Experimental Modern and Quantum Physics for Do-It-Yourself Science Enthusiasts 

Facebook Flickr LinkedIn YouTube RSS
magnify
Home Instrumentation Photomultipliers Construction of a d.i.y. Thermoelectrically-Cooled Photomultiplier Tube (PMT) Housing
formats

Construction of a d.i.y. Thermoelectrically-Cooled Photomultiplier Tube (PMT) Housing

diy thermoelectrically cooled PMT housing David Prutchi PhD www.diyPhysics.com

The photomultiplier tubes (PMT) is the workhorse detector in particle physics and many other fields that require detection of light at extremely low levels.  However, the long-wavelength response of PMTs is not only low because of low quantum efficiency, but also because thermionic emission at room temperature causes swamps low-level signals with noise.

Reducing dark counts is especially important in photon-counting applications, especially when attempting to detect photons in the near-infrared. For example, the dark count of many PMTs rated for a wavelength range from 400 to 1200 nm, is in the hundred of thousands of counts when not cooled—making it virtually useless for detecting almost anything but the strongest signal. When cooled to -20 °C, the dark count is reduced to just a few tens counts. As such, in general, the use of PMTs that detect above 600 nm almost mandate a cooled housing.

We constructed a thermoelectrically-cooled housing to experiment with cooling a standard 2” face-on PMT. Although appropriate PMT noise reduction was achieved (one order of magnitude), the thermal efficiency of the do-it-yourself housing design was low, so lessons learned from this build will be used in a second-generation cooled housing.

Temperature dependence of PMT noise

In the dark and at room temperature, the dominant mode of photocathode emission is thermionic. The figure above shows the effect of temperature on dark counts for typical photocathode types, demonstrating that PMT photocathode cooling is highly advantageous in low level detection. This figure also shows that for standard photocathode materials there is little advantage in cooling below -25 °C.

diy thermoelectrically cooled PMT housing David Prutchi PhD www.diyPhysics.com

For this project we retrofitted the PMT probe that we described in Chapter 2 of our book “Exploring Quantum Physics Through Hands-On Projects” with thermoelectric cooling (TEC) modules. We started the conversion by tightly wrapping the PMT with CONETIC magnetic shielding sheet. The shield was held together with aluminum adhesive tape. The outer diameter of the shield was made to fit snuggly inside a Thorlabs XT66-100.

diy thermoelectrically cooled PMT housing David Prutchi PhD www.diyPhysics.com

Then, four 2-stage thermoelectric cooling modules were glued with Artic Silver epoxy adhesive to the Thorlabs XT66-100 construction rail enclosing the PMT. The hot sides of the TEC modules were glued with the same adhesive to two machined aluminum blocks that serve as thermal interfaces to the die-cast enclosure.

diy thermoelectrically cooled PMT housing David Prutchi PhD www.diyPhysics.com

The PMT/thermoelectric cooling module assembly fits tightly inside the die-cast aluminum enclosure. Silver-loaded heat-conductive paste is used to ensure appropriate heat transfer between the aluminum blocks and the enclosure. we placed one K-type and one T-Type thermocouple in good thermal contact with the construction rail at the location of the PMT’s cathode. The thermocouples were held in contact with the rail using Kapton tape, and a dab of Artic Silver epoxy was used to ensure proper thermal contact.

diy thermoelectrically cooled PMT housing David Prutchi PhD www.diyPhysics.com

We used sheets of 0.079″-thick aerogel thermally insulate the PMT assembly from the aluminum blocks and die-cast enclosure. Lastly, we sprayed expanding foam insulation into all cavities prior to closing the enclosure.

diy thermoelectrically cooled PMT housing David Prutchi PhD www.diyPhysics.com

The figure above shows the effect of cooling on a RCA6655A PMT. The dark counts due to thermoelectrons were reduced by an order of magnitude when temperature was reduced from ambient 23⁰C to -13.5⁰C.

Although we believed that the design would provide effective cooling of the PMT, and were originally hoping that the large top heatsink and fan plus only two of the four TEC modules would be needed to cool the PMT to the target -20°C. That was obviously not the case, requiring the use of all four TEC modules, and the addition of two forced-air tunnel heatsinks to the sides of the enclosure to barely reach -13°C at a 23°C room temperature.

Under ideal conditions with no heat load and in vacuum, the 2-stage TEC modules that we chose are capable of producing temperature differentials as great as ΔTmax= 83°C. The enclosure we built however is far from ideal, reaching a ΔTmax≈ 40°C. The difference between the ideal and actual performance is mostly the result of gradients that exist at the various interfaces between the photomultiplier and the heat exchange medium. In addition, there is also a heat load from dynode chain dissipation and an important amount of heat transferred from the housing case through the insulation.

