum of photons goes in there?
Thank you for this excellent tutorial! 👏👏👏
Detecting single photon events in a thermal background of a few 100 counts per seconds will be a challenge. However, if the laser is pulsed and you gate the amplifier with the laser pulse the background may be reduced. The amplitude of the amplifier output then should vary with the number of primary photoelectrons.
I am curious if this will show up. Eagerly awaiting your next video.
Really cool little units with all the needed parts integrated, and at a very reasonable price. What do you think they came from - a university undergrad optics lab being cleared out, or is there a commercial use?
Some kind of classified apparatus: the seller could not disclose the application because of a confidentiality agreement with the company that sold the apparatus to him that contained these items.
@@HuygensOptics Maybe check it with a radioactivity detector!!
It's very impressive to be able to detect single photons.
One supercool commercial application with PMT was the 1953 thru early 1959 Autronic Eye from General Motors. This device would automa6tic control your headlights (high beam/ low beam)! But the type PMT is unknown, it was an RCA or Delco tube.
If you are looking for an easy way to block light - use tin foil or foil tape. The metallization does a great job blocking light when compared to fabric or similar (cardboard), and the stiffness makes for it staying in place without too much extra effort.
Worth mentioning that a grounding clip to earth ground could help too - when measuring such small currents, it's not so difficult for a floating metal piece to act as an antenna and couple in some stray EMF to your sensing lines, so having it grounded dramatically reduces any chance of electrical interference.
Does the photoelectric effect work the same way in a bulk material compared to like, a nanostructure/nanoparticle? Assuming the substance it's made of stays the same
I love how this old tech vacuum tube is superior to semiconductor sensors. A great example for electronic engineers that newer isn't always better. The amplifier circuit you mentioned is more a current to voltage converter. We don't realy speak of amplification here because we can't compare voltage with current even though the power gets amplified.
Semiconductors are pretty close. They have a small edge on QE and aren't affected by magnetic fields. However they have their own problems.
Curious, when a photon of enough energy lands on a material surface (with a given work function), does the "kick" to the photoelectron always send the electron flying out above the surface? Can the electron be lodged further into the material's volume instead?
What would happen if a photon hits a super-thin sheet of photoelectric material? Could the electron come out on the other side?
I wonder if the 1.5pF input capacitance of the AD817 contributes to the long trailoff of the pulses, given the relatively high impedance path to ground on that side.
This is exactly what i wanna know! Thank you!
These are super videos, beautifully explained and diagrammed. A question I have is that these random pulses-described as thermal electrons with the photo multiplier might not be triggered also by random particles as detected by a Geiger-Müller tube detecting cosmic events?
Today is a good day...
As side note, I have wondered if is possible to use one of those as a camera.
In that case you need to add an XY scanner.
A Gamma camera uses an array of the Hamamatsu miniature square PMTs with Cesium iodide crystals made into an X-Y array. They are used in CAT scan medical instruments. The company I worked for built a 16x16 gamma camera to be used in oilfield pipe inspection. This was back when the 3/8 diameter PMTs were just coming out.
In the past, to develop compact gamma cameras I used position-sensitive PMTs like Hamamatsu's H8500 or H9500, but recently, I've switched to SiPM arrays coupled to CsI(Tl).