John Curl's Blowtorch preamplifier part III

Status
Not open for further replies.
Ah, okay, evaporative cooling still works at these temps. :) I don't remember what the record is nowadays, but pretty close to zero (but definitely not zero)

I grabbed one of the papers, here's the key sentence to your question:



From: Electronic noise due to temperature differences in atomic-scale junctions | Nature

4.2K...that's more like it. Anybody can do that.

I had to put a pair of 100 HP motors to use on a pump to get down to 1.8K in a big dewar. A two liter or so chamber with fins, put 4.2K liquid in it, pull a vacuum, it boils, and in the process, the fins cool down the helium it's immersed into. We maintain a liquid level in the chamber as gas leaves, and it eventually cools down the large volume of liquid to 1.88.

It gets really expensive at these temps, 1 watt into 4.2K liquid requires 1000 watts of electricity at the fridge. 1 watt into 1.8K requires 2 kilowatts at the fridge.

I suspect it isn't even exponential, that Carnot guy is very unforgiving. An probably much more than 1/ delta t.

jn
 
...I had to put a pair of 100 HP motors to use on a pump to get down to 1.8K in a big dewar...

Yikes! I never used He evap chillers! I did have fun installing and maintaining two other types of vacuum systems:

1) Turbopumped vacuum systems which only achieve a moderate vacuum (10^-6 torr) in a fully baked-out system, but which had really high recovery rates for metallizing discs at a 10^-2 torr pressure in a 2 second (!!) cycle time. We had over 50 of these systems.

2) Cryopumped systems for stamper master glass metallizing using 10" cryopumps. In a fully regenerated and baked out system with new shields we would see ultimate vacuums in the 10^-10 torr range. Their outstanding characteristics were pumping speeds better than 3000 L/s after roughing to 10^-2 torr with a mechanical pump, it was awesome to watch the pressure plummet on the ion gauge after crossover! The second stage of these cryopumps could eventually settle down to the 4° K range. We had LN H2O traps pre-cryo that would grow big snowballs the kids (young operators) would play with in the clean room during regen...

I really loved working with vacuum systems, and after I was able to stop the kids from metallizing cheap sunglasses, watching marshmallows grow and explode contaminating the chamber, etc., those big CHA bell-jar units were really reliable.

Kids....what ya gonna do with them...they do work cheap though...sorry for the off-topic musings...

Howie
 
That seems backwards in terms of the pumping efficacy of turbomolecular pumps vs cryo. The latter typically peters off in the 10^-7 but gets there quickly and deals with water vapor quite well. Turbos can get down to MBE territory, which is your 10^-8. Of course this is all system dependent, so I'm not arguing your results at all!

I think cryo's can go lower but typically they're tuned for quick pump-down speeds from roughing rather than ultimate base pressure.

https://people.rit.edu/vwlsps/LabTech/Pumps.pdf

Diffusion pumps are a trip when you think about their mechanism of action. Then again, molecular flow is not necessarily the most intuitive thing.
 
Last edited:
That seems backwards in terms of the pumping efficacy of turbomolecular pumps vs cryo. The latter typically peters off in the 10^-7 but gets there quickly and deals with water vapor quite well. Turbos can get down to MBE territory, which is your 10^-8. Of course this is all system dependent, so I'm not arguing your results at all!

I think cryo's can go lower but typically they're tuned for quick pump-down speeds from roughing rather than ultimate base pressure...

Sorry to have been incomplete, the missing variables in the statement I made were the volume of chamber being pumped as well as the target (for a plasma metalizer) or source (for a vapor phase metalizer) to substrate distance. The critical factor in successful metalizing is ensuring the atoms being metalized arrive at the substrate with enough energy to bond to it. In a practical vacuum there are atoms floating around to intercept the desired metal atom, causing it to lose kinetic energy or even return it back to the target. The average distance the desired atom travels before impacting another is called the MFP (Mean Free Path), so the closer the target and substrate are or the lower the pressure, the fewer impacts the metal atom will see on its way to the substrate. In all practical metalizers the required pressure is inversely proportional to the target to substrate distance, so chamber design is the determining factor in cycle time and required pressure.

In-line optical disc manufacturing uses plasma metalizers with exceedingly short cycle times, under 2 seconds. In this time the chamber has to be opened, loaded, roughed with a mechanical pump, crossed over to turbo pump for fine vacuum, argon vented, plasma struck and then continuously maintained against arcing with an ultra-fast arc detecting 20 KW HV supply for 200 mS, then vented and unloaded, all in under 2 seconds. In order to achieve these cycle times the chamber is barely larger than the disc substrate itself, accounting for the fast pumping speeds. With less than a 10 mm target to source distance, a much higher pressure can be tolerated given the MFP at the pressures of ~10^-2 torr.

In the 24" dia bell-jar metalizers we used, the glass mastering substrates are metalized three at a time using vapor-phase nickel by heating a tungsten source boat. The source to substrate distance is nearly 400 mm, and the chamber volume so great it took two minutes for a mechanical pump to rough the chamber to crossover. If I remember correctly it took another 5 minutes for the cryopump to achieve the 10^-6 need to metalize.

At these pressures there is little atom-to-atom interaction which is usually described as the mechanism behind gas pressure. It is more of a statistical ballistic issue whether or when a molecule will bounce into the cryo and stick as it gives up it's kinetic energy.

Sorry if this seems off topic, but this technology is utilized in the manufacturing of many parts of audio hardware.
 
Last edited:
That seems backwards in terms of the pumping efficacy of turbomolecular pumps vs cryo. The latter typically peters off in the 10^-7 but gets there quickly and deals with water vapor quite well...

The reason for a LN H2O pre-cryo trap was to increase the length of time between cryo regenerations which takes a few hours during which the metalizer is out of service. H2O vapor is the major constituent which condenses out of the air. With the LN trap, after a few weeks of cryo operation it would be holding several thousand liters of condensed air constituents. Indeed we evacuated the clean rooms due to the large volume of nitrogen initially evolved: liquid nitrogen has a 1:700 volumetric expansion ratio when it returns to room temperature gas and we didn't want to asphyxiate the operators. By using the LN H2O trap we were able to run the metalizer 24/7 for a few weeks between regens as opposed to a few days without. Mass production demands very specific process procedures for sure!

Cheers!
Howie
 
Last edited:
scott wurcer said:
Without particles/molecules what does degree of freedom mean?
Oscillation mode?

I take your point. Maybe 'fluid world' is so different that we cannot really imagine it. Perhaps at some deep level shot noise and thermal noise are connected, but one depends on charge carrier charge and the other depends on temperature and these two are not obviously connected.
 
Takes me back to a Balzers CDI600, more than 10 meters long and had 5 turbopumps. Twice I melted a hole in a target with it. Easy to forget the amount energy that is involved during the process of sputtering.

Hi Mark!

Were you involved in the optical disc industry? Those disc sputtering metalizers are wonders for sure...inputting tens of KW into that tiny toroidal plasma required those targets to be water cooled so they wouldn't melt down...and don't get me started on sputtering silicon for the partly reflective DVD-9 L0...

Cheers!
Howie
 
Perhaps he worked at ODM. It was later acquired by ODME from Philips.
I did spend sometime in Oliphant, Commerce and Minomonie.

Not a fan of ODC mastering. Fuji had serious problems with theirs. They had to shutdown for almost a year to fix high jitter that I had discovered while working at one of their customers.
 
Last edited:
Status
Not open for further replies.