In a future design, the following will be considered to improve the cooling efficiency of the housing:

  • Tighter coupling between the TEC’s hot side and the heat sink.
  • Use of a water-cooled heat-sink.
  • Use of a metal-jacket PMT

A complete white paper describing our build is available here: Prutchi diy PMT Cooled Enclosure March 2013

The following are useful references:

Find more useful information through the Internet Archive

 

Please visit www.diyPhysics.com for other cutting-edge d.i.y. projects, and remember to check out our new d.i.y. Quantum Physics book:

 
 Share on Facebook Share on Twitter Share on Reddit Share on LinkedIn
7 Comments  comments 

7 Responses

  1. Stephen Young

    Hi!

    A simple improvement, (which you might have tried already): Run your experiments in a closed room with air conditioning running full blast. Or, depending where you live, with the windows wide open. As we all well know, it is easier to go down in the basement if you are already halfway down the stairs. (this lousy analogy is not reversible, way tougher for me to go back upstairs than for the hot side of the Peltier to leak back to the cool side when you cut off the supply, as you can see by using one of those small car travel coolers…)

    Also, have you experimented with stacked Peltier modules, run at different voltages? I think it helps at least by increasing the distance between the cooled side and the heat dissipating side, less leakage through whatever you use as insulation there.

    That brings us to having the whole setup inside an aluminum case in contact with the hot side of the Peltier, maybe not a good idea… Have the hot sides stick out like engine cylinders on an old aircraft…

    This is interesting, I will be following this development, while digging out my related hardware, thanks for the re-inspiration!

    Stephen Young

    P.S. Custom machine your heat sink blocks, use mechanical pressure instead of that lousy epoxy, indeed use water cooling.

    • Thank you very much Stephen!
      This is not the first time that reality contradicts my guesses about heat flow inside an enclosure, and I now realize the many mistakes I made in this design.
      Regarding your comments:

      - “A simple improvement, (which you might have tried already): Run your experiments in a closed room with air conditioning running full blast.” – Yes, but my 3 daughters and wife vetoed my A/C settings…

      - “have you experimented with stacked Peltier modules, run at different voltages? I think it helps at least by increasing the distance between the cooled side and the heat dissipating side, less leakage through whatever you use as insulation there.” – Yes, the TEC that I used is a stack of 2 modules, but the way in which I used it didn’t take advantage of the potential increase in performance. I am currently working with Shanni on an APD-based SPCM for which we have a stepped stack of 4 TEC modules. We changed our enclosure design completely after the less-than-stellar results of my cooled PMT housing.

      - “That brings us to having the whole setup inside an aluminum case in contact with the hot side of the Peltier, maybe not a good idea… Have the hot sides stick out like engine cylinders on an old aircraft…” – Agree completely with you. The hot side really has to be in direct contact with the heat sink. The reason I did it this way was so that I could avoid breaching the enclosure and allowing photons to leak through.

      - “Custom machine your heat sink blocks, use mechanical pressure instead of that lousy epoxy, indeed use water cooling.” I did custom machine the aluminum blocks and used both large machine screws and artic silver. However, the loss through the blocks and interfaces is too large. The hot sides really need to be mounted straight on the heatsink.

      What are you working on?

      Best Regards,

      David

  2. Great job! Did you ground the rail that houses the PMT?

    Be wary of those banana jacks – they don’t look like it, but they are a bit translucent – you can get some light through them. It’s most likely not a concern when you have the PMTs sensitive parts shielded like that though.

  3. From my experience – when using generic chineese peltiers, there is no use for 2 stages. 1 stage goes down to ~-17, and no improvement in dual stage setup. Peltier itself works poorly.

    I tried to cool it with -10C air, and still got same ~ -17C result.
    I tried to cool one side to -70 with dry ice, and peltier was unable to cool anymore at all.

    I guess true 2-stage setup must use peltiers made of different materials, they can go to -50C and below.

    • Thanks Mikhail,
      I’m now playing with a stepped 4-stage Chinese TEC, attempting to cool an APD. The hot side is 4 cm x 4 cm and the cold side 1.5 x 1.5 cm. I’ll report on how this works out. Of course, cooling a tiny APD to -20C should be much, much easier than cooling a standard PMT.
      Returning to the PMT, I’m still on the fence between trying again with TEC, or simply building a dry ice or LN2 cooler…
      Cheers,
      David

  4. Brett_cgb

    Experiments I’ve run using identical Peltier modules stacked suggest that the the hotter module is easily saturated by the output of the colder module. This makes sense as the output of the colder module not only includes the heat from whatever is being cooled, but also includes heat from the electrical power dissipated within the colder module. The hot module would need to be able to remove all the electrical power diving the cold module, as well as whatever heat the cold module removed.Bottom line, the hot module needs to have a Qmax greater than the power injected into the cold module, just to be worthwhile.I would also place the heat sink (liquid cooling? heat pipe?) in contact with the hot module, if possible.




Leave a Reply

Your email address will not be published. Required fields are marked *


*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>