Operation of an ADR using helium exchange gas as a substitute fora failed heat switch
P. Shirron a,⇑, M. DiPirro a, M. Kimball a, G. Sneiderman a, F.S. Porter a, C. Kilbourne a, R. Kelley a,R. Fujimoto b, S. Yoshida c, Y. Takei d, K. Mitsuda d
aNASA/Goddard Space Flight Center, Greenbelt, MD 20771, USAb Institute of Science and Engineering, Kanazawa University, Ishikawa 920-1192, Japanc Sumitomo Heavy Industries (SHI), Niihama-shi, Ehime 792-0001, Japand Institute of Space and Astronautical Science (ISAS), Japan Aerospace Exploration Agency (JAXA), Sagamihara, Kanagawa 252-5210, Japan
a r t i c l e i n f o
Article history:Available online 26 April 2014
Keywords:Adiabatic demagnetizationADRSpace cryogenicsExchange gas
a b s t r a c t
The Soft X-ray Spectrometer (SXS) is one of four instruments on the Japanese Astro-H mission, which iscurrently planned for launch in late 2015. The SXS will perform imaging spectroscopy in the soft X-rayband (0.3–12 keV) using a 6 � 6 pixel array of microcalorimeters cooled to 50 mK. The detectors arecooled by a 3-stage adiabatic demagnetization refrigerator (ADR) that rejects heat to either a superfluidhelium tank (at 1.2 K) or to a 4.5 K Joule–Thomson (JT) cryocooler. Four gas-gap heat switches are used inthe assembly to manage heat flow between the ADR stages and the heat sinks. The engineering model(EM) ADR was assembled and performance tested at NASA/GSFC in November 2011, and subsequentlyinstalled in the EM dewar at Sumitomo Heavy Industries, Japan. During the first cooldown in July2012, a failure of the heat switch that linked the two colder stages of the ADR to the helium tank wasobserved. Operation of the ADR requires some mechanism for thermally linking the salt pills to the heatsink, and then thermally isolating them. With the failed heat switch unable to perform this function, analternate plan was devised which used carefully controlled amounts of exchange gas in the dewar’s guardvacuum to facilitate heat exchange. The process was successfully demonstrated in November 2012,allowing the ADR to cool the detectors to 50 mK for hold times in excess of 10 h. This paper describesthe exchange-gas-assisted recycling process, and the strategies used to avoid helium contamination ofthe detectors at low temperature.
Published by Elsevier Ltd.
1. Introduction
Astro-H [1], Japan’s sixth X-ray astronomy mission, is currentlythe only major X-ray facility under development for use in space.Its four instruments will make observations across the X-ray spec-trum (0.3–600 keV) to investigate such topics as how matterbehaves in extreme gravitational fields, how the largest structuresin the Universe grew, and the spin rate of black holes. The mostsensitive instrument is the Soft X-ray Spectrometer (SXS) [2],which will perform imaging spectroscopy with a resolution of bet-ter than 7 eV in the 0.3–12 keV band. This will be accomplishedwith a 6 � 6 array of microcalorimeters operating at 50 mK. Thearray is cooled by a 3-stage ADR [3] which thermally connects toboth a 40 l superfluid helium tank and a 4.5 K Joule–Thomson cryo-cooler. This configuration allows the ADR to operate using the
liquid helium as a heat sink, and the JT cooler after the helium isdepleted. The basis of the ADR’s design was a hold time of 24 hat 50 mK, and a recycle time of 1–2 h. These values are consistentwith a system level observing efficiency of >90%.
The SXS dewar, including the cryocoolers, was designed andbuilt by Sumitomo Heavy Industries, Inc., (SHI) Japan, and thedetector array and ADR (collectively referred to as the CalorimeterSpectrometer Insert (CSI)), were designed and built by NASA/GSFC.The SXS instrument has significant heritage in the X-Ray Spectrom-eter (XRS) instrument on Astro-E [5] and Astro-E2 [6]. The majordifference is that the solid neon dewar used for both XRS instru-ments has been replaced by a complex of Stirling and JT cryocool-ers. This was done to achieve two objectives: redundancy in thecryogenic system, and longer mission life.
