Micro and Nano Scale NMR E-Book

• Herausgeber: Wiley-VCH Verlag GmbH & Co. KGaA
• Kategorie: Wissenschaft und neue Technologien
• Serie: Advanced Micro and Nanosystems
• Sprache: Englisch

Seitenzahl: 848

Veröffentlichungsjahr: 2018

Beschreibung

This must-have book is the first self-contained summary of recent developments in the field of microscale nuclear magnetic resonance hardware, covering the entire technology from miniaturized detectors, the signal processing chain, and detection sequences. Chapters cover the latest advances in interventional NMR and implantable NMR sensors, as well as in using CMOS technology to manufacture miniaturized, highly scalable NMR detectors for NMR microscopy and high-throughput arrays of NMR spectroscopy detectors.

Leseprobe

Table of Contents

Cover

Editor's Preface

Series Editor's Preface

Chapter 1: Magnets for Small‐Scale and Portable NMR

1.1 Introduction

1.2 Compact Permanent Magnets

1.3 Magnet Development

1.4 Concluding Remarks

References

Chapter 2: Compact Modeling Techniques for Magnetic Resonance Detectors

2.1 Introduction

2.2 Fast Simulation of EPR Resonators Based on Model Order Reduction

2.3 System Level Simulation of a Magnetic Resonance Microsensor by Means of Parametric Model Order Reduction

2.4 Conclusions and Outlook

References

Chapter 3: Microarrays and Microelectronics for Magnetic Resonance

3.1 Introduction

3.2 Microarrays for Magnetic Resonance

3.3 Integrated Circuits

3.4 CMOS Frequency Division Multiplexer

3.5 Summary

References

Chapter 4: Wave Guides for Micromagnetic Resonance

4.1 Introduction

4.2 Wave Guides: Theoretical Basics

4.2 Designs and Applications

References

Chapter 5: Innovative Coil Fabrication Techniques for Miniaturized Magnetic Resonance Detectors

5.1 Wire‐Bonding – A New Means to Miniaturize MR Detectors

5.2 Microcoil Inserts for Magic Angle Spinning

5.3 Micro‐Helmholtz Coil Pairs

5.4 High Filling Factor Microcoils

5.5 Coil Fabrication Using Inks

References

Chapter 6: IC‐Based and IC‐Assisted

NMR Detectors

6.1 Technological Considerations and Device Models

6.2 Monolithic Transceiver Electronics for NMR Applications

6.3 Overview of the State‐of‐the‐Art in IC‐Based and IC‐Assisted

NMR

6.4 Summary and Conclusion

References

Chapter 7: MR Imaging of Flow on the Microscale

7.1 Introduction

7.2 Methods – Flow Imaging

7.3 Applications of Microscopic Flow Imaging

7.4 Discussion

Acknowledgments

References

Chapter 8: Efficient Pulse Sequences for NMR Microscopy

8.1 Introduction

8.2 Spatial Encoding

8.3 Contrast Mechanisms

8.4 Basic Pulse Sequences

8.5 Special Contrasts

References

Chapter 9: Thin‐Film Catheter‐Based Receivers for Internal MRI

9.1 Introduction

9.2 Catheter Receivers

9.3 Thin‐Film Catheter Receivers

9.4 Thin‐Film Device Fabrication

9.5 Magnetic Resonance Imaging

9.6 Conclusions

Acknowledgments

References

Chapter 10: Microcoils for Broadband Multinuclei Detection

10.1 Introduction

10.2 Microcoil‐Based Broadband Probe NMR Spectroscopy

10.3 An Engineer's Answers to the Questions

10.4 Conclusion and Outlook

Acknowledgment

References

Chapter 11: Microscale Hyperpolarization

11.1 Introduction

11.2 Theory

11.3 Microtechnological Approaches

11.4 Conclusion

References

Chapter 12: Small‐Volume Hyphenated NMR Techniques

12.1 Different Modes of Hyphenation

12.2 Types of Radio‐Frequency Coils Used for Small‐Scale Hyphenation

12.3 Hyphenation of NMR and Pressure‐Driven Microseparations

12.4 Electrically Driven Microseparations

12.5 Off‐Line Hyphenation of Microsamples with Microcoil Detection

12.6 Continuous Monitoring of

In Situ

Biological Systems

12.7 Studies of Microfluidic Mixing and Reaction Kinetics

12.8 Measurement of Flow Profiles in Flow Cells and Microchannels

12.9 Conclusion

References

Chapter 13: Force‐Detected Nuclear Magnetic Resonance

13.1 Introduction

13.2 Motivation

13.3 Principle

13.4 Force versus Inductive Detection

13.5 Early Force‐Detected Magnetic Resonance

13.6 Single‐Electron MRFM

13.7 Toward Nano‐MRI with Nuclear Spins

13.8 Paths Toward Continued Improvement

13.9 Comparison to Other Techniques

13.10 Outlook

13.11 Conclusion

References

Index

End User License Agreement

List of Tables

Chapter 01

Table 1.1 Compact permanent magnets and their characteristic summarized in a figure of merit

R

.

Chapter 02

Table 2.1 NMR and EPR sensor model order reduction.

Chapter 03

Table 3.1 The main parameters of the imaging experiment in a 9.4 T MR scanner employing the CMOS FDM.

Chapter 05

Table 5.1 Comparison of sensitivities obtained between conventional CPMAS probes and MACS.

Table 5.2 Overview of the casting procedure.

Chapter 06

Table 6.1 Table summarizing the small‐signal and small‐signal noise parameters of a MOSFET.

Table 6.2 Tables summarizing the small‐signal hybrid‐

model and the small‐signal noise parameters of a BJT in the forward active region.

Chapter 08

Table 8.1 Parameter combinations to achieve some typical values for the spatial resolution assuming

nx

= 1000.

Chapter 10

Table 10.1 Summary of the pros and cons of the conventional matched excitation and detection schemes of Figure 10.8.

Chapter 11

Table 11.1 Parameter descriptions for Eqs. (11.6)(11.7)(11.8)(11.9).

Table 11.2 Description of interactions generating alkali metal atomic energy levels.

Table 11.3 Collection of relevant data for the alkali metals and noble gases.

Chapter 12

Table 12.1 Typical column separation parameters and resulting peak concentrations for 1 nmol of analyte.

List of Illustrations

Chapter 01

Figure 1.1 NMR in inhomogeneous and in homogeneous fields. (a) Magnetic field strength

B

linearly varying with pixel position

x

. (b) Three pixels containing different numbers of NMR‐active nuclei at different positions

x

. (c) NMR spectrum observed in an inhomogeneous field (gray). For the case that the magnetic field

B

is homogeneous across each pixel (broken lines in a), the peak integral is proportional to the total magnetization at each pixel (black). (d) Spatially homogeneous magnetic field. (e) In a homogeneous field, the resonance signals from each pixel sum up at the same frequency.

Figure 1.2 Types of compact NMR magnets. (a) A stray‐field magnet is placed close to the object, here a car tire, for analysis of material properties. (b) A center‐field magnet accommodates the sample inside, here one of the 5 mm diameter sample tubes (foreground) containing the sample solution. The magnet is the most voluminous component of the NMR spectrometer (red). The sample tube is inserted into the magnet from the top.

Figure 1.3 Resonance curves illustrating the relation between quality factor

Q

and excitation bandwidth

Δω

in a resonance circuit

.

The larger the quality factor, the lower is the observed excitation bandwidth relative to the resonance frequency

ω

.

Figure 1.4 Compact stray‐field NMR magnets with iron yokes (gray) to shield the stray field on the bottom side. The object (not shown) approaches the sensor from the top. The stray field emanating from the top passes through the object with its field lines either parallel to the magnet surface (a,c) or perpendicular to it (b,d). Gaps are a simple way to shim the field by bending the field lines adjacent to the gap (c,d).

Figure 1.5 Stray‐field magnet with a planar sensitive slice. (a) Arrangement of magnets on an iron yoke. Both gaps are adjusted to define a planar region of constant magnetic field at a fixed distance above the magnet. (b) MiniMOUSE (left) and MicroMOUSE (right) fitted with a microcoil and suitable for depth profiling. (c) MiniMOUSE mounted on an optical displacement table to scan a 1 mm distance with 0.2 µm precision. The sensor has a gradient of 68 T/m and a resonance frequency of 17.1 MHz. (d) Photo of a painted car fender section showing the paint layer. (e) Depth profile through the paint layer of the car fender.

Figure 1.6 Stray‐field NMR magnets generating a sweet spot (light grey) external to the top surface of the magnet. (a) Cylindrical barrel magnet. The stray field is aligned perpendicular to the magnet surface. (b) NMR‐MOUSE with shim magnets generating a homogeneous gradient in a slice. The stray field is parallel to the magnet surface. The sensitive slice is centered above a RF coil.

Figure 1.7 Center‐field magnets for NMR. These magnets enclose the sample, which has to be inserted into the assembly. (a) Classical C‐shaped geometry. (b) Cylinder magnet with an iron housing and iron pole shoes. (c) Halbach magnet from trapezoidal magnet blocks (left). It can be approximated with six identical hexagonal bar magnets magnetized transverse to their axes (right). (d) Halbach magnet with shim plates that can be moved in and out in radial direction.

Figure 1.8 Magnetic field strength

B

from finite element numerical simulations evaluated in a circular center plane for three different radii

r

. When the radius is one‐third of the inner diameter of the magnet, the variation in the magnetic field is less than 1 mT. By increasing

r

to one‐third of the inner diameter and eventually to approximately the bore radius, roughly one order of magnitude is lost in each step.

Figure 1.9 Schematic drawings of compact NMR magnets (cf. Table 1.1 for magnet characteristics and literature references). The magnets are referred to by the name of the first author in the publication. (a) Moresi magnet assembled from rotatable cylinder magnets magnetized transverse to their axes. Metal plates (blue) with high permeability further homogenize the field in the center. (b) Armstrong magnet with tunable field strength. It is assembled from two Halbach magnets that can be rotated against each other. (c) Danieli magnet I. It is a Halbach magnet for magnetic resonance imaging similar to one ring of the Armstrong magnet but with shim magnets (blue) in the center. (d) Manz magnet approximating a simple spherical Halbach magnet. (e) Hugon magnet, which produces a magnetic field aligned with the direction of the magnet bore. (f) Sun magnet with the smallest dimensions of all magnets shown. (g) Windt magnet, better known as the NMR‐CUFF, which can be opened and closed to fit around a long pipe or a plant stem. (h) Danieli magnet II, a Halbach magnet from trapezoidal magnet elements with shim plates (blue) that can be moved for shimming. This design yields the highest figure of merit.

Figure 1.10 Simulated field maps of Halbach magnets constructed from hexagonal bar magnets and shimmed with ideal bar magnets placed in the third shell. (a) Halbach magnet with the inner two shells constructed from ideal magnet elements. (b) Halbach magnet with the inner two shells constructed from flawed magnet elements.

Figure 1.11 Approaches for active shimming of C‐shaped magnets. (a) The Anderson method employs sets of coils that produce magnetic field contributions following the terms of an expansion of the field variation in spherical harmonic functions. (b) The Kose method employs a two‐dimensional array of individually driven ring currents. (c) The McDowell method employs the magnetic fields from current‐driven individual wires crossing each other in parallel arrays.

Chapter 02

Figure 2.1 (a) Yee cell. (b) Perfect matching at the boundary of the computational domain.

Figure 2.2 Planar coil (a) EPR resonator with feed line. Layout on R6010LM, coil diameter

, trace width

, resonator length

, gap width

, and substrate thickness 0.635 mm. Electric (b) and magnetic (c) field distribution in the microstrip resonator.

Figure 2.3

‐field positions relative to the material and PML being modeled. The PML is four cells away from the EPR resonator top surface. The ground plane is not shown in the figure. The

and

fields at

are of metallic character because they are part of the EPR resonator.

Figure 2.4 Error due to input signal positioned inside PML (a). Original and reduced model reflection coefficients for EPR resonator (b). The blue curve is the relative difference between the original and reduced model.

Figure 2.5 Lumped source inside the FDFD mesh.

Figure 2.6 For each iteration, the error is accumulated over the desired frequency range. After 10 Arnoldi iterations, corresponding to a reduced‐order model of dimension 10, the model shows an error that has already decreased by four orders of mangnitude.

Figure 2.7 Compact coil models: (a) Lumped element‐based compact model represented as an equivalent electrical network; (b) Compact model derived from full‐scale FEM model via MOR.

Figure 2.8 (a) Scanning electron micrograph of a typical microcoil. (b) Schematic view of microcoil with main dimensions and material domains of the 2D coil model.

Figure 2.9 Inductance and resistance versus frequency for the microcoil: solid line – measured; dashed line – FE simulation.

Figure 2.10 System parameters to be preserved in the compact model.

Figure 2.11 Inputs and outputs of the coil model from [14].

Figure 2.12 (a) The results of the harmonic simulation of the reduced model for coil voltage output

, for different values of the electrical conductivity

, and with

. (b) The relative error of the compact model output with respect to the full model output.

Figure 2.13 (a) The results of harmonic simulation of the reduced model for magnetic flux density

at the coil center as an output for different values of relative permeability of sample

with

MS/m. (b) The relative error of the compact model output with respect to the full model output.

Figure 2.14 System‐level representation of the NMR detector.

Figure 2.15 (a) RF microcoil mounted on a PCB [13, 16]. (b) Sensor model connected to tuning and matching capacitors. The capacitors

and

should be adjusted to deliver 50

impedance at port

–

and resonance at

MHz.

Figure 2.16 Reflection coefficient

for different capacitor values. As the values of capacitors increase, the minimum point moves toward lower frequencies. For

and

,

is minimized, indicating resonance.

Figure 2.17 Sensitivity of

at 400 MHz to both the variation of the relative permeability of the sample and the electrical conductivita of the sensor's wire. The system is tuned for

and

.

Figure 2.18 Harmonic response of the circuit output impedance

. The relative errors of the circuits with 15 DOF and 30 DOF compact models connected to the circuit with respect to the full model are less then 1.25% over the entire frequency range.

Figure 2.19 Harmonic response of the magnetic flux density

output. The relative errors of the coil's

‐field with 15 DOF and 30 DOF compact models with respect to the full model are less than 1.3% over the entire frequency range.

Figure 2.20 (a) Minimal lumped model and (b) compact model representation of the receiver coil.

Figure 2.21 LNA circuit from [31] connected to the reduced‐order model of the receiver coil.

Figure 2.22 AC Gain of the LNA for both compact model and lumped model.

Chapter 03

Figure 3.1 Various head imaging arrays.

Figure 3.2 MR microscopy images acquired of (A) fish eggs. (B) Human skin patches.

Figure 3.3 SNR comparison of different single and multicoil configurations.

Figure 3.4 (a) Photo of McDougall's

et al

. 64‐element coil array. (b) Cross section of a standard printed circuit board.

Figure 3.5 Photo of the Watzlaw's

et al

. coil array.

Figure 3.6 Photo of Gruschke's

et al

. wirebonded mircocoil array.

Figure 3.7 Photo of Badilita's

et al

. hybrid MEMS and CMOS coil array.

Figure 3.8 Images of Anders's

et al

. CMOS coil array.

Figure 3.9 Block diagram of the entire MRI system employing eight‐channel frequency division multiplexer. (a) MR‐phased array of eight coils. (b) Eight‐channel frequency multiplexer. (c) Signal‐processing unit includes Analog-to-Digital Conversion (ADC) and digital signal processing.

Figure 3.10 Common‐gate common‐source LNA.

Figure 3.11 Double‐balanced Gilbert mixer.

Figure 3.12 Schematic diagram of the bandpass filter. The second‐order filter is realized using a multi‐feedback RC topology based on a two‐stage operational amplifier with indirect compensation.

Figure 3.13 Chip microphotograph of the CMOS FDM.

Figure 3.14 Characterizing the chip performance inside and outside the high magnetic field. (a) Channel's gain. (b) Channel's linearity.

Figure 3.15 MRI experiment setup.

Figure 3.16 Results of MRI experiment using the CMOS chip. A Matlab graphical interface was developed to compare the SNR of the original (image acquired using the Bruker system) MR image with that of the image reconstructed using the CMOS chip.

Chapter 04

Figure 4.1 Cross section of some typical planar transmission line geometries, including magnetic field lines of the TEM mode. (a) microstrip, (b) stripline, (c) microslot, and (d) parallel plate transmission line.

Figure 4.2 Single‐turn microstrip transmission line surface coil for MRI at 7 T (a,b) and images of the human brain (c).

Figure 4.3 Microstrip transmission line resonator (a,b) and images of the rat spine (c,d).

Figure 4.4 (A) Microslot probe. (a) Probe with housing removed, (b) scanning electron micrograph and light micrograph of the slotted microstrip detector. (B) Spectrum of sucrose in

acquired with a conventional (a) and the slotted microstrip probe (b).

Figure 4.5 Tumor spheroid microdevice. (a) CAD rendering of the microslot probe without (top) and with (bottom) the culture device in place; (b) Optical micrographs of tumor spheroids; (c) 600 MHz

NMR spectra obtained with the system.

Figure 4.6 Nonresonant microstrip detector (a) and MR images obtained from a sample of corn oil at 7 T with a resonant and a nonresonant microstrip detector (b).

Figure 4.7 Microfluidic NMR chip and stripline probe. (a) Conductor and dielectric layers of the stripline detector; (b) calculated RF magnetic field distribution; (c) photograph of the stripline detector and sample channel; (d–f) NMR flow probe assembly.

Figure 4.8 Stripline resonator for pulsed EPR (a) and nutation diagrams obtained with this system (b).

Figure 4.9 PTL resonator for microfluidic NMR. (a) CAD rendering of the resonator and the lab‐on‐a‐chip device; (b) experimental nutation curve at 10 W; (c) 150 mM acetate in

spectrum demonstrating excellent baseline resolution; (d)

spectrum of 2 µl of cell growth medium containing 20 mM glucose, and less than 1 mM concentrations of various amino acids. The spectrum was acquired in about 20 min.

Figure 4.10 Liquid‐state DNP system based on a Fabry–Perot microwave resonator combined with a stripline RF resonator. Adapted with permission from Ref. [113].

Chapter 05

Figure 5.1 (A) 3D microcoils with rectangular cross section fabricated by surface micromachining followed by post‐release folding. (B) Hollow core solenoidal microcoils fabricated in borosilicate glass – SEM images. (C) Process flow to fabricate solenoidal microcoils on capillaries by microcontact printing.

Figure 5.2 Photographs of solenoidal microcoils obtained by maskless lithography on capillary substrates and copper electroplating: (a) solenoidal coil; (b) tilted coil.

Figure 5.3 Bicone microcoils fabricated by 3D printing and Cu electroplating.

Figure 5.4 (a) Wire‐bonded microcoil fabricated on glass substrate and mounted on a PCB for tuning and matching. (b) wire‐bonded microcoil integrated with a microfluidic channel (filled with blue ink for visibility). (c) MRI of polymer 50 µm diameter polymer beads (SNR higher than 50 for 20 acquisitions).

Figure 5.5 (a) Schematic of magic angle coil spinning insert. The microcoil insert is fabricated by handwinding a copper wire around a capillary and the leads of the wire are soldered to a high

capacitor.

Figure 5.6 (a) Ratio of sensitivity of inductively coupled system against the direct microcoil NMR system, as a function of the quality factor of the microresonator and the volume ratio

/

. The quality factor of the probe,

, is assumed to be 200 [33]. (b) Effect of imperfect coil tuning on the enhancement of the sensitivity achieved by MACS. The Larmor frequency in this case is 300 MHz.

Figure 5.7

Comparison of sensitivity on

1

H MAS NMR spectra using minute samples of powdered

L

‐alanine.

(a) Spectrum acquired (total acquisition time

12 h) using a standard 2.5 mm rotor. The sample, centered on the rotor, weighed around 0.41 mg. Most of the signal observed shows up from the housing and rotor background. (b) Spectrum after subtraction of the background signal (total acquisition time

24 h), using the same setup as in (a) (spinning sidebands are labeled with asterisks). The residual signal comes from the

L

‐alanine signal including some artifacts and has a signal‐to‐noise ratio (SNR) of 270. (c) Spectrum obtained using the MACS technique on a 7 mm rotor, using 0.15 mg of sample (acquisition time

8.5 min). The signal (SNR

110) originates only from the

L

‐alanine sample and contains a center band (shown expanded in the inset) as well as spinning sidebands at multiple intervals of the spinning frequency.

Figure 5.8 Process flow for the fabrication of wire‐bonded MACS NMR detectors.

Figure 5.9 (a) Coil‐winding process using an automatic wire‐bonder; (b) wire‐bonded MACS resonators. The wire‐bonded coil contributes to the inductance,

, and the interdigitated finger structures contribute to the capacitance,

, of the resonant LC circuit.

Figure 5.10 (a) Fabricated MACS inserts having a dimensions of

mm

3

; (b) MACS arrangement with the insert being mounted on the plug and sealed inside the rotor.

Figure 5.11 (a) Comparison of the signal collected from water sample using the probe coil of the Bruker scanner and the MEMS microcoil as pickup coils, respectively, proving that the NMR signal comes from the microcoil. (b)

1

H NMR spectra of

Drosophila

pupae. The spectra was acquired with a total of three pupae in a

m‐microcoil placed inside a standard 7‐mm MAS probe, spinning at 500 Hz for the single‐pulse experiment, and at 371 Hz for the PASS experiment, respectively.

Figure 5.12 Double‐resonant MACS insert: (a) The circuit diagram used for the double‐resonance MACS experiments; (b) MACS resonator for double‐resonance experiment; (c) description of the arrangement inside the Shapal‐M holder.

Figure 5.13 (a) Nutation frequency dependencies for the

1

H and

13

C nuclei with the RF power at

 T in a Varian 5‐mm CP‐MAS probe. Blue color depicts the

13

C nutation frequencies with (filled squares) and without (open squares) the doubly tuned MACS resonator, and the brown color depicts the

1

H nutation frequencies with (filled circles) and without (open circles) the MACS resonator. The data points are fitted by square root function,

, and the coefficient of fitting is shown for each curve; (b) A

13

C CP‐MAS spectrum of polycrystalline

13

C

15

N‐labeled

L

‐alanine obtained without the doubly tuned MACS resonator; (c) A

13

C spectrum of the identical sample obtained by employing the doubly tuned MACS resonator. The spinning frequency is

 kHz in both cases.

Figure 5.14 (a) Schematic sketch of a Helmholtz coil pair. The current direction in both loops is identical. The total magnetic field is composed by the superposition of the two individual fields, whereas the homogeneity of the summed field depends on the spacing and the radii of the coils employed. (b) Current distribution of the fundamental, in‐phase resonance at

(corresponding to (a)) and of the second, out‐of‐phase resonance at

for one instant in time. The latter case results in a linear field gradient.

Figure 5.15 The micromachined Helmholtz pair. (a) Schematic illustration of the fabrication steps for the multiturn Helmholtz pair. (b) Micrograph of a fabricated device showing the spherical sample chamber in between the two coils.

Figure 5.16 The SU‐8‐based designs presented. (A) Process steps for the fabrication of a micro‐Helmholtz coil. (B) Photographs of a fabricated solenoid (a), a planar (b), and a Helmholtz coil (c), derived from the process illustrated in (A).

Figure 5.17 The flow cell design presented. (a) Photograph of the upper half of the coil and the tubing channel below. (b) Photograph showing the tubing in between the coil pair.

Figure 5.18 (a) The silicon cylinder spiral coil consists of various individual micromachined elements, arranged in a concentric manner. (b) SEM closeup of the conductors traces.

Figure 5.19 (a) Photograph of the wire‐bonded Helmholtz coil chip. An empty ASSAI was placed in between the wire‐bonded coil pair. (b) SEM image of a microcoil, suspended on a post base plateau made from photoresist. The coil was wire‐bonded from

 µm diameter insulated copper wire.

Figure 5.20 Illustration showing microcoils with mechanical support structures inside and outside of the detection volume.

Figure 5.21 Decrease of SNR with increase of coil wall thickness.

Figure 5.22 (a) Fabricated microcoil; (b) pink: magnetic resonance spectrum of water, gray: nutation spectrum of water.

Figure 5.23 (a) Unplated ink‐jet‐printed silver coil. (b) Ten micrometer electroplated gold on the coil. (c) Details of the electroplated coil.

Figure 5.24 Structuring of coil layer 1. (a) Direct, droplet‐based silver nano‐particle printing. (b) Sintered, conductive silver tracks. (c) Thick‐film permanent photoresist deposition. (d) Backside exposure and resist development. (e) Fabricated electroplating molds. (f) Copper electroplating.

Figure 5.25 Structuring the coil interconnect that closes the loop on layer 2. (a) Passivation of the electroplated structures via lamination of photo‐definable dry resist. (b) Ink‐jet printing of a wet removable optically opaque mask. (c) Front‐side, flood exposure. (d) Developed passivation layer, showing the vias. (e) Silver ink‐jet printing of the interconnect seed layer. (f) Copper electroplating to reduce the electrical resistance of the printed interconnect.

Figure 5.26 (A) The developed trenches in the resist before they are filled with copper. (B) The final coil with two electroplated layers. One can see the huge difference between a confined (layer 1) and an unconfined (layer 2) electroplating process.

Figure 5.27 (a) Rapid prototyped MR coil insert, made using laser‐cut PMMA, a self‐etched PCB, and nonmagnetic tuning and matching capacitors attached. (b) Comparison between an optical (left) and a high‐resolution MRI (right).

Chapter 06

Figure 6.1 Illustration of the integration of both the detection coil and the transceiver electronics into a single ASIC in the IC‐based NMR approach.

Figure 6.2 (a) Small‐signal equivalent model of the noise free MOS transistor and (b) noise model of the intrinsic noise sources of a MOSFET including the channel noise and induced gate noise around the noise‐free model of (a).

Figure 6.3 (a) Small‐signal equivalent model of the noise‐free BJT transistor and (b) noise model of the intrinsic noise sources of a BJT including the base and collector shot noise as well as a noise source corresponding to the white noise of the base contact resistance.

Figure 6.4 Architecture of a conventional NMR detection system.

Figure 6.5 Equivalent circuit diagram for an LNA directly connected to the detection coil.

Figure 6.6 Simplified schematics of (a) a wideband single‐ended common base/gate amplifier, (b) a wideband single‐ended common emitter/source amplifier, (c) a wideband differential pair amplifier, and (d) a wideband differential common source amplifier with current reuse.

Figure 6.7 (a) Simulated noise figure for all wideband LNA topologies as a function of frequency when used together with a high‐

microcoil (

at 500 MHz) and for a bias current of 10 mA in each individual input transistor. (b) Simulated noise figure for all wideband LNA topologies as a function of frequency when used together with a low‐

microcoil (

at 500 MHz) and for a bias current of 10 mA in each individual input transistor. The dashed line in each figure indicates an operating frequency of 500 MHz.

Figure 6.8 Simulated noise figure of all wideband LNAs as a function of the bias current in each individual input transistor at an operating frequency of 500 MHz when used together with (a) a high‐

microcoil (

at 500 MHz) and (b) a low‐

microcoil (

at 500 MHz).

Figure 6.9 Simulated noise figure for the different narrowband LNA topologies used in combination with (a) a high‐

coil (

and

) and (b) a low‐

coil (

and

). In both cases, the bias current is 0.5 mA for each individual input transistor. In all simulations, the optimum tuning capacitors for the operating frequency of 500 MHz were used.

Figure 6.10 Simulated noise figure for all narrowband LNA topologies used in combination with (a) a high‐

detection coil (

and

) and (b) a low‐

coil (

and

) as a function of the bias current in each individual input transistor.

Figure 6.11 (a) Photograph and micrograph of the palm NMR system and the utilized ASIC presented in [19]. (Sun

et al

. 2011 [19]. Reproduced with permission of IEEE.) (b) Photograph, micrograph, and concept drawing of the point‐of‐care NMR system, the utilized ASIC, and the target application presented in [38].

Figure 6.12 (a) Concept drawing of the probe head and micrograph of the fully integrated NMR detection chip presented in [37]. (b) MR image with a nominal resolution of 9.6

m of a phantom consisting of a small copper sulfate doped water‐filled capillary inside a larger one taken in an imaging time of 7.5 h with the chip presented in [37].

Chapter 07

Figure 7.1 Schematic representation of the ToF sequence. Two different labeling methods are shown. Method 1 labels use a single pulse to label a plane, whereas method 2 uses a DANTE sequence to label a set of parallel planes. The subsequent imaging sequence is a FLASH sequence. The labeling methods can be repeated with different evolution times

and different directions.

Figure 7.2 Schematic representation of the PC sequence. The black part of the gradients is designed to generate flow‐compensated images. The blue gradients generate the flow‐encoding.

Figure 7.3 Schematic representation of outflow effects during slice selection.

Figure 7.4 The center part of the figure shows the reconstructed geometry of model A, reconstructed from the proton density data. The two red boxes represent the areas where the tagging pulse was applied. On the left‐and right‐hand side, the resulting images of the tagged areas are shown for three different evolution times with and without flow. The color represents the signal intensity, which corresponds to the spin polarization (blue – no polarization, red – full polarization).

Figure 7.5 The left‐hand side shows the complete model B, including the connecting tubes. The center shows the reconstructed geometry of the aneurysm. On the right‐hand side, the resulting images of the tagged area are shown for three different evolution times for the measurement and a CFD simulation (two upper rows). The color represents the signal amplitude, which corresponds to the spin polarization (blue – no polarization, red – full polarization). The bottom row shows the difference between both images relative to the maximum signal of all measured images. Here, the color encodes the difference (blue = 0%, red = 100%).

Figure 7.6 Velocity profiles for a cylindrical tube with a diameter of

mm. The left‐hand side shows an

cross section, the lower right‐hand side the velocity of one slice (no. 12), and the upper right‐hand part shows the cross section through the center of this slice.

Figure 7.7 The left‐hand side shows a picture of the aneurysm model C. The center part shows the reconstructed geometry. The green box marks the area of the aneurysm. The right‐hand side shows the absolute velocity vector field in a streamline representation. The color encodes the absolute value of the velocity.

Figure 7.8 Color‐encoded absolute value of the WSS for aneurysm model C from two different perspectives. The flow is from bottom to top. Areas with high and low WSS are labeled.

Figure 7.9 ToF images for three different delays

during the pressure pulse (

520, and 1000

). Each image shows the displacement of the tagged slice (dark, low signal intensity) inside of the tube (bright, strong signal intensity). The tagged slice is tracked by 32 points, represented in yellow. The reference slice without flow is shown by the green points.

Figure 7.10 (a) Velocity profiles for 52 different delays

. The color encodes the velocity; negative velocities are represented by pink color. (b) The volume flow and the maximum velocity are shown as a function of the delay

Chapter 08

Figure 8.1 Principle of spatial encoding with a magnetic field gradient: (a) in a constant magnetic field, the Larmor frequency of all spins is identical, so all “sing with the same voice.” (b) If a constant gradient is applied, the field changes linearly with position, the frequency spectrum will correspond to a projection over the body.

Figure 8.2 Basic principles of

k

‐space imaging. Signal is measured in the frequency domain called

k

‐space, the image is related to the acquired data by a two‐dimensional Fourier transformation. Explanation of parameters are found in the text.

Figure 8.3 Radial sampling in NMR imaging: the basic relationship between the acquired data and the image remains the same as in 2DFT imaging. As radial sampling leads to a non‐rectilinear grid of data points, transition from

k

‐space data to the image domain is performed using a nonuniform Fast Fourier Transform (NuFFT).

Figure 8.4 Pulse profiles (a–d) and the corresponding frequency response in the low‐tip angle approximation. A rectangular pulse (a) will have a sinc‐shaped frequency response (e), a truncated sinc pulse (b) will create pronounced wiggles (f), filtering with a half‐sine leads to (c), which creates the frequency profile shown in (g) (blue line). The red profile in (g) corresponds to the profile for a 90° pulse generated by simulation of the Bloch equation and shows that the small‐tip angle approximation holds reasonably well. (d) shows a Gaussian pulse, which will generate a Gaussian frequency response (h).

Figure 8.5 Frequency profile of a Gaussian pulse with variable flip angles. Deviations from the small‐tip angle approximation become visible at ∼

α

 = 90°.

Figure 8.6 Dependency of the slice profile Δ

z

1,2 on the frequency profile Δ

f

and the gradient slope

G

z

1

and

G

z

2

. Note that changing the gradient will also change the position of the slice.

Figure 8.7 (a) Effect of a local susceptibility disturbance: the susceptibility‐induced field will be superimposed on the readout gradient, which will result in local distortions and/or ambiguities. (b) Signal attenuation due to intravoxel dephasing as a function of spatial resolution. The gradually flatter susceptibility‐induced field modulation (black line) leads to a monotonically decreasing off‐resonance phase d

ϕ

. A homogeneous distribution of spins is assumed. The resulting signal attenuation

I

T

2

* will be appreciable at a course resolution (blue) and gets less with increasingly finer resolution (red, green).

Figure 8.8 Susceptibility‐induced field by a spherical (a) and elongated (b) body with a susceptibility difference of 10 · 10

−6

to its surroundings. Note that the elongated body oriented along

B

0

shows a susceptibility effect only around its ends.

Figure 8.9 Image of a mouse brain slice with a small air bubble (red arrow). Image was acquired on a 7 T system (Bruker Biospin, Germany) with a cryocoil. Parameters: TE = 16 ms, TR = 300 ms,

n

av

 = 16, four slices, 22 × 22 × 100 µm

3

, flip angle = 50°, matrix: 550 × 680, FOV: 12 × 15 mm

2

, 54 min total acquisition time.

Figure 8.10 Sequence diagram of a (a) single‐spin echo and (b) multiecho (RARE) imaging sequence. RF denotes the radio frequency pulses and signals, Acq the acquisition time,

G

S

,

G

R

, and

G

P

denote the slice selection‐, readout‐, and phase‐encoding gradients, respectively.

Figure 8.11 Comparison of a FISP image (a) with a RARE image (b) of a kiwi fruit. Experimental parameters: (a) RF‐spoiled FLASH 2D: TR 60 ms, TE 4 ms, flip angle 30°, SLTH 1 mm, matrix: 256 × 256, FOV: 60 × 60 mm

2

, 234 × 234 µm

2

resolution, acquisition time 15 s. (b) RARE 2D: TR 3500 ms, TE 60 ms (ESP 20 ms), Rare Factor 6, SLTH 1 mm, matrix: 256 × 256, FOV: 60 × 60 mm

2

, 234 × 234 µm

2

resolution, acquisition time 2 min 27 s.

Figure 8.12 Sequence diagram of a FLASH sequence. TR represents the repetition time and S refers spoiler gradients to suppress unwanted residual transverse magnetization.

Figure 8.13 Steady‐state signal intensity SI for RF‐spoiled FLASH as a function of the phase d

φ

increment between successive pulses for (a)

T

1

 = 200 ms,

T

2

 = 40 ms, TR = 10 ms, flip angle 30°, (b) same parameters, but 60° flip angle. The horizontal red line represents the signal intensity of a spoiled FLASH measurement, the black line represents a refocused FLASH (FISP).

Figure 8.14 Two slices of a 3D FISP data set of a cotton fruit for visualizing the cotton seed anatomy. Parameters: 7T Bruker Biospec with cryocoil, flip angle 20°, TR 40 ms, TE 6.8 ms, 8 averages, acquisition time 14 h 13 min. Matrix size 500 × 400 × 400, FOV 25 × 20 × 20 mm

3

, resolution 50 × 50 × 50 μm

3

.

Figure 8.15 Sequence diagram of a CE‐FAST(PSIF) sequence. The readout and slice selection gradients are time reversed; therefore, no signal is generated from the RF pulse immediately preceding the acquisition time, all signals come from indirect refocusing pathways over more than one TR. A spoiler gradient

S

is normally added to avoid interference with the tail of the free induction decay formed by the preceding RF‐pulse.

Figure 8.16 Sequence diagram of a DESS sequence, in which both direct and indirect signal pathways are read out. The diagram shows that the sequence consists of a back‐to‐back arrangement of a FLASH (Figure 8.12) with a CE‐FAST (Figure 8.15) sequence separated by the shaded part S of the readout gradient.

Figure 8.17 Sequence diagram of a fully balanced steady‐state free precession (bSSFP) sequence (trueFISP, FIESTA), in which all pathways collapse into one signal pathway.

Figure 8.18 Sequence diagram of an echo planar imaging (EPI) sequence. Transverse magnetization is generated by a single excitation pulse, multiple gradient echoes are generated by time reversal of the readout gradient. Phase encoding is applied in short blips during reversal of the readout gradient. An initial gradient lobe along GP leads to initial dephasing such that the

k

‐space zero point is acquired in a later echo (typically at the center of the echo train).

Figure 8.19 Sequence diagram (a) and

k

‐space sampling pattern (b) of a UTE sequence. Data acquisition starts immediately after the short (typically rectangular) RF pulse, gradients are varied such that

k

‐space is sampled in a “Koosh‐ball” pattern (b).

Figure 8.20 Axial images through the stem of a

Hamamelis

fruit acquired with (a) UTE compared with (b) FISP acquired at 7 T (Bruker Biospec) using a quadrature coil. UTE was chosen to reduce the quite severe susceptibility artifacts caused by air enclosures. Parameters: UTE: matrix: 256 × 256 × 256, FOV: 25.6 × 25.6 × 25.6 mm

3

, 100 × 100 × 100 µm

3

resolution, TR 5 ms, TE 20 ms, flip angle 5°, 205 634 projections, no averaging, total acquisition time 17 min 8 s. FISP: matrix: 256 × 256 × 256, FOV: 25.6 × 25.6 × 25.6 mm

3

, 100 × 100 × 100 µm

3

resolution, TR 30 ms, TE 4 ms, flip angle 20°, total acquisition time 32 min 46 s.

Figure 8.21 (a) Basic Stejskal–Tanner sequence with two diffusion gradient lobes

G

D

placed symmetrically around the refocusing pulse in a spin echo sequence. (b) shows the phase graph of the sequence: an arbitrarily selected spin will acquire a linear phase during the constant first gradient lobe, the phase will be inverted by the refocusing pulse leading to refocusing at the echo time. (c) illustrates that the diffusion weighting of the signal is proportional to the volume of a rotational body generated by rotating (b) around the time axis.

Figure 8.22 An example of DTI with NMR microscopy. Images show a comparison between a hippocampal slice preparation of a normal mouse compared to a mouse treated with kainate. (a) the reference image, (b) the images generated by fiber tracking of a DTI measurement, and (c) the histological comparison. The diffusion of the cortical layers and the sprouting of the mossy fibers are clearly demonstrated in the DTI images. Parameters: (a) RARE acquisition, TE = 17 ms, TR = 2000 ms, av = 32, 3 slices, 20 × 24 × 100 µm

3

, RARE factor 6, matrix: 512 × 480, FOV: 10 × 11.38 mm

2

, total acquisition time 1 h 25 min. (b) DTI‐EPI, TE = 39 ms, TR = 3000 ms, four segments, av = 16, rep = 1, four slices, 60 directions, resolution 39 × 39 × 100 µm

3

with interpolation from 50 × 50 × 100 µm

3

,

b

‐value: 1000 s/mm

2

, matrix: 320 × 190, FOV: 12.48 × 7.41 mm

2

, total acquisition time 3 h 31 min.

Figure 8.23 Diffusion weighting in an idealized spin echo experiment, in which all pulses are assumed to be infinitesimal small and the diffusion gradient

G

D

is “on” throughout the sequence. Panel (a) shows the sequence diagram, (b) the diffusion weighting WD(

t

). The gradient is chosen such that the spatial resolution of around 9.6 µm equals the diffusion displacement for free water over the acquisition time

t

acq

 = TE. Panel (c) shows the resulting point spread function PSF(

ω

). The dots represent the discrete sampling grid, the red sinc function represents the PSF of the unfiltered signal. The full width at half maximum (fwhm) is about twice as broad as the mere displacement.

Figure 8.24 Diffusion weighting as a function of the dimensionless spatial resolution represented as multiples nd of the diffusion displacement over the acquisition time. The three cases shown below the plot represent different values for the spatial resolution and the corresponding gradient amplitude for different values for

t

acq

and different diffusion coefficients representing free water and bound water.

Figure 8.25 Diffusion weighting in a UTE‐type acquisition. Signal is read out using a radial out approach with a single gradient lobe GD. The corresponding diffusion weighting function (free water over 20 ms) has a similar shape compared to that in Figure 8.23, but without the initial signal attenuation. As the trajectory covers only half of

k

‐space, it has to be complemented by a symmetric signal (dotted line in b), which effectively doubles the spatial resolution. The fwhm of the PSF is only slightly broader than the diffusion displacement.

Figure 8.26 Observed linewidth for acquisition under a constant gradient as a function of the acquisition time

t

acq

. Graphs represent gradient amplitudes of

G

 = (a) 50 mT/m, (b) 100 mT/m, (c) 200 mT/m, (d) 600 mT/m, and (e) 2000 mT/m. Although with a constant gradient the nominal spatial resolution increases with

t

acq

, the observed line width flattens out due to diffusion weighting.

Figure 8.27 Principle of (a) flow encoding and (b) flow compensation. A bipolar gradient will refocus the phase graph of stationary magnetization (red) but generate a velocity‐dependent phase

φ

FLOW

for moving spins (blue).

Figure 8.28 Susceptibility‐weighted image of the trabecular structure of a piece of bone. In addition to some susceptibility artifacts induced by air (bottom arrow), the trabecular structure is intensified by the susceptibility effect between bone and soft tissue. The top arrows point on a region where susceptibility got enhanced by binding iron oxide particles to the bone tissue. Parameters: images acquired at 7 T with a cryocoil, 3D FLASH, matrix: 800 × 256 × 128, FOV: 30 × 10 × 12 mm

3

, 38 × 39 × 94 µm

3

, TR 40 ms, TE 6.19 ms, flip angle 20°, no averaging, total acquisition time 21 min 50 s.

Chapter 09

Figure 9.1 MRI signal reception: (a) elongated loop catheter coil and (b) precessing magnetization vector.

Figure 9.2 Detection sensitivity patterns for a two‐wire coil: (a)

H

x

, (b)

H

y

, and (c)

.

Figure 9.3 Equivalent circuits of resonant detectors: (a) and (b) series capacitor and mutual inductance matching, (c) and (d) passive and active decoupling, and (e) coaxial output.

Figure 9.4 Excitation of standing surface waves on an immersed linear conductor by an electric field: (a) physical arrangement and (b) equivalent circuit model.

Figure 9.5 Prevention of surface wave resonances: (a) insertion of current‐blocking impedances, (b) transformer segmentation, and (c) use of magneto‐inductive waveguides.

Figure 9.6 Planar spiral inductors with (a) hybrid and (b) integrated capacitors.

Figure 9.7 Planar interconnects: (a) microstrip, (b) coplanar waveguide, (c) electromagnetic bandgap waveguide, (d) complete receiver‐combining resonant detector with EBG output cable.

Figure 9.8 Magneto‐inductive receivers: (a) single figure‐of‐eight element and (b) complete receiver‐combining resonant detector with MI output cable.

Figure 9.9 (a) Double‐sided patterning for thin‐film circuit fabrication and (b) catheter construction.

Figure 9.10 Completed thin‐film circuits: (a) spiral microcoil, (b) and (c) EBG waveguide unmounted and mounted on catheter, (d) and (e) MI waveguide unmounted and mounted on catheter, (f) and (g) resonant transducer, flat, and attached to MI catheter.

Figure 9.11 Magneto‐inductive catheter receiver: (a) mounted on guidewire and (b) passing through biopsy channel of nonmagnetic duodenoscope.

Figure 9.12 Frequency variation of scattering parameters for resonant spiral microcoil, (a) during and (b) after tuning and matching for operation at 1.5 T.

Figure 9.13 Frequency variation of scattering parameters for resonant spiral microcoil, (a) after tuning and matching for operation at 3.0 T and (b) during active decoupling.

Figure 9.14 Frequency variation of scattering parameters for (a) 2 m long EBG waveguide and (b) EBG receiver tuned for operation at 1.5 T.

Figure 9.15 Frequency variation of scattering parameters for (a) 2 m long MI waveguide and (b) MI receiver tuned for operation at 1.5 T.

Figure 9.16 Microcoil evaluation at 3.0 T: (a) phantom‐imaging experiment and (b) axial array coil image showing cylindrical gel phantom with microcoil at center.

Figure 9.17 Microcoil evaluation at 3.0 T: (a) and (b) axial images of gel phantom obtained using microcoil, before and after correction for radial sensitivity variation and (c) comparative variations of SNR for microcoil and array coil.

Figure 9.18 EBG catheter receiver evaluation at 1.5 T: (a) catheter receiver inside

in vitro

liver specimen and (b) uncorrected axial slice image of duct al system.

Figure 9.19 MI catheter receiver evaluation at 1.5 T: (a) phantom specimen with catheter spiraled on cuboid and (b) coronal slice image beneath catheter.

Figure 9.20 MI catheter receiver evaluation at 1.5 T: (a) phantom specimen with catheter in dummy duodenoscope tip and (b) sagittal slice image through catheter.

Chapter 10

Figure 10.1 Various coil geometries that have been exploited for small‐volume NMR spectroscopy.

Figure 10.2 (a) Back view (microfluidics) and (b) front view (RF‐electronics) of a 25 nl detection volume NMR chip (1.5 cm × 1.5 cm); (c) front view (electronics) of a 6 nl detection volume NMR chip (0.9 cm × 0.9 cm) with side inlets for the microfluidics connections.

Figure 10.3 Multipurpose NMR chip with two detection areas. (a) Schematics of the chip, front view with fluidic channels (top), and back view with copper electrodes and coil (bottom). (b) Photographs microfluidic 1.5 cm × 4.5 cm NMR chip inside slider (left) and aluminum holder on top of a sacrificed probe (right). (c) RF circuit of the BBC setup, a microcoil having a DC resistance

ρ

in series with an optional variable resistor (

R

). See Fratila

et al

. [36] for further details of the chip and coil. Tx = transmitter; Rx = receiver.

Figure 10.4 NMR spectra with the broadband microcoil circuit in the full Larmor frequency range at 9.4 T, from 400 MHz (

1

H, b) to 40 MHz (

15

N, b). The amine group of the

15

N‐labeled urea shows characteristic doublet and triplet coupling patterns, respectively, because of the

1

J

NH

coupling (R.M. Fratila and A.H. Velders, unpublished data).

Figure 10.5 Heteronuclear 1D and 2D NMR BBC experiments on a single non‐resonant microcoil of neat acetic acid‐2‐

13

C (a,b) and aqueous NaPF

6

solution (c,d). 1D: (a)

13

C‐NMR spectra obtained using different pulse sequence schemes: coupled (1), coupled with nuclear Overhauser effect (NOE) enhancement (2), decoupled without NOE enhancement (3), and decoupled with NOE enhancement (4); (c) 1D

19

F‐NMR spectra coupled (1) and decoupled (2). 2D: (b)

1

H

13

C‐HSQC spectra and (d) 2D

19

F

31

P‐HSQC spectra; the 2D spectra have been acquired in coupled (red) and decoupled (blue) mode.

Figure 10.6 (a) Drawing of a classical saddle‐shaped NMR coil and (b) its simulated resistance and inductance versus frequency.

Figure 10.7 (a) Simple three‐element lumped equivalent circuit modeling the frequency behavior of an NMR coil, (b) broadband excitation of an NMR using a voltage source, and (c) broadband readout scheme for an NMR coil.

Figure 10.8 (a) Classical excitation scheme for an NMR coil using a remotely placed, impedance‐matched power amplifier (PA), (b) classical detection scheme for an NMR coil using a remotely located, impedance‐matched, low‐noise amplifier (LNA), and (c) classical NMR front‐end combining the matched excitation and detection using cross‐coupled diodes in combination with a

λ

/4‐TRL‐based impedance transformer for decoupling.

Figure 10.9 (a) Illustration of the non‐matched connection between probe head and PA for transmit and (b) illustration of the effective impedance presented by the probe had to the PA, which is largely affected by the length of the connecting TRL.

Figure 10.10 (a) Illustration of the non‐matched connection between probe head and LNA for receive and (b) illustration of the effective impedance presented by the LNA to the probe had, which is identical to the TRL's characteristic impedance

Z

0

.

Figure 10.11 (a,b) Simulated effective impedances versus frequency originating from a coil model according to Figure 10.7a and transformed by TRLs of three different electrical lengths, (c) simulated absolute value of the coil current during transmit versus frequency for three different TRL lengths, and (d) simulated absolute value of the LNA input voltage versus frequency for three different TRL lengths.

Figure 10.12 (a,b) Broadband excitation and detection of an NMR coil with the power amplifier (PA)/low‐noise amplifier (LNA) being directly connected to the coil and (c) circuit realization of the PA as an H‐bridge with LNA decoupling switches to prevent damage to the LNA input transistors during transmit.

Chapter 11

Figure 11.1 (a) The polarization of electron and proton spin states in percent plotted versus temperature. (b) Plot of the Overhauser enhancement

and coupling factor

as a function of

. Shown are the curves for the dipolar (solid) and scalar coupling as well as the influence of the rotational correlation time

and the loss factor

.

Figure 11.2 Hyperpolarized gas example applications. (a) Principle of xenon as a biosensor. Free Xe in water appears at 193 ppm, whereas cryptophane‐bound Xe appears at 70 ppm. The signal of bound Xe is distributed between linker‐free and functionalized cryptophane. With addition of a ligand (avidin), a new signal appears attributed to cryptophane‐bound Xe

linker binding. (Spence

et al

. 2001 [21]. Reproduced with permission of PNAS. Copyright (2001) National Academy of Sciences, USA.) (b) Example of

3

He lung imaging. Three mice (naive and two ovalbumin (OVA) sensitized) were subjected to methacholine (MCh) challenge. Airway closure and ventilation loss were observed in the OVA‐sensitized mice (arrows). Experiments were possible in the considerably smaller mouse lung as a consequence of using hyperpolarized

3

He MRI.

Figure 11.3 Fractions of ortho (dashed black curve) and para (solid red curve) molecules in hydrogen gas as a function of temperature.

Figure 11.4 Pictorial representation of PASADENA (a) and ALTADENA (b) experiments. Comparison between use of thermal H

2

and

p

‐H

2

is provided: populations of energy levels are drawn as blue‐colored stacks for H

2

and red‐colored stacks for

p

‐H

2

. The top row shows statistical populations of spin levels in both H

2

and

p

‐H

2

molecules before hydrogenation. The middle row displays the result of hydrogenation and adiabatic transport (for ALTADENA) on the populations in the hydrogenated target molecule, and the direction of allowed NMR transition between molecular spin states. On the bottom row: NMR spectra resulting from PHIP, highlighting the distinctive antiphase character of the doublets.

Figure 11.5 Schematic of the SEOP process. Circularly polarized light enters an optically thick cell containing alkali metal vapor, noble gas, and buffer gas. The entire cell is immersed in a weak magnetic field of strength

(approx. tens of Gauss). Important spin polarization‐dependent interactions include gas–wall, alkali–alkali, and alkali–noble gas collisions.

Figure 11.6 Energy level diagram for

87

Rb (I = 3/2). Level splitting occurs as a result of (a) spin–orbit coupling, (b) hyperfine coupling, and (c) Zeeman splitