Hydrogen Cooling Options for MgB₂-based Superconducting Systems

Abstract

With the arrival of MgB₂ for low-cost superconducting magnets, hydrogen cooling has become an interesting alternative to costly liquid helium. Hydrogen is generally regarded as the most efficient coolant in cryogenics and, in particular, is well suited for cooling superconducting magnets. Cooling methods need to take into account the specific quench propagation in the MgB₂ magnet winding and facilitate a cryogenically reliable and safe cooling environment. The authors propose three different multi-coolant options for MRI scanners using helium or hydrogen within the same design framework. Furthermore, a design option for whole-body scanners which employs technology, components, fueling techniques and safety devices from the hydrogen automotive industry is presented, continuing the trend towards replacing helium with hydrogen as a safe and cost efficient coolant.

INTRODUCTION

Hydrogen is the most abundant element on our planet. Not only is it the best and most cost efficient cryogenic coolant we know of for superconducting magnets (see Tables 1 to 3), “indirect liquid hydrogen cooling” (i-LH₂) for MRI magnet systems has also cost advantages, as recognized by NIH (Grant 5R01EB009360) and others thinking ahead, like Glowacki: “A cryogen-free, or if interfaced with a hydrogen economy, an i-LH₂-cooled 1.5 T MgB₂ MRI system, could have a large social impact” [1].

TABLE 1.

SC magnet cryogenic coolants [8] – temperature range and associated enthalpy change¹.

Helium		Hydrogen		Nitrogen
4K – 20K	86.94 J/g	20K – 77K	590 J/g	77 K – 300 K	233.5 J/g
4K – 77K	383.72 J/g	20 K – 300 K	3490 J/g
4K – 300K	1541.83 J/g

¹Change in enthalpy of gas at 1 atm pressure.

TABLE 3.

1 W dissipated in a cryogenic coolant at its boiling point.

Fluid	Volume of liquid boiled off	Flow of gas at 0°C, 1 atm.
He	1388 ml /h	16.05 1/min.
H2	114.5 ml/h	1.506 1/min.
N2	22.5 ml/h	0.243 1/min.

Hirabayashi et al. studied the feasibility of using liquid hydrogen in superconducting energy storage systems [2] and Nakayama et al. show superconducting cable modules would be benefiting from the future hydrogen economy with using either HTS or MgB₂ superconductors [3]. Paul Grant’s vision for a hydrogen based infrastructure expands this thinking process even further in great detail [4]. Hydrogen cars and buses are now available and safe techniques for storing gaseous and liquid hydrogen have been established, pioneered by the space and automotive industry. Some cars already use pressurized liquid hydrogen in their reservoir (e.g. BMW Hydrogen 7, and the first production of Hyundai’s ix35 FuelCell car). The Department of Energy (DOE) Hydrogen and Fuel Cells Program is supported by the NHA (National Hydrogen Association) and synergies to cryogenics are included there as well. Since early on, NASA worked with hydrogen, related to storage and handling and defined “best practices” and guidelines [5]. It goes without saying we can learn a lot from tapping into this knowledge. As pointed out succinctly by the late Edeskuty: “The public perception of the safety problems of working with hydrogen, the so-called “Hindenburg syndrome” has hindered the development of systems that could safely use hydrogen as an ideal fuel” [6]. Given the rise of cost of helium [7], together with the availability of new types of superconductors, it is now about time to change this perception. In the following we will make a technology leap and analyze what design strategies are promising and what might still be missing. Clearly, for the cost-cautious engineer the cryogenic world looks different at an operating temperature of 20.3 K compared to 4 K. To a certain extent we can then benefit from thermal conductivity peaks for copper and aluminum as heat drains in superconducting magnets or from increased quench stability limits against disturbances. Tables 1 to 3 refer to cryogenic advantages when substituting helium with hydrogen as a coolant for large thermal mass cooldown, its capability absorbing large heat loads or when used in heat pipes. The gas / liquid expansion ratio is only slightly higher for hydrogen as compared to helium which enables us to keep a liquid reservoir similar sized. Another great advantage is the low boil-off rate (liter/hour) per Watt heat load. Moreover, the cryogenic design simplifies if the low boil-off allows leaving out the thermal shield (magnet size dependent), provided ride-through time is less of a concern. For long ride-through times for coldhead service or during power outages and shipments the thermal shield will still be beneficial. Should even longer ride-through times be required thermal batteries are readily available in the 20 K range that are difficult to implement at 4 K and only at high cost.

MgB₂ Magnet Cooling Overview

The typical MgB₂ magnet operating temperature is slightly below 20.3 K @ 1 bar. At that temperature a small MRI magnet system will be able to maintain a magnetic field of 1.5 T. However, as long as the engineering current density J_eng of MgB₂ is still low a multi-coolant approach is required for higher field strengths above 1.5 T. During this transition phase toward a wire with higher critical current I_c and coil J_eng, a small amount of liquid helium may still be required with MRI systems for fields of 3 T or higher. Table 4 gives an overview of superconducting (SC) magnet cooling methods for hydrogen and helium:

TABLE 4.

Cooling method comparison: hydrogen vs helium.

Hydrogen (LH2)	Helium (LHe)
Thermosiphon w/o helium as barrier gas	Thermosiphon
Heat pipe (w/o) wicks	Heat pipe (nanowicks)
Pulsating heat pipe	Pulsating heat pipe (capillaries) [10]
Bath-cooling (not recommended)	Bath-cooling

General 20 K cooling considerations for HTS and MgB₂ magnets

Small MRI limb systems based on low temperature superconductors can work very well in conduction-cooled mode [11]. Quench propagation in MgB₂ or HTS is however rather slow with a normal zone propagation (NZP) typically in the range of 2.5 to 10 cm/s for MgB₂ depending on wire/tape type with minimum quench energies (MQE) in the range of 3 to 5 J for 300 to 450 A. Quench protection and detection is therefore more difficult and hot spots within a coil are not easily detectable. Local hot spot heating in the windings could well lead to a distorted magnetic field and must be avoided. Good conduction cooling at 20 K would require costly high-purity aluminum formers or heat sinks and heat spreaders. This has other downsides apart from the high cost namely eddy current generation and with MRI magnets, consequently high heat loads during gradient / magnet interaction. Furthermore, for solenoids cooling needs to be provided as close as possible to the winding. There are several ways of achieving that, either by using coldplates, and/or thermosiphon / heat pipe technology. In particular with pulsating heat pipes (PHPs) one obtains thermal conductivity values in the region of 8500 to 11480 W/mK as Mito el al. report: “For HTS magnets which can operate at 20 K or higher, the thermal time constant becomes long due to the decrease of the thermal diflusivity, which is necessary for cooling and the anomaly detection in the magnet, and the magnet protection becomes more difficult. A cryogenic oscillating heat pipe, that enhances the heat removal characteristics of HTS magnets by embedding them close to the magnet windings is proposed” [12]. From the cryogenic system design point of view here are some typical “Must haves” for MRI systems in a clinical environment:

• Cryogenics needs to be invisible for the user, requiring no maintenance and service

• Requires a highly efficient, hermetically closed cooling circuit

• Vibration transmission to magnet caused by coldhead should be eliminated

• Cryocooler needs to be serviceable in the hydrogen enviromnent and at field

• Needs to sport a long ride-through capability at power outages and for coldhead exchange

• Guaranteed safety for all MRI operating modes (tubes need to be in vacuum)

• All tubing external to the cryostat may require an protective helium gas barrier

• Warm gas purge of tubing needs to be possible either using GHe or GN₂

• Fast cold mass pre-cooling with LN₂, preferably making use of built-in hydrogen cooling loop

• Automatic coldmass cooldown

• In case of magnet quench, safely capture gaseous hydrogen

In the following a small-sized system design implementation is shown based on this strategy.

MgB₂ Small-Sized System (Dual Coolants)

With the arrival of long lengths of MgB₂ wire e.g. from Columbus, it is now possible to wind superconducting coils (rather than pancakes) [13]. The figure below shows a 6 coil MRI magnet design for extremity scans. The coils are individually wound using the paint and brush method and without coil former. Due to superconducting magnet stability reasons we need to provide adequate radial and circumferential cooling. For this, cold plate technology is introduced as shown below in Fig. 1.

FIGURE 1.

MRI magnet cold mass cooling with fitted cold plates.

Table 5 gives the magnet design parameters. First test results of the small coil performance as well as on MgB₂ SC joint technology have recently been presented [13]. The magnet consists of 2 large coils, 2 medium coils and 2 small coils, performance optimized for 1.5 T and 3 T. All coils are fitted with Al6061-T1 aluminum cold plates, Stycast 2850-FT bonded to both faces. Additionally, L to M (Large to Medium coil) sized cold plates each feature a welded-on aluminum cooling loop tube, whereas M to S (Medium to Small) plates and the SS (Small) plate are only thermally linked to each other via tinned copper braids (inner and outer coil circumference). Slotting of cold plates is required to locally confine eddy currents avoiding any closed current loop that otherwise would be transiently induced by the gradient system. The coils are pre-stressed to compensate for the 1 mm thermal shrinkage in z-direction using machined G10 end plates and threaded SS 316L suspension rods. The end plates also have grooves towards the coil surface to hold cooling strips for end plate and support ring cooling. Figure 2 shows the cryogenic structure with its magnet (left), the cryostat cut open (center) and the overall design (right).

TABLE 5.

Magnet design parameters.

Item	Unit	3 T magnet
Operating temperature	K	4.2 / 20
Warm bore diameter	mm	280
Coil ID	mm	300
Coil OD	mm	412 (max)
Magnet length	mm	558 (total)
Number of turns	-	73,120
Operating current	A	250
Operating temperature	K	4.2
Center field at 3 T	T	3.0
Max. field at 3 T	T	3.85
Wire length	m	8080
Wire size	mm x mm	0.71 × 3.1
Design	19 filaments/Ni matrix / Cu tape
Twist pitch	mm	500
Insulation	Dacron braid

FIGURE 2.

Cryostat with MgB₂ magnet, front view (left), outer shell omitted for clarity (center), side flange removed (right).

Description

As mentioned, the current cryogenic design of the MgB₂ magnet is based on the use of dual coolants either with hydrogen or helium. The design consists of two distinct and separate primary and secondary cooling circuits. The primary, hermetically closed piping circuit employs thermosiphon technology with cooling loop tubes welded around the large coil cold plates (L) (see Fig. 2 center). This circuit supplies liquid cryogen to the cooling loop tubes via a manifold above the magnet coils and also liquefies upstream gas bubbles using a finned heat exchanger tube within a separate cooling reservoir (condensing tube closed at both ends). The fins are in contact with helium gas to liquefy hydrogen within this tube (see Fig. 3). This allows constant 20 K hydrogen cooling without the need of a pool of liquid hydrogen in the reservoir. The GHe containing reservoir enables precise temperature control of the liquefaction fins by using a heater attached to the second stage of the Sumitomo RDK 408 cryocooler. This way, cryocooler vibrations are not transmitted which otherwise would result in “ghosting” effects in an MRI image. For operation at higher magnetic fields at 3 T and 4 K the reservoir (secondary circuit) is filled with 2 1 liquid helium by initially supplying warm gas from a helium gas bottle that is precooled by the thermal shield, with the flow then passing the liquefaction fins of the cryocooler. The liquid helium reservoir also provides sufficient ride through time for coldhead exchange. For hydrogen and helium use an external 4 liter gas reservoir is filled with an overpressure of 2 MPa and sealed (not shown). During a magnet quench, liquid cryogen in the thermosiphon tubing and in the manifold evaporates and is captured in the external gas reservoir. The warm gas is then reclaimed from the gas reservoir for initiating magnet re-cool. For 4 K operation a constant LHe level is maintained in the reservoir, whereas in the 20 K operating mode only GHe is kept in the reservoir. The condensing tube with its heat exchange fins is shown in Fig. 2 center. Above LHe temperature single coolant use reduces the design effort considerably. In this case a reservoir as shown in Fig. 2 (secondary circuit) is no longer required and the reservoir reduces to the size of a liquefaction cup attached to the 2^nd stage of the cryocooler with gas/liquid inlet/outlet. For HTS type BSCCO magnets and other coolants, e.g. neon [14] this approach works well. The general difficulty with neon apart from its cost is its small operating temperature range window (>24.6 to 27.1 K) for thermal mass cooling, making this hard to control properly without plugging the tubes in the presence of magnet temperature transients, which would then result in an immediate magnet warm up and quench. This is much easier to control electronically with hydrogen and a dT of 6.3 K that gives a better active control for maintaining the liquid state, avoiding subatmospheric operation. Gaseous helium is also used as buffering gas against any possible air ingress.

FIGURE 3.

Dual cryogenic working fluid design (4 K thermosiphon or alternatively 20 K pulsating heat pipe).

Cooldown

As Fig. 2 indicates, the magnet cooling tubes are broken and plugged at the end for eddy current reasons mentioned, avoiding a closed current loop. This offers the added benefit of efficiently using the thermosiphon tubing in “heat pipe mode” for magnet cooldown. Oomen et al. have shown that the so-called evaporator of a traditional heat pipe can assume part of the circumference of the magnet coil [15]. Furthermore, in the presented design the liquid reservoir at the top of the tubing assumes the role of the traditional liquefaction fins. Once temperatures and pressures in the cooling loop tubes are close to the critical point, liquefaction commences following the vapor pressure curve for hydrogen and all 4 heat pipes switch to thermosiphon mode. As mentioned, gas supply is provided by an external gas storage that then produces 150 ml of liquid hydrogen. Depending on the coolant chosen and the targeted magnet field the circuit pressure rises to 2 MPa for helium and 2.5 MPa for hydrogen.

Hydrogen gas storage

The quest for a self-contained MRI cryostat system (push-button approach) that regulates and controls itself in all operating modes requires either external or internal storage of so-called “Compressed Gaseous Hydrogen” (CGH₂) using a hydrogen tank. Below the most common hydrogen storage options are listed:

• Pressure storage systems (250 to 300 bar, 700 bar possible)

• Liquid storage system (requires cryostat and liquid container, boil off)

• NaAlH₄ storage systems

  Operational temperature: 125 to 160°C, pressures up to 100 bar [16]

• LiBH₄/MgH₂ storage systems

  Operational temperature: 350 to 400°C, pressure approx. 50 bar [16]

• FeTiH₂ hydrides

  Operational temperature: approx. 20°C, pressure approx. 5 bar [17]

• Getters (like SAES hydrogen for room temperature), synthesized alloys and nano-structure based materials

Hydrogen storage is an exciting, rapidly forward moving field of science with new storage methods thought of by the automotive / fuel cell and other industries with Jepsen focusing on the economic aspects and potential of those complex hydrides as compared to conventional hydrogen storage systems [16]. A good overview can also be found in a DOE report compiled by Ford in 1996 comparing different storage methods [18]. The dynamic filling characteristic of hydrogen adsorptive storage systems operating at cryogenic temperatures has been investigated and analyzed numerically and experimentally [19]. Filling an automotive hydrogen storage tank should not be fundamentally different from temporarily storing cold hydrogen in an adsorption matrix during a magnet quench. Fill time constants are similar but the stored hydrogen mass (10 to 20 g) is quite low compared to hydrogen storage tanks (4 to 5 kg within 3 min fill) for fuel cell powered cars. Some selected properties of intermetallic compounds are given in Table 6:

TABLE 6.

Intermetallic compounds and their hydrogen-storage properties [17].

Fluid	Metal	Hydride	Mass %	peq, temperature
Elemental	Pd	PdH0.6	0.56	0.020 bar, 298 K
AB5	LaNi5	LaNi5H6	1.37	2 bar, 298 K
AB2	ZrV2	ZrV2H5.5	3.01	10–8 bar, 323 K
AB	FeTi	FeTiH2	1.89	5 bar, 303 K
A2B	Mg2Ni	Mg2NiH4	3.59	1 bar, 555 K
Body-centered cubic	TiV2	TiV2H4	2.6	10 bar, 313 K

MgB₂ Small-Sized System (Dual Coolants)

Figure 3 depicts an alternative cooling layout with meandered, multi-loop structure for magnet cooling at 20 K or 4 K. For 1.5 T at 20 K operation this design allows conveniently routing the thin-walled stainless steel tubing with a typical internal diameter of 5 to 7 mm preferably next to the coil winding. This is advantageous especially for magnet coils wound on a former or coils with G10 spacers and small width, weight optimized, aluminum cold plates. For medium sized systems operating at 20 K thermal diffusion has slowed down considerably and cooling needs to be made even more efficient against disturbances. In this case cryogenic pulsating heat pipes would be recommended to improve removal of heat from magnet coils as discussed under cooling strategies. Should it then be necessary to work at 4 K the same tubing diameter and tubing routing can be used for operating the cooling loop tubes in thermosiphon mode, whereas each single meandered loop essentially acts as a thermosiphon. This design makes use of the known fact that pulsating heat pipes for hydrogen and thermosiphons filled with liquid helium own the same tube geometry with good efficiency.

Using a single cryogenic working fluid (hydrogen, neon or nitrogen) makes the above design simpler. The 4 K cryocooler can be replaced by a low cost shield cooler with much higher cooling power, with thermal straps heat sinking bottom and top of the primary circuit (meandering structure). This suffices for heat removal for medium-sized magnet systems. Furthermore, the secondary circuit is also no longer required, as mentioned earlier. For medium-sized magnet systems cold plate cooling does become expensive even with a powerful 10 K cooler at hand and PHPs seem to provide a cost efficient alternative.

MgB₂ Large-Sized Systems (Single Coolant)

Small and medium-sized systems finally results in an envisaged, helium-free whole-body scanner design as conceptually shown in Fig. 4, although hydrogen thermosiphons or PHPs may still be required, keeping in mind the magnet size of a high-field whole-body system. The projected system design is completely embedded in a presumably existing hydrogen infrastructure and its supporting environment. Many suitable, hydrogen qualified components are now available or are about to appear on the horizon (e.g. valves, fill stations, pressure tanks, level meter etc). Commercially available 10 K type cryocoolers with typical cooling powers of > 8 W @ 20 K (50 W @ 40 K) for a dual-stage Sumitomo RDK210 or the Cryomech Pulse Tube Cooler PT815 with 22 W @ 20K (46 W@ 48 K) compared to present LTS MRI systems with an RDK408 and 1 W @ 4 K are capable of opening up cryogenic design space by changing the internal cryostat structure, resulting in bigger warm bores for better patient comfort and providing cost reduction in system manufacturing and ownership. Since hydrogen is readily available from the MRI system itself, part of it could be directed for use in fuel cells to provide convenient ride-through for the cryocooler in case of power outages. Further details of the cryogenic cooling system engineering are described in [20] and [21].

FIGURE 4.

MgB₂ based whole-body MRI magnet embedded into hydrogen environment.

Cryostat safety design aspects in a clinical environment

Hydrogen is the lightest and smallest molecule known and has the ability to diffuse through almost any solid structure. It is odorless, colorless and non-toxic. Table 7 below shows selected, characteristic properties compared to fuels [17].

TABLE 7.

Selected physical and chemical properties of hydrogen, methane and petrol.

Properties	Hydrogen (H2)	Methane (CH4)	Petrol (− CH2 −)
Self-ignition temperature (°C)	585	540	228-501
Flame temperature (°C)	2045	1975	2200
Ignition limits in air (vol. %)	4-75	5.3-15	1.0-7.6
Minimal ignition energy (mW s)	0.02	0.29	0.24
Flame propagation in air (m/s)	2.65	0.4	0.4
Diffusion coefficient in air (cm2/s)	0.61	0.16	0.05

Compared to Methane and Petrol it is 10 times easier to ignite hydrogen for an air vol. % between 4 to 75 %. The cryostat design needs to address the above safety aspects. To satisfy these requirements thermosiphon cooling tubes are hermetically sealed, surrounded by helium as a barrier gas and are therefore not exposed to atmosphere. Tubing connections, as well as attached safety features (valve, burst disk, buffer volume, etc) need to be rated for liquid / gaseous hydrogen service and require absolute hydrogen leak tightness. Leak detecting devices need to be qualified and tested. The majority of detected hydrogen accidents in industry relate to those undetected leaks, most of them can be traced back to flange/gasket problems [22]. Ignition sources are mechanical sparks (rapidly closing valves) and electrostatic discharges, or sparks from electrical equipment and lightning strikes in the vent stack [23]. First design concepts showed that the liquid reservoir for the thermosiphon tubes around the superconducting magnet can be as low as 150 to 200 ml. Initially, this reservoir is supplied from an attached pressurized hydrogen gas buffer volume and liquefied using a cryocooler in the presence of helium as an exchange gas. This enables maintenance of the cryocooling device without compromising the integrity of the hydrogen cooling path. It is envisaged to externally store the quench gas leaving the thermosiphon tubes using metal hydrides and to reclaim the hydrogen gas after a quench. In particular one of those metal hydrides, namely FeTIH2, is capable to safely store hydrogen gas at 300 K and pressures as low as 1.5 bar (see Table 6). Alternatively, a 4 liter “party balloon” sized gas pressure tank for initial cooldown can be used for internally storing quench gas. Any leakage of gaseous hydrogen from the cooling loop tubes will the poison vacuum only, an additional safety means for the safe operation of hydrogen cooled superconducting magnets.

Cryostats were developed and their safe performance proven in space technology, for example Ariane and Space Shuttle - both use hydrogen fuel during their launching process. Best working practices have been for example established by PNNL and LANL with DOE funding and NASA who developed safety standards for hydrogen and hydrogen systems, including guidelines for system design materials selection, operation, storage and transport [5]. Further engineering guidelines can also be found in free e-books [23].

A collaboration of the Pacific Northwest National Laboratory and Los Alamos National Laboratory with funding from the U.S. Department of Energy maintains a “Best practices website”: http://h2bestpractices.org/default.asp and ISO (International Organization for Standardization) hydrogen technology publications are available.

CONCLUSION

When following outlined and known design principles, hydrogen is safe to use in a medical environment and eventually will have to replace helium. Using small reservoirs together with thermosiphons/heat pipes, hydrogen gas can be stored in the vicinity of the cryostat. Compact low-cost design studies are based on safe and hermetically sealed tubing with helium as buffer gas. The proposed indirect cooling method eliminates cryocooler vibrations and thermal contact issues.

TABLE 2.

Amount of cryogenic coolant required for cold mass cooling [9].

Fluid	Helium				Hydrogen				Nitrogen
Initial metal temperature K	300		77		300		77		300
	σ2	Liters	σ	Liters	σ	Liters	σ	Liters	σ	Liters
		per lb		per lb		per lb		per lb		per lb
Using latent heat of vaporization only
A1	8.3	30.2	0.4	1.45	0.38	2.42	0.018	0.12	0.83	0.46
SS	4.2	15.1	0.18	0.65	0.2	1.28	0.0085	0.05	0.43	0.24
Cu	3.9	14.1	0.27	0.98	0.17	1.08	0.012	0.08	0.37	0.21
…. and using the enthalpy of the gas
A1	0.2	0.73	0.028	0.1	0.075	0.48	0.0097	0.06	0.51	0.29
SS	0.1	0.36	0.013	0.05	0.037	0.24	0.0045	0.03	0.27	0.15
Cu	0.1	0.36	0.02	0.07	0.037	0.24	0.0065	0.04	0.23	0.13

²σ = weight of fluid required to cool the same weight of metal to the fluid boiling temperature.