The extreme sensitivity of the SXS detectors, and the criticalityof the SXS instrument to Astro-H mission science, led the project topropose building a full engineering model (EM) instrumentthat could be used to investigate possible coupling between the
http://dx.doi.org/10.1016/j.cryogenics.2014.04.0110011-2275/Published by Elsevier Ltd.
⇑ Corresponding author.E-mail address: [emailprotected] (P. Shirron).
Cryogenics 64 (2014) 207–212
Contents lists available at ScienceDirect
Cryogenics
journal homepage: www.elsevier .com/locate /cryogenics
https://ntrs.nasa.gov/search.jsp?R=20150010114 2018-06-14T17:21:43+00:00Z
cryocoolers and the detector signals – from electromagneticinterference, exported vibration or other mechanisms. The EMprogram included extensive testing of the SXS instrument, bothas an isolated subsystem and after integration onto the spacecraft.The latter provided an opportunity to investigate interferencewith the SXS detectors and other instruments, particularly thecryocooler used on the Soft X-ray Imager (SXI) instrument, andspacecraft components such as the reaction wheels. Correctiveaction could be taken prior to committing to the flight instrumentdesign.
In late 2011, the EM CSI was assembled and subjected to qual-ification level consisting of pre-vibe performance testing, vibrationtesting, and post-vibe performance testing. At the time, there wereno obvious changes in performance that could not be attributed todisassembly and re-assembly of the 2-stage ADR just prior tovibration testing. The EM CSI was delivered in March 2012 to SHIfor integration into the EM dewar. The first cooldown of the EMinstrument occurred in July 2012.
During the cooldown, one of the ADR’s four gas-gap heatswitches was observed to thermally decouple as the entire assem-bly slowly cooled through about 100 K. The faulty heat switch (HS2in Fig. 1) links the ADR’s 2nd stage to the helium tank. The behaviorof the heat switch suggested that its containment shell, consistingof an outer 2-ply layer of T300 carbon fiber composite and an inner12 micron foil of Ti15333, had developed a significant permeationleak, and that all of the initial helium-3 charge had been lost andreplaced by some fraction of an atmosphere of air. In any event,the ADR could not operate without the function of this heat switch,and the whole EM program was at risk.
Since the EM SXS hardware was also needed for spacecraft-leveltests such as EMI/EMC compatibility testing and vibration qualifi-cation testing, and the schedule for these had little to no slack, dis-assembly of the dewar and repair or replacement of the failed heatswitch was not practical. Therefore, any and all options for recov-ering some operation of the CSI were developed, including the useof helium exchange gas in the dewar’s guard vacuum to thermallycouple across the failed switch. While the latter is clearly a non-standard operating mode for a high-performance dewar system,it represented – ultimately – the best and lowest risk approachto enabling the EM program to move forward. This paper describesthe exchange gas-assisted ADR recycling process, and the results ofits successful implementation.
2. SXS cryogenic system
The SXS cryogenic system [4], shown schematically in Fig. 1,consists of a small (40 l) helium tank, a 4.5 K JT cryocooler, twopairs of 2-stage Stirling cryocoolers, a 3-stage ADR, and the detec-tor assembly. Two of the Stirling cryocoolers connect to an innerand an outer vapor cooled shield (IVCS and OVCS), which are alsocooled by the helium boiloff gas. The other two Stirling cryocoolersact as pre-coolers for the JT system, which also cools a shieldimmediately surrounding the helium tank. The ADR and detectorassembly are an integral unit that mounts to the top of the heliumtank. The ADR is functionally divided into a 2-stage ADR (inside the1.2 K shield) which cools the detectors using the liquid helium as aheat sink, and a 3rd stage ADR which pumps heat from the heliumtank to the JT cooler.
With the failed heat switch, the salt pills in the 2-stage ADR donot have a direct, high-conductance path to the heat sink. Instead,they are thermally isolated from the helium tank except for theweak connection through the containment shell of heat switchHS2, sensor wiring, and Kevlar suspension components. This con-figuration is schematically shown in Fig. 2.
2.1. Possible recovery options
After discovery of the heat switch failure, three options forrecovering some operation of the ADR were developed andevaluated.
