Practical Considerations for LN2 cooled, O-Ring Sealed, Vacuum Insulated Dewars for Optical and IR Detectors

Bruce Atwood and Thomas P. OBrien

Imaging Sciences Laboratory

Astronomy Department

The Ohio State University

Columbus, Ohio 43210-1106


We consider the problem of vacuum insulation in O-ring sealed, LN2 cooled Dewars for optical and IR detectors. In addition to the obvious requirement of basic vacuum integrity, some form of Cryo-sorption pump is essential to maintain effective vacuum insulation in these systems. Our experience is with Cryo-sorption pumps based on Zeolites and activated charcoal although other Cryo-sorption materials may exist. Water has an important role in these systems, and will receive special attention even though its partial pressure is completely irrelevant at LN2 temperatures.


The Mathcad® program given in the appendix demonstrates that it is only necessary to assume a binding energy (or heat of adsorption) for a gas onto a surface and an effective surface area to calculate the properties of a Cryo-sorption pumping material such as Zeolites or activated charcoal. The basic model is that the binding energy leads directly to a temperature dependent residence time which combined with the frequency of collisions between gas molecules and the adsorbent surface leads to a quantity of "trapped" gas. That quantity of gas varies with pressure in the same way as it would in a virtual volume. The Cryo-sorption material does not pump in the sense of removing gas from the system but lowers the pressure by sharing the gas between the real and virtual volumes. This virtual-volume model is very successful in predicting the performance of Cryo-sorption pumps.

The virtual volume model does not address the rates of adsorption and desorption. Two rates that have particularly important practical consequences are the uptake of water and the release of air. The binding energy for water in zeolite is so high that, at temperatures below about 200 C, adsorbed water is essentially permanent. When exposed to STP and normal humidity the surface of Zeolites will be covered with water, saturating the Zeolite, and therefore preventing other gasses from binding to the surface. It is therefore necessary to transfer the dry zeolite to the Dewar in such a fashion that it is not exposed to ambient humidity for too long a period. We have measured the rate of weight gain of previously dried Zeolites held in ambient air and find that typical time constants for water absorption into canisters of a few ten of cm3 are of the order of 10 hours. The binding energy for water on charcoal is low enough that it will quickly dry under vacuum at room temperature. The other important rate is how quickly absorbents release air when they are put in a vacuum. We have measured time constants of approximately an hour for air release of typical Zeolite systems.


Air diffuses through the O-rings into a "sealed" Dewar at a continuous rate after equilibrium has been established. This rate is a function of the total length of O-rings in the Dewar, the type of rubber, the "squeeze" of the O-ring, and the pressure across the O-ring. The diffusion rate is nearly independent of the O-ring cross-section since the diffusion increases linearly with the area exposed to air and decreases linearly with the distance that the air has to diffuse.

The Parker O-Ring Handbook provides data on the gas permeability of air through various O-ring materials. Butyl rubber has the lowest permeability with a value of about 210-9 Std cm3/(cm sec). Nitrile rubber (Buna-N), which is more widely available, has permeability of about 510-9 Std cm3/(cm sec). The amount the O-ring is squeezed and greased have a minor affect on the diffusion rate. The appendix shows the calculation of diffusion into any Dewar. Our instruments fall into two classes. Large instrument Dewars, approximately meter in diameter by 1 meter long, with O-rings for the main vacuum shell, a window, a detector port, and a dozen other small ports, have a diffusion rate of 510-5 Std cm3 sec-1. This produces a steady rise in pressure for an un-pumped Dewar of about 5 millitorr/day. The diffusion rate for small CCD Dewars of 20 cm diameter by 35cm long with two main O-rings and several small O-rings is about 10-5 Std cm3 sec-1 giving an un-pumped pressure rise of about 35 millitorr/day. (In practice we generally see actual diffusion rates that are two to four times lower than these theoretical rates.) Note that this calculation only includes the diffusion of air through the O-Rings and does not include pressure rises from the desorption of gases from either the walls of the Dewar nor from any cryo-sorption material that might be present. No allowance had been made for any "virtual leak", that is the slow escape of gas from a nearly sealed sub volume in the vacuum space. Trapped volumes and virtual leaks are not a problem since even 110-5 Std cm3 sec-1 corresponds to almost one cm3 of STP gas per day. So as long as the trapped volumes are only a few cm3 at most only a few days operation are lost. Note that these leak rates of 1 STP cm3 day-1 are the best that can be done with O-ring seals . Furthermore, in a Dewar that is reasonably clean, one that has been wiped with acetone for example, water is the only significant species to evolve from the interior surfaces of the Dewar and water is very effectively pumped as long at the Dewar is cooled with LN2.


Water is the principal contaminate in most vacuum system because all materials adsorb some water when stored in an environment with typical humidity level. Most materials adsorb a LOT of water. The time constant for water to desorb from room temperature surfaces in a vacuum can be very long, hundreds of hours for aluminum surfaces and the age of the universe for Zeolites.

