Development of a 0.5-T/240-mm MgB₂ MRI Magnet: Assembly Design and Module Coils

Abstract

This paper presents the latest results from our continued development of a 0.5-T/240-mm MgB₂ MRI magnet at the MIT Francis Bitter Magnet Laboratory. Because we have successfully developed our superconducting joint technique with a monofilament MgB₂ wire, manufactured by Hyper Tech Research, Inc. (Columbus, OH), we have decided to use a monofilament wire to wind our MgB₂ MRI magnet. The magnet, comprising eight module coils, has a winding inner diameter of 276 mm, an outer diameter of 290 mm, and a total height of 460 mm. Each coil has its own persistent-current switch (PCS) and a superconducting joint. In order to guard against a few bad coils forcing the entire magnet to be inoperative, each coil will be heat-treated and tested individually. After eight coils are successfully operated, they will be assembled into an MRI magnet and series-connected with soldering joints between adjacent coils. The PCS in each coil is designed in such way that it will also serve as a detect-and-heat protection absorber when the magnet quenches over a small “localized” region: The conductor volume in the eight switches is designed to absorb the entire magnet energy while still remaining below 200 K. This paper reports 1) the design of the whole magnet and 2) the fabrication and test results of the two real-size test coils, with their PCSs and superconducting joints. The tests were conducted in gas helium in the temperature range of 10–15 K and in the self-field of the coils.

I. INTRODUCTION

MRI magnets are used worldwide in the early detection and efficient treatment of diseases or injuries. Nowadays, most diagnostic MRI magnets are wound with NbTi wires. MgB₂ has a higher critical temperature of 39 K, which makes it easier for the MgB₂ magnet than the NbTi magnet to operate helium-free, a reduced-cost mode. In addition, a high critical temperature results in a greater energy density margin, making MgB₂ magnets more stable than NbTi magnets. Thus, for the MRI magnets, the MgB₂ wire is a promising option[1]–[3].

Because the reliability of our MgB₂ joint technique is higher with mono- than multifilament wires [4], we chose a monofilament MgB₂ wire, supplied by Hyper Tech Research, Inc. (Columbus, OH), for our 0.5-T/240-mm MRI magnet. We have demonstrated that a monofilament MgB₂ wire, of 0.29-mm-thick superconducting core, is flux-jump-free [4]. In addition, due to a high critical temperature of MgB₂, even if there is flux motion in the wire, the magnet will still be quench-free and safe. The wire has diameters of 0.84 mm (bare) and 1 mm (S-glass insulation). Outside the superconducting core, it has from inner to outer a niobium layer, a copper layer, and a monel layer. Of the entire cross section of the wire, MgB₂ and Cu occupy, respectively, 12% and 35%.

II. MAGNET DESIGN

A. Electromagnetic Design

The eight-coil 0.5-T/240-mm MgB₂ MRI magnet is shown in Fig. 1. Today, there are three chief reasons to modularize a large MgB₂ magnet: 1) It is still difficult to acquire a uniform-performance piece of a monofilament MgB₂ wire that is long enough for the entire magnet; 2) heat treatment is not 100% reliable to risk heat-treating the entire magnet; and 3) to limit the extent of damage in case of an unscheduled quench. In this modular design, the joint of each coil can be placed in the space at the outermost turn between adjacent coils, where the field is about an order of magnitude lower than a center field of 0.5 T at an operating current of 102 A.

Fig. 1.

Configuration of a 0.5-T/240-mm MgB₂ MRI magnet, consisting of eight module coils. b1 and b2 are measured from the midplane.

The magnet has a winding I.D. of 276 mm with a 240-mm room temperature bore. Each coil uses less than 300-m-long wire, over which we believe the wire properties are uniform. The total eight-piece conductor length is 2.1 km. Operating in the range of 10–15 K, the magnet generates a center field of0.5 T at 102 A. The magnet has a computed field error less than 200 ppm over a 12-cm diameter spherical volume (DSV). The parameters of the coils are summarized in Table I.

TABLE I.

DESIGN PARAMETERS OF THE MODULE COILS

Parameters	Units	Coil 1	Coil 2	Coil 3	Coil 4
al	[mm]	138	138	138	138
a2	[mm]	145	145	145	145
bl	[mm]	15	89	142	193
b2	[mm]	52	128	179	230
Turns; layers		36; 8	38; 8	36; 8	36; 8
Total turns		288	304	288	288
Operating current	[A]		102
Current density	[kA/cm2]		11.3
Conductor length	[m]	259	273	259	259
Total conductor	[km]		2.10
Field error (pre-shim)	[ppm]		<200 in 12 cm DSV

The parameters a1, b1, b2 and coil numbers are defined in Fig. 1.

In order to maintain the electrical performance of the wire, the wire hoop strain should be smaller than 0.5% [5]. If we use the simplest BJR model to estimate the hoop stress, valid for α = a2/a1 ≈ 1.0 [6], with a Young’s modulus of 100 GPa for MgB₂ [6], a computed safe hoop strain is ≤0.02%.

B. PCS Design

The magnet has a computed inductance of 0.74 H. At a ramping rate of 0.1 A/s, the induced voltage across the magnet will be 74 mV. To limit the bypass current through the copper part of each of eight “open” (≥ 40 K) persistent-current switches (PCSs) to below 1 A, each open PCS must have a resistance greater than 10 mΩ, i.e., it requires its wire to be longer than 5 m.

The magnet will be protected by the detect-and-activate-the-heater—or more concisely, detect-and-heat—technique [6]. The detect-and-heat technique has been designed and tested for this MgB₂ MRI magnet [7]. In this magnet, in place of an implanted quench-triggering protection heater, each PCS acts as not only a quench-trigger heater but also an energy absorber. Upon full discharge at 102 A, even if the entire magnet energy of 3.7 kJ is released only into the PCSs, their final temperature will be less than 200 K. As a < 200-K energy absorber, the MgB₂ wire, having 3 × 10⁵ kJ/m³ enthalpy from 40 K to 200 K, in each PCS must be 3 m long, which is shorter than a length of 5 m required as a bypass-current-limiting (< 1 A) PCS. With eight PCSs open, the magnet will be discharged with a time constant of ~10 s.

III. COIL FABRICATION

A. Pre-Heat-Treatment Fabrication

Each module coil consists of three essential components: the coil itself, the PCS, and the joint. A complete coil is shown in Fig. 2.

Fig. 2.

Completed module coil. It consists of a coil, a PCS, and a joint. A schematic drawing on the right shows the wiring of the PCS and current leads.

So far, we have fabricated two real-size coils, test coils #1 and #2. The coils were wound with a regular winding machine, with a winding tension of ~5 N. We left a 5-m wire at both ends to make the PCS and joint. The winding was hexagonal, with the same number of turns in each layer. After making the PCS and joint, we reinforced the coil with stainless-steel wires before heat treatment.

The PCS topology requires the outlet wire of the coil to be U-bent to meet and cowound with the inlet wire. The two2.5-m-long parallel wires, wound along a stainless-steel strip at the outermost turn of the coil, formed the PCS. A segment of the outlet wire was aligned parallel to the inlet wire of the next coil below, electrically connecting two coils in this eight-stacked-coil magnet.

We made each joint in the same way as we made sample joints [4], at the end of each PCS. The joint was also attached to the stainless-steel strip to prevent damage during handling.

B. Heat Treatment

To date, two coils have been heat-treated in a box furnace in the MIT Department of Materials Science, with the same temperature-time schedule used in the previous test coils [4]. Because the box furnace can keep neither a pure-gas environment nor positive pressure, for heat treatment, the coil is housed in a stainless-steel can of ~7-kPa argon gas, a pressure that we consider to be less than optimal making good joints [8] but the upper pressure limit of this can. This pressure limit issue will be resolved in the near future.

C. Post-Heat-Treatment Fabrication

After the coils were heat-treated, we epoxy the PCS and the joint, in order to enhance the heat conduction within the PCS and prevent possible damage during handling and wire motion during operation. We wound two pieces of manganin wire toroidally around the PCS as heaters: One is for normal switch operation, and the other is for detect-and-heat protection. Each heater has a resistance of 34 Ω at 4.2 K. A 10-mm-wall Styrofoam jacket insulated the PCS thermally.

IV. TEST RESULTS

When finishing the fabrication, we mounted each coil on the test setup and inserted it into a 430-mm-diameter 1-m-deep cryostat. Voltages of both the coil and the joint were monitored. The temperature was monitored by a pair of Cernox sensors, mounted both on top and at the bottom of the coil. Two pairs of type-E thermocouples were mounted oppositely on the PCS loop to monitor the PCS temperature. We used a cryogenic Hall sensor to measure the center field of the coil.

A. Test Coil #1: Persistent Mode

Local heating at the Styrofoam-insulated solder joints to the current leads prevented Test Coil #1 to be charged beyond 85 A, 17 A short of its operating current of 102 A. Coil #1 was operated at 70 A in persistent mode.

Fig. 3 shows a charging profile of Coil #1. The solid circle line is the coil voltage, matching the calculated value very well. The solid square line shows the supply current profile. After the external current was ramped down to zero, the field still kept constant at 0.09 T. Fig. 4 shows the field kept at 0.09 T (70 A) for 4 h, in the temperature range of 10–20 K, in self-field. In Fig. 4, t = 0 coincides with t = 0 in Fig. 3. The loop resistance, estimated from the linear regression of the curve, is about 3 × 10⁻⁹ Ω. A step response of the field happened at ~30 min when the temperature was increasing from4.2 K to 10 K. The mechanism is not completely understood, but we believe this transition is associated with the normal-state transition, at 9.8 K, of niobium that shields the MgB₂ core.

Fig. 3.

Supply current, coil voltage, and center field versus time plots of Test Coil #1 during charge-up. On the right axis, the 0–10 range is the coil voltage in mV, and the 0–0.1 range is the center field in T.

Fig. 4.

Center field versus time plot of Test Coil #1 in persistent mode. The coil sustained a persistent current of 70 A in the temperature range of 10–20 K. The calculated loop resistance is less than 3 × 10⁻⁹ Ω.

B. Test Coil #1: PCS Open and Close

Fig. 5(a) and (b) shows the open and close operations of the PCS, respectively. The vertical dashed lines indicate when the heater was turned on or off. In the steady open state, the heater current was 0.7 A, generating a power of 17 W to keep the PCS above 40 K. We will further reduce this power through better thermal insulations.

Fig. 5.

Temperature versus time plots of opening and closing the PCS in Test Coil #1. (a) PCS opening at 40 K in 10 s. (b) PCS closing below 10 K in 20 s.

In Fig. 5(a), we can see that the PCS warmed up from about 4.2 K to 40 K in 10 s after the heater had been turned on. Fig. 5(b) shows that the PCS temperature dropped from above 50 K to below 10 K in about 20 s. These results were obtained when the PCS was in gas helium, right above the liquid level. In other cryogenic environment, for example, in solid nitrogen, the operation time may vary accordingly.

C. Test Coil #2: Coil and Joint

Because Test Coil #1 did not reach our target current, we wound and heat-treated Test Coil #2.

We were able to charge this coil in driven mode to the target current of 100 A at 15 K, as shown in Fig. 6. Unfortunately, the PCS of this coil could carry only 50 A of superconducting current at 15 K, because it was possibly damaged after heat treatment by mishandling.

Fig. 6.

Supply current, coil voltage, and center field versus time plots of Test Coil #2 during charge-up. On the right axis, the 0–10 range is the coil voltage in mV, and the 0–0.15 range is the center field in T. We were able to charge the coil to 100 A in 15 K, in self-field.

D. Test Coil #2: Persistent Mode

Since the critical current of the PCS is 50 A at 15 K, we decided to operate the coil in persistent mode at 45 A. Fig. 7 shows that the field kept almost constant for about 6 h in the temperature range of 10–20 K, in self-field. The field, namely, the current in the loop, started to decay at ~20 K and dropped to zero at ~30 K. The loop resistance calculated % from the slope of the field–time curve is 4 × 10⁻⁹ Ω.

Fig. 7.

Center field and coil voltage versus time plots of Test Coil #2 in persistent mode. The coil was able to sustain a persistent current of 45 A in the temperature range of 10–20 K. The current started to decay at 20 K and dropped to zero at 30 K. The calculated loop resistance is 4 × 10⁻⁹ Ω.

E. Test Coil #2: PCS Protection

Since the active detect-and-heat protection technique has been designed and proven, we did not test the detect mechanism but only the dump performance of Test Coil #2.

The PCS was tested with a current of 100 A flowing in the coil loop semipersistently, with a joint resistance of 3 μΩ. In Fig. 8, we can see that the PCS temperature increased immediately after we turned on the protection heater. In about 2 s, when the PCS temperature reached about 30 K, the field started to decay. Including the detection time proved in previous experiments, the magnet should start dumping energy in 3 s after an unrecoverable local hot spot occurs. This is prompt enough to protect this MgB₂ wire from permanent damage.

Fig. 8.

Center field and PCS temperature versus time plots of Test Coil #2 after a dump. At ~2 s, the heater was activated, warming up the PCS right away. Two seconds later, the coil started to decay, releasing the magnetic energy in the PCS, causing the PCS to warm up more rapidly.

Fig. 9 shows that during the dumping process, the PCS temperature only went up to 60 K, well below the designed peak temperature of 200 K. The nonoptimized thermal insulation and convection of gas helium conduced to keeping the PCS cool.

Fig. 9.

Center field and PCS temperature versus time plots of dumping Test Coil #2. The coil was discharged in ~30 s. The temperature of the PCS only reached 60 K, well below a designed temperature of 200 K.

V. CONCLUSION

We have made steady progress on the MgB₂ MRI magnet program at the MIT Francis Bitter Magnet Laboratory. To minimize risk in working with as-yet fully developed MgB₂ wire and its magnet technology, we adopted a modular design for this magnet that comprises eight module coils, each requiring a MgB₂ monofilament wire of ≤300-m length. To date, we have fabricated and tested two module coils. The coil part of Test Coil #2 successfully carried a targeted current of 100 A in driven mode. The PCS can be opened and closed in a reasonable time of 30 s. The energy absorber function of PCS also worked well. Proper handling of the coils of this scale, including their PCSs and joints, remains to be perfected. High pressure during heat treatment appears to improve joint performance. In addition, we are developing a technique to make MgB₂ joints with reacted monofilament wires, which may make the react-and-wind route a realistic option. We believe that success of this magnet is a major milestone in MgB₂ MRI magnet technology and will likely serve to promote further R&D work in this technology and, ultimately, the proliferation of the MgB₂ MRI magnet in the near future.