3D-printed RF Probeheads for Low-cost, High-throughput NMR
R. Adam Horch1,2 and John C. Gore1,2

1Vanderbilt University Institute of Imaging Science, Vanderbilt University, Nashville, TN, United States, 2Department of Radiology & Radiological Sciences, Vanderbilt University, Nashville, TN, United States

Synopsis

3D printing is demonstrated as a new means to fabricate complete RF probeheads for solution-state NMR. Current 3D printing methods yield mm-scale RF coils with integral sample chambers for self-contained NMR probes, and 3D-printed microcoils are imminent given ongoing advances in technology. The unique properties of 3D printing enable facile construction of potentially thousands of coils at low cost, giving way to dense coil arrays for high-throughput NMR and novel coil geometries.

Introduction and Purpose

Conventional probeheads for solution-state NMR are constructed using manual1 or photolithographic techniques2 that incorporate RF coils, tune/match circuitry, and liquid sample holders into spatially-compact housings. These techniques are suitable for probeheads with up to a few 10s of RF coils but suffer from expensive, time consuming, and labor-intensive manufacturing practices. As an alternative, we have been exploring stereographic 3D printing that provides rapid, low-cost fabrication with minimal geometric constraints. 3D printing renders topologically-complex structures in a wide variety of NMR-compatible polymers, with print resolutions of ~10µm across build volumes exceeding one cubic meter3. Herein, we demonstrate complete, 3D-printed NMR probeheads that contain RF resonators, variable tuning/matching capacitors, and liquid NMR sample cavities. Fluidically-addressable channels and cavities are included in the probehead structure, providing a means for hydraulically-controlled RF tuning/matching and liquid NMR sample loading/unloading. Electrically conductive RF circuitry is defined within the 3D printed polymer bodies by metallizing certain structures with silver ink. In this way, multiple independent RF coils and NMR sample cavities can be fabricated into a single probehead for high-throughput NMR. The unique scalability of 3D printing supports fabricating thousands of millimeter-scale coils in a single build session, providing a manufacturing pathway towards a new generation of low-cost, high-throughput probeheads.

Methods

Stereographic 3D printing utilizes liquid polymer resins that are photo-crosslinked in a layer-by-layer manner. With a 3D Systems iPro 9000XL Printer using Somos 11122 Watershed material (DSM Desotech), probeheads were formed with an in-layer resolution of 40 µm and a through-layer resolution of 150 µm. Probeheads were printed with a number of non-connecting hollow channels, defining RF circuitry, hydraulic control lines for variable capacitors, or NMR sample loading pathways. After printing, the RF circuit channels were selectively metallized by injecting PELCO Silver Paint (Ted Pella Co.) to form ≈10µm silver coatings for electrically conductive pathways. In this way, RF-resonant solenoids were constructed and tuned/matched using parallel-plate variable capacitor networks included in the probehead. These capacitors contained a partially hollowed dielectric (Figure 1), which was filled with a variable amount of Fomblin fluorocarbon oil (Solvay Solexis) or D2O (99.9%, Sigma Aldrich). Since Fomblin and D2O are immiscible and have dielectric constants that differ by 40-fold, a moveable liquid interface between the two was placed within the capacitor’s dielectric gap to provide hydraulic control of net capacitance. Additionally, Fomblin and D2O are invisible to 1H NMR and bear similar magnetic susceptibility to H2O, thus allowing the parallel plate capacitors to be placed in the vicinity of the NMR sample for a compact RF coil footprint. Probehead sample chambers were filled with 40 µL saline (1x PBS), and 1H FIDs were collected on a 9.4T Varian/Agilent small animal imaging system (10kHz BW, 40000 points, 1 acquisition).

Results

Probeheads were successfully printed with 0.75mm-diameter hollow channels to define RF coils, each consisting of a 5mm-diameter/5-turn solenoid and three parallel-plate capacitors for tuning/matching (Figure 2). The complete RF coil circuit occupied a 2 mL footprint, which was printed into an 8mL probehead body with fluid couplings for separate hydraulic control of the capacitors and NMR sample (Figure 3). RF bench testing showed a broad tuning range of 40 MHz as the capacitors were hydraulically varied, with Q ≈ 10 at 400 MHz (Figure 4). A match of only -12dBm (S11) could be achieved at 400 MHz, and efforts are underway to optimize coil geometry and capacitor range for an improved match. However, 1H FIDs on 40 µL saline were observed at 400 MHz (Figure 5), with a ≈ 20 Hz linewidth acceptable for high-throughput NMR processes such as drug screening. The coil gave a 1 kHz nutation rate under 125 mW input RF power, indicating compatibility with low-cost, low-power RF front-end electronics.

Discussion and Conclusions

3D printing and selective silvering was demonstrated as a new means to fabricate complete NMR probeheads for solution-state NMR. Importantly, RF coil tuning/matching and NMR sample loading/unloading could both be controlled hydraulically, which is compatible with high-throughput infrastructure from the microfluidics industry. Recently-developed 3D printers have driven the minimum build resolutions to 5-fold smaller than what was used herein, so work is now underway to fabricate single probeheads with high-density arrays of independent RF coils and NMR sample pathways. Ultimately, the metallizing process will be combined with the 3D-printing process for facile construction of large numbers of coils. Given its low cost and build scalability, 3D printing offers a unique means to construct probeheads of unprecedented complexity, enabling drastically denser arrays for high-throughput NMR and novel coil geometries that are not available with conventional fabrication techniques.

Acknowledgements

Vanderbilt University Department of Radiology, NIH T32 EB001628.

References

1) e.g. Webb, A. G. Progress in Nuclear Magnetic Resonance Spectroscopy 31, 1–42 (1997).

2) e.g. Lee, H., Sun, E., Ham, D. and Weissleder, R. Nat Med 14, 869–874 (2008).

3) see manufacturer specifications at www.stratasys.com and www.3dsystems.com

Figures

Figure 1: Schematic of 3D-printed, parallel-plate variable capacitor for hydraulic tuning. Two conductive plates (left) span a dielectric gap consisting of two 3D-printed matrix layers (d­M) and a hollow hydraulic layer filled with Fomblin or D2O (dH). The immiscible Fomblin/D2O interface can be moved across the capacitor via hydraulic pressure, varying total capacitance by way of the 40-fold-difference in dielectric constant (k). An equivalent electrical schematic is shown at right.

Figure 2: Probehead electrical schematic and physical layout. A conventional RF resonant circuit was used (upper left), consisting of three variable capacitors for matching/tuning/balancing and a solenoid. This circuit was 3D-printed with hollow structures (shown as solid bodies in color-coded views), which were rendered in a solid polymer body. Electrically-conductive channels defining the RF circuit (red, purple, yellow) were selectively metallized.

Figure 3: Representative photographs of prototype 3D-printed probehead. The complete probehead with RF transmission line and hydraulic/sample fluid lines is shown at left. Luer lock interfaces adapted the probehead fluid channels to eight 1mL syringes. A close-up schematic view of the probehead core is shown at upper-right (color-coded to Figure 2), above corresponding photographs of the probehead body before and after metallization with silver ink using a number of fill ports.

Figure 4: Network analyzer S11 sweep. Probehead S11 response (400MHz center, 400MHz span) is shown before (top) and after (bottom) the tuning capacitor is varied hydraulically by injection of D2O. In the presence of a 40 µL saline load, the best observed match was -12dBm at 400 MHz with a Q of ~10. Since this is a proof-of-concept, optimizations are underway to improve performance.

Figure 5: Representative 1H NMR spectrum. A single resonance was observed from a saline load, with shoulders likely arising from a difference between sample and probehead magnetic susceptibility. A narrow, unshimmed linewidth is conducive to high-throughput probeheads which contain multiple coils that may not be individually-shimmed.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
2147