If a new "wet" Dewar is quickly pumped and cooled with LN2 the water that desorbs from the warm surfaces will be "pumped" by the cold surfaces. LN2 cooled Zeolite can be used to pump the air that diffuses through the O-rings since the water that would otherwise saturate the Zeolite will be condensed on the LN2 cooled surfaces. When the Dewar is warmed, however, all the water that had been frozen on the LN2 cooled surfaces will be quickly pumped by the Zeolite, even though the Zeolite is warm. If the Dewar is so designed it can be warmed to room temperature, backfilled with dry gas, and the Zeolite changed, removing the water from the Dewar. In typical aluminum CCD Dewars with a few liters of vacuum, changing the Zeolite a few times over the first few months of operation will effectively dry the Dewar. Further Zeolite changes are then unnecessary as long as the Dewar is not exposed to atmospheric humidity since the air that diffuses through the O-rings is desorbed from the zeolite when warm and can be pumped from the Dewar without changing the Zeolite.

The situation is quite different if charcoal is used. As with Zeolite the water will be pumped by the LN2 cooled surfaces but when the Dewar is warmed the partial pressure of water will rise to saturation. If the Dewar is left untouched the water will be slowly re-adsorbed in the interior surfaces of the Dewar and the partial pressure of water will return to roughly the atmospheric value before the Dewar was pumped. A dramatic situation will occur if a charcoal pumped Dewar is vented to air shortly after it is warmed. The partial pressure of water will now be at saturation for room temperature. When the partial pressure of water in the introduced air is added super-saturation occurs, leading to cloud formation. The cloud will rapidly condense on the interior surfaces, including any optics or detectors present. A similar transient situation will occur with any object in the Dewar that has a thermal time constant considerably longer that the bulk of the cold surfaces. During warm up any water that is not pumped by Zeolite will condense on these cold objects until they warm.


Linde type 4A or 5A Zeolites have a capacity for adsorbing water at room temperature equal to about 15% by weight. This adsorption process is fairly rapid, resulting in totally saturated Zeolite in less than 24 hours when exposed to air with a relative humidity of about 40%. This water adsorption reduces the capacity of the Zeolite for Cryo-pumping air to nearly zero. The Zeolite must be regenerated by baking at 350 C for several hours and then stored in an airtight container prior to Cryo-pumping service. Temperatures higher than 350 C will damage the Zeolite.

An advantage of this large capacity for water is that Zeolite acts as a very effective desiccant for drying Dewars. The operational disadvantage is that the Zeolite must be replaced every time the interior of a Dewar is exposed to ambient conditions for more that a few tens of minutes.

Any organic solvents adsorbed by the Zeolite will be driven off during the 350 C regeneration cycle required to desorb the water.

Activated charcoal, such as EM Scientific #CX0640-1 activated coconut charcoal, available through Fischer Scientific, has very little capacity to adsorb water at room temperature. This lack of water capacity means that it provides no desiccation or drying function in the Dewar.

Under some circumstances it is an operational convenience that the charcoal does not require replacement or regeneration after a Dewar is cycled to room temperature and exposed to air.

The detailed computation of the Cryo-pumping capacity of charcoal is given in the appendix. The heat of adsorption and surface area constants are assumed to be the same as for Zeolite as no references for these constants were found. These values are consistent with our experience with charcoal Cryo-sorption systems. It is worth noting that typical virtual volumes are 10 8 cm3 for each gram of adsorbent.


As is calculated in the appendix the pressure in an evacuated Dewar without pumping will typically rise several millitorr each day. It is frequently convenient for Dewars to operate for a year without service. Thus the Cryo-sorption pumping must be sufficient to lower the pressure in the Dewar from pressures of the order of one torr to values where the thermal conductivity of the residual gas is negligible, on the order of 10-5 Torr. It is therefore not necessary to rough pump the Dewar to very low pressures before cooling. For each few millitorr of gas left in the Dewar before cooling only one day is lost from the maximum run length. In order to have the maximum capacity of the Cryo-sorption pump it is necessary to wait for the air to desorb from the Zeolite or charcoal. It is not necessary, however, to pump the Dewar continuously during this time. If fact, long periods of pumping run the risk of back-streaming of oils from the pump into the Dewar with possible contamination of the detector or optics. For a CCD Dewar a very effective protocol is ten minutes of pumping, valving off the Dewar for 24 hours followed by an additional 10 minutes of pumping. For a new Dewar, or one that has never been desiccated by changing the Zeolite (or by aggressive baking) the pressure will be dominated by water vapor below about 500 millitor.

If it is desirable to pump and cool a Dewar quickly, as is the case for some detectors that require UV flooding to increase the sensitivity, a larger diameter pump line and valve will yield much greater gains than a higher performance pump. Even if the largest, and most expensive, turbo-molecular pump can maintain 10-7 Torr at its inlet the pressure inside the Dewar is limited by the impedance of the line and valves. Simple mechanical pumps can quickly remove the air in a Dewar and if they are not operated for long periods in the molecular flow pressures, that is below about 100 millitor for a system with 15 mm diameter lines, contamination of the Dewar by pump fluids is all but nonexistent. Since it would be very unusual that that one is willing to wait the hundreds of hours that are required for substantially all the water to desorb from the interior, drying the Dewar by changing a Zeolite canister is a much better strategy.