Option A involved using the weak residual thermal couplingbetween the ADR and the helium tank to recycle the ADR. In thiscase, the ADR’s salt pills would be magnetized to full field, warm-ing them to as high as 20 K, then they would eventually cool to atemperature at which the ADR could be demagnetized to 50 mK.The stages would have to cool to at least 5 K, at which point the2nd stage could precool the 1st stage to about 2 K, and then the1st stage could reach 50 mK with about 2 h of hold time – severelylimited by the high boundary temperature of the 2nd stage and thehigh heat flow through HS1. Based on measurements of heat flowversus salt pill temperature, the time needed to reach 5 K was esti-mated to be 20 days.
In Option B, exchange gas would be admitted to the dewar’sguard vacuum to enhance the thermal contact between the saltpills and components anchored to the helium tank – principally
Fig. 1. Schematic representation of the SXS cryogenic system.
208 P. Shirron et al. / Cryogenics 64 (2014) 207–212
their magnets. The ADR’s salt pills would be magnetized andcooled to close to the helium bath temperature. After eliminatingthe exchange gas, the ADR could be progressively cooled as out-lined in Option A. If the salt pills could be cooled to about 3 K at fullmagnetic field, the systems cooling capacity would be sufficient toallow it to reach 50 mK with more than 10 h of hold time. Based onestimates of gas conductance in the �1 x 10-3 Pa ra-oled to 3 K inless than 1 h.
The concept of Option C was to backfill the dewar’s guard vac-uum with up to 1 atmosphere of helium-4 and allow up to 1 weekfor helium to diffuse into the switch, possibly recovering full oper-ation of the ADR. The main difficulty for this option was to boundthe helium leak/permeation rate through the shell and assesswhether the leakage would occur only near room temperature(characteristic of permeation) or at all temperatures (indicativeof a leak). In either case, any helium injected into the switch wouldpartially leak back out during the long period needed to clean upthe guard vacuum. With a large enough leak (permeation or directleak), the helium charge could dissipate before the ADR could becooled down. If a direct leak were present, the heat switch mightcontinue to degrade at low temperature at too fast a rate to provideuseful function.
The relative merits of each option involved both objective mea-sures of how well the ADR might perform, and subjective assess-ments of risk or compromise to the entire SXS system. Option Awould yield the poorest ADR performance, but except for the needto run the ADR magnets continuously at full current for weeks, theoperation imposed no apparent risk to the SXS hardware. Option Bsignificantly increased the amount of ‘‘cold time’’ that could beachieved, but risked contaminating the guard vacuum in a waythat could seriously degrade detector performance at 50 mK, andpossibly compromise future measurements of the dewar’s thermalperformance. Option C could conceivably restore the ADR to fulloperation, but it might also yield nothing if the leak was too large.In either case, it was certain to heavily contaminate the guardvacuum.
The decision ultimately depended on the discovery of a verysmall helium leak in the dewar itself, which appeared after anextended time at superfluid conditions. At that point, the deliber-ate introduction of exchange gas for Options B and C no longer rep-resented an irreversible compromise to the guard vacuum. OptionB was chosen for implementation based on its higher likelihood ofsuccess and the lower likelihood of helium contamination once thedetectors were cooled.
3. ADR cycling using exchange gas
3.1. Gas thermal conduction
Referring to Fig. 2, the introduction of helium exchange gas intothe guard vacuum of a dewar containing an ADR will, at some level,
thermally connect the salt pills to surrounding components. Thestrongest coupling will be to the magnets and magnetic shieldswhich surround the salt pills, and from which they are separatedby a uniformly small gap. For heat flow at low temperature in themolecular regime, we use an expression by Kumagai et al. [7],although similar formulations may be found in White and Meeson[8]. Heat flow per unit area between surfaces at T1 and T2 is
_QA¼ 1
2cþ 1c� 1
� � ffiffiffiffiffiffiffiffiffiffiffikB
2pm
rT1 � T2
ð ffiffiffiffiffiT1
p þ ffiffiffiffiffiT2
p Þ=2 PRT ð1Þ
where c is the ratio of specific heats at constant pressure and vol-ume, m is the mass of atoms/molecules of the gas, and PRT is thepressure in the system, measured at room temperature. In SI units,where pressure is given in Pa and temperature in K, the expression,for helium gas, reduces to
_QA¼ ð74:0 W=m2Þ T1 � T2ffiffiffiffiffi
T1p þ ffiffiffiffiffi
T2p PRT ð2Þ
This expression specifies the maximum possible heat flowbetween the surfaces, since transfer of heat requires the gas tostick to each surface it strikes. In practice, this expression is foundto closely approximate heat transfer between the salt pills andmagnets, possibly due to the small gap which enhances the fre-quency of bounces if a helium atom is not adsorbed.
For example, for a pressure of 10�3 Pa, and a salt pill at 5 Kinside a magnet at 3 K, the heat flow per unit area would be37 mW/m2. The 1st stage salt pill has a surface area of 0.027 m2,and the 2nd stage salt pill 0.015 m2. The combined heat flow fromboth salt pills can exceed 1.5 mW. At this rate, the several Joules ofheat that must be rejected from the ADR can be transferred in lessthan 1 h.
However, the exchange gas causes heat flow between othercomponents of the system, particularly the shields connected tothe cryocoolers. At too high a pressure, the heat flow will over-whelm their cooling power. Thus a balance of adequate heatexchange for the salt pills and acceptable heat loads on the cryoco-olers and helium tank demanded that the pressure be kept in the1–2 � 10�3 Pa range.
In fact, because the JT cooler shield (at 4.5 K and >1 m2 area)directly faces a shield at �26 K, even for pressures as low as10�4 Pa, its heat load would exceed the approximately 20 mW ofexcess cooling capacity. For this reason, Option B was not consid-ered feasible without liquid helium in the tank to stabilize thesystem.
3.2. Helium contamination
While the exchange gas is clearly vital to charging the ADR, itmust be eliminated before cooling to sub-kelvin temperatures.Any residual gas in the guard vacuum would condense onto theADR and detectors as they cooled below the tank temperature. Sur-face coverages of even one monolayer would severely degrade theresolution of the detectors through the added heat capacity.
Pumping on the guard vacuum is eventually effective, buteven in the most open systems, several days would be requiredto reach sufficiently low pressure. The tightly layered shieldsand MLI in the SXS dewar necessitated a different approach:the use of the helium tank as a gettering surface for the exchangegas. The vapor pressure above the adsorbed film on the coldsurfaces is an extremely strong function of temperature, andvery modest cooling of the helium tank will reduce the pressureto levels that are undetectable on either an external pressuregauge or a helium leak detector.
This can be seen by casting the chemical potential, l, of theadsorbed film in the Frenkel–Halsey–Hill [9] formulation as
Fig. 2. Effective thermal schematic of the SXS helium tank and 2-stage ADR withthe failed heat switch. Heat switch HS1, and the two salt pills (S1 and S2) and theirmagnets remained functional, but thermally decoupled from the helium tank.
P. Shirron et al. / Cryogenics 64 (2014) 207–212 209
P ¼ Psvpe�l=kBT ð3Þwhere P is the pressure above the film and Psvp is the saturatedvapor pressure at temperature T. Both terms in Eq. (3) have an expo-nential dependence on T, resulting in P potentially having an extre-mely strong dependence on temperature. The dependence can befar less dramatic if the film thickness, d, increases significantly ashelium gas is adsorbed, given that l–d�3 [10]. However, in theregime required for this operation, vapor pressures are necessarilysmall, on the order of 10�8 atm. Even with a very conservative esti-mate of the effective (i.e. cold) volume of the guard vacuum, theamount of He in the gas phase is less than 1% of the amountadsorbed onto the cold surfaces, and therefore the film thicknessdoes not increase appreciably as gas is adsorbed.
The effect on vapor pressure of reducing tank temperature canbe seen in Fig. 4, in which an external pressure gauge was usedto monitor guard vacuum pressure after dosing approximately 1standard cm3 of He gas into the engineering model dewar. Coolingthe tank by only 10–15% produced an order of magnitude changein vapor pressure, and when cooling below 2 K, the helium signalfrom the dewar was undetectable.