To accurately monitor the pressure in a Dewar with a Cryo-sorption pump it is necessary to be able to measure pressures from 10-7 Torr to 10-3 Torr. Ionization gauges provide a very accurate, although not inexpensive, method for making measurements in this range. (Cold cathode gauges, while less expensive, require some maintenance, are much less precise, and are prone to operational problems such as difficulty starting at low pressures.) Typical measurements of interest are the initial pressure after a cool-down and monitoring of the long term pressure rise in the Dewar due to gas diffusion into the Dewar. Regular readings of the pressure in a cold Dewar can indicate exactly when a Dewar will require pumping to remove the air that has diffused through the O-rings. We normally equip CCD Dewars with a Convectron Gauges (available from Granville Phillips) which are reliable from atmospheric pressure to 1 millitor, (Thermocouple gauges are less expensive but are useful only over the pressure range from 1000 to 10 millitorr.) We install both a Convectron and an Ionization gauge on our large instruments.

The total heat load on a vacuum insulated Dewar can be measured by monitoring the flowrate of nitrogen boil-off gas. A simple and robust Rotometer type (ball suspended in a tapered tube) gas flowmeter available from Dwyer Instruments provides adequate accuracy. Once the typical baseline boil-off flowrate has been determined any increased conductive heat load caused by pressures > a few 10-5 Torr will be clearly indicated by and increased flow. Thus the flowmeter can indicate a problem with the vacuum insulation long before a Convectron or thermocouple gauge can measure the increased pressure. The flowmeter is also a valuable tool during system development, allowing a direct measurement of the effect of any change on total power input to the Dewar and directly indicating the hold time. If pumping and cool-down are done with a fixed protocol the N2 boil-off rate as a function of time provides an inexpensive, early, and robust indication that the Dewar system is functioning normally.

The flowmeter should be calibrated in "Standard Cubic Feet Air/Hour" because this fortuitously reads directly (within a few percent ) in liters/day of LN2 consumed, a very convenient unit. The directly read value can be multiplied by 2 to convert to watts.

Dewars which have been open to air desorb a substantial amount of water during pump down. Since the mechanism for pumping water and air are so different it is useful to be able to measure the partial pressure of water and partial pressure of air separately. If a Dewar is rough pumped until the pressure drops below a few hundred millitorr a Convectron or thermocouple gauge will read the total pressure reading. Adding a small amount of LN2, just a whiff, will pump virtually all of the gaseous water and produce a rapid drop in pressure. After a minute or two the pressure will stabilize, the final value is a reasonably accurate measure of the partial pressure of air in the Dewar.


The OSIRIS instrument has been in the field for six years and has used a Zeolite molecular sieve canister as a pump. The canister contains about 35 grams of Linde Type 4A and 5A Zeolite. The canister is mounted to the LN2 cooled optical bench in a cold radiation environment. This pump has worked well in practice and has maintained high vacuum in the Dewar for runs of several months duration.

The canister is accessible through an access port in the instrument and the Zeolite must be replaced every time the instrument is exposed to ambient humidity. This is because of the irreversible adsorption of water by the Zeolite at room temperature which renders the Zeolite ineffective for pumping air. The requirement for good access to the canister necessitated the location of the canister at the "dry" end of the LN2 reservoir, away from the liquid LN2. The result is that the canister is typically at temperatures as high as 85 Kelvin which is not optimal.

The TIFKAM instrument has a Dewar similar to OSIRIS but uses an activated charcoal canister mounted inside the cold volume of the Dewar at the "wet" end of the LN2 reservoir and is therefore at very nearly 77 Kelvin. The canister contains 70 grams of coconut activated charcoal. Since charcoal doesn't irreversibly adsorb water, the charcoal does not need to be replaced every time the instrument is warmed up. The TIFKAM instrument also has an ionization gauge which allows accurate pressure measurements to be made. TIFKAM is typically pumped warm to a total pressure of about 400 millitorr which will whiff to approximately 50 millitor. Two hours after filling with LN2 the pressure is typically 2x10-6 Torr and after 24 hours the pressure is about 4x10-7 Torr.

The pressure will then rise very slowly over a long run. Periodic pressure measurements with the ionization gauge can accurately monitor the vacuum performance of the Dewar. The charcoal will maintain high vacuum (P < 10-5 Torr) for more than a year.

Our current CCD Dewar design has a cold charcoal canister and an easy to change warm Zeolite canister. Fresh Zeolite will hold the water partial pressure below 20 millitorr and guarantee that no condensation will form. For detectors and optics that are sensitive to water (or just very expensive) good practice is to use heaters to warm them to room temperature before the LN2 is depleted.

The appendix provides a detailed analysis of the charcoal pump performance, gas diffusion through O- rings, initial pump down pressure, and charcoal service time for the ANDICAM instrument. This is contained within a MATHCAD 6.0 document which available for download at our website,