The strong dependence of vapor pressure on temperature alsomeans that the amount of gas absorbed onto the tank when theexchange gas is present is a strong function of temperature. Conse-quently, there was concern that if the temperature chosen fortransferring heat from the ADR was too low, the thicker film wouldnot allow the vapor pressure to be reduced low enough to preventcontamination of the detectors at 50–60 mK. On the other hand, ifthe transfer temperature is too high, the cooling capacity achievedmight not be enough to reach 50 mK with acceptable hold time. Asa compromise, cycling was performed at 3 K, and the tank was sub-sequently pumped to less than 1.3 K.
After cooling the tank, a very thin film of helium would almostcertainly still be adsorbed on the detectors, and potentially enoughto degrade their performance through the added heat capacity.Consequently, a ‘‘degassing’’ operation was included in which thedetectors were warmed above 6 K for 10 min before demagnetiza-tion to low temperature. In the high vacuum environment pro-duced by the pumped helium tank, the helium coverage on thedetectors at 6 K was calculated to be negligible – a fact that wasconfirmed during the first ADR cycle.
3.3. Complete ADR cycling process
The ADR cycling process itself consisted of only a few basicsteps. The starting point was to pump the helium tank to 3 K, asmeasured at the top where the ADR is located. The salt pills weretypically about 4–5 K at the end of the dewar cooldown. In caseswhere they were warmer, they began cooling once the dosing ofexchange gas began, and cooled below 5 K in less than 30 min.
Exchange gas was dosed into the guard vacuum until the pres-sure on an external gauge read steadily at the desired value, as evi-denced by steady cooling of the salt pills toward 3 K. For themajority of cycles performed, the target value was 2 � 10�3 Pa.One of the complicating factors was that the exchange gas causedthe top of the tank, which is not in direct contact with liquid, towarm relative to the bottom. Over time, the gradient tended toincrease, causing exchange gas to adsorb onto the tank bottomand reduce the heat flow. Careful control of the pumping ratewas needed to keep the tank top at 3 K, and a mass gauge heaterat the bottom of the tank was powered in order to minimize thegradient across the liquid.
The controllers for the two ADR stages were set to control thesalt pill temperatures at 5 K. At this point, turning on HS1 is advan-tageous since far more heat is generated by the magnetization ofthe 2nd stage, and the 1st stage has the larger surface area and heat
flow capability. Magnetization continues until the magnets reachfull current, and the salt pills cool close to 3 K. The helium tankis then pumped to <1.3 K, and the exchange gas is completelyadsorbed onto the tank surface. Fig. 3 shows the pressure readout of the external gauge during this process.
At this point, the entropy (or cooling) capacity of the ADR isfixed. That is, the maximum hold time for the detectors at low Tis now determined. In subsequent steps, the cooling capacity ismoved between the stages in order to accomplish two objectives:to degas any residual helium from the detectors, and to prepare the1st stage for final demagnetization cooling. Cooling capacity istransferred between stages by using one of the stages to the otherdown, using each stage’s temperature controller to maintain amoderate temperature difference (to maintain high transfer effi-ciency) across HS1.
To perform the degassing operation, (1) the 1st stage transfersall of its cooling capacity to the 2nd stage through HS1, (2) HS1is opened, (3) the 1st stage is magnetized to full current, whichraises its temperature from �3 K to >6 K, for 10 min, and (4) thestage is demagnetized to zero current, back to �3 K.
At the end of the degassing process, the 2nd stage has essen-tially all of the stored cooling capacity of the 2-stage ADR (chargedto 3 T at typically 2.5 K). The majority of this cooling capacity istransferred back to the 1st stage to set up for its demagnetizationto 50 mK. Heat is flowed between the stages, charging the 1st stageto 2 A/2 T and cooling it to below 2 K. Some capacity is retained inthe 2nd stage so that it can demagnetize to a lower temperatureduring the hold and significantly reduce the parasitic heat flowthrough HS1.
Once the transfer is complete, both stages are demagnetized totheir hold temperatures: 0.9–1.0 K for the 2nd stage and 50–60 mKfor the 1st stage.
The complete recycling process, and the subsequent holdperiod are shown in Figs. 5 and 6. The time required for the fullrecycling process exceeded a normal work shift, so it was dividedinto three periods. The initial transfer of heat from the ADRstages to the helium tank was performed during one work shift.After pumping the helium to <2 K, the ADR was left idle for upto 16 h, during which time some additional cooling occurredthrough the Kevlar suspension and heat switch structures. Duringthe subsequent work shift, the detectors were degassed and theADR was prepared for cooldown. Combined, the recycling processtook up to 30 h to complete – as opposed to 1 h for a normallyoperating ADR.
Since each cycle achieved a hold time of only 9–16 h at 50–60 mK (compared to 42–74 h demonstrated before the failure ofthe heat switch), test sequences were carefully scheduled to makefull use of the available cooling capacity. Fortunately, breaks intesting could be introduced by temporarily warming the ADRand detectors to �0.5 K, where the usage of entropy capacity wasnegligible.
When the 1st stage ADR ran out of cooling capacity, the nextcycle could be started by warming the helium tank back to 3 K.As the tank warms above about 2.4 K (as seen in Fig. 3), theadsorbed exchange gas is released, thermally connecting the saltpills and helium tank.
After a few cycles were performed, some improvements weremade to make the process more time-efficient. Since the transferrate depends on surface area, more heat is rejected from Stage 1than Stage 2. As the salt pills cooled, a faster time-average transferrate could be achieved by periodically transferring the coolingcapacity in Stage 1 to Stage 2 by flowing heat from Stage 2 to Stage1 through HS1, then opening HS1 and magnetizing Stage 1 to ahigher temperature. The process allows both salt pills to be cooledslightly below the tank temperature in about 2 h, as opposed tocooling to within about 0.1 K of the tank in 4–5 h. The oscillation
210 P. Shirron et al. / Cryogenics 64 (2014) 207–212
in Stage 1’s current and temperature before the helium tank iscooled below 3 K reflects this periodic transfer.
4. Results
Recycling the SXS EM ADR using exchange gas was successfullydemonstrated on the first attempt, in November 2012. Since then13 cycles have been completed, each time providing 8–13 h of testtime at 50–60 mK. The cycles have been conducted in groups of 2–3, each time requiring about 1 week of schedule. The primary goalwas to identify coupling, if any, between the cryocoolers anddetectors on the SXS instrument, and secondarily to identify cou-pling between other instruments and spacecraft components. Con-sequently, ADR cycles were conducted with the SXS instrument atseveral different levels of assembly, from the dewar level up to thefull spacecraft level with all Astro-H instruments present.
On the first cycle, after cooling to 50 mK, coupling between thecryocoolers and the SXS detectors was clearly observed. Itappeared as both excess noise in the measurement of X-ray ener-gies – resulting in a degraded resolution of 20–30 eV (for 5.9 keVFe55 X-rays) instead of the expected 4–7 eV – and as thermal noisein both the SXS detector and ADR stage temperatures. The latterled to speculation that the coupling was primarily mechanicalheating of components at 50 mK. What made the mechanical heat-ing problematic was that it exhibited a randomness over timescales of seconds to tens of seconds that could not be dampedout by the ADR’s temperature controller.
Fig. 3. The 2-stage ADR (left) and HS2 (right) with its T300 carbon fiber composite shell. The shell is approximately 8 cm long and 1.8 cm diameter.
Fig. 4. Vapor pressure in the guard vacuum, measured on an external gauge, as afunction of the average tank temperature, for both increasing and decreasingtemperature. The background reading for the gauge was 2 � 10�4 Pa.
Fig. 5. Temperature and current during the recycling portion of one gas-assistedADR cycle. Time is zero at the beginning of the hold time. Subscripts for T and Idenote the respective stage; Ttank is the temperature of the tank where the ADR islocated.
Fig. 6. Temperature and current in the 1st stage during a hold time. Thetemperature setpoint was either 50 mK or 60 mK.
P. Shirron et al. / Cryogenics 64 (2014) 207–212 211
When all cryocoolers were turned off, as seen in Fig. 7, the tem-perature stability immediately improved, nearly to the levels dem-onstrated in subsystem tests at Goddard Space Flight Center beforedelivery to Japan. An effort is now underway to develop mechani-cal isolation structures for the cryocooler compressors to reduceexported vibration, and to minimize the overlap of sensitive fre-quencies for the detector system with peaks in the cryocoolervibration spectra. These will be partially implemented on the engi-neering model hardware, and, if successful, fully implemented inthe flight design.
5. Summary
The failure of one of the four heat switches in the engineeringmodel 3-stage ADR built for the Astro-H SXS instrument led toan exhaustive search for recovery options and evaluation of eachone’s potential to damage or compromise the SXS hardware.Ultimately the option chosen, both for its higher potential forsuccessfully cooling the ADR to 50–60 mK and the ability to obtainsignificantly more test time at low temperature, was to usehelium-4 exchange gas to replace the function of the failedswitch. While it provides beneficial heat exchange between ADRcomponents and the helium tank, the exchange gas also generates
unwanted heat flows within the remainder of the cryogenic sys-tem. Fortunately, pressures in the 1–2 � 10�3 Pa range provideadequate thermal conduction within the ADR assembly to allowrecycling, but not so much conduction among warmer componentsthat the cryocoolers or helium tank are overwhelmed.
The process relied critically on being able to eliminate theexchange gas after charging the ADR stages, and this was accom-plished by cooling the helium tank and adsorbing the gas ontothe tank surface. The residual pressure was low enough that, aftera degassing operation at 6 K, detectors sensitivity at 50–60 mK wasindistinguishable from that observed in performance tests whenthe ADR was functioning normally. Even after extended operationat low temperature, the detectors never showed any sign of heliumloading, which would have been evident if even a very small frac-tion of a monolayer of helium had condensed on their surface.
The successful implementation of this ADR recycling techniqueallowed the main goals of the Astro-H/SXS engineering model pro-gram to be met, including probing coupling between the SXS cryo-coolers and detectors. As a result, significant compatibility issueswere identified early, and mitigation strategies could be demon-strated before finalizing the design of the flight instrument.
References
[1] Takahashi T et al. The ASTRO-H mission. In: Proc SPIE 7732, space telescopesand instrumentation 2010: ultraviolet to gamma ray, 2010. p. 77320Z–77320Z-18.
[2] Mitsuda K et al. The high-resolution X-ray microcalorimeter spectrometersystem for the SXS on ASTRO-H. In: Proc SPIE 7732, space telescopes andinstrumentation 2010: ultraviolet to gamma ray, 2010. p. 773211–773211-10.
[3] Shirron P, Kimball M, James B, Wegel D, Martinez R, Faulkner R, et al. Designand predicted performance of the 3-stage ADR for the Soft-X-ray Spectrometerinstrument on Astro-H. Cryogenics 2012;52(4–6):165–71.
[4] Fujimoto R et al. Cooling system for the Soft X-ray Spectrometer (SXS) onboardASTRO-H. In: Proc SPIE 7732, space telescopes and instrumentation 2010:ultraviolet to gamma ray, 2010.
[5] Breon SR, Branch HD, Blount GJ, Jackson ML, Boyle RF, Tuttle JG. Design of theXRS helium insert. Adv Cryog Eng 1996;41:1129–34.
[6] Kelley R et al. The Suzaku high resolution X-ray spectrometer. Publ Astron SocJpn 2007;59:S77–S112.
[7] Kumagai H, Tominaga G, Tuzi Y, Horikoshi G. Vacuum science andengineering. Tokyo: Shokabo; 1970. p. 72.
[9] White GK, Meeson PJ. Experimental techniques in low-temperature physics.4th ed. Oxford: Oxford Science Publications; 2002. p. 82.
[9] Frenkel J. Kinetic theory of liquids. Oxford: Oxford University Press; 1946.p. 332.
[10] Sabisky ES, Anderson CH. Phys Rev A 1973;7:790.
Fig. 7. Temperature noise is dramatically reduced as the cryocoolers wereprogressively turned off.
212 P. Shirron et al. / Cryogenics 64 (2014) 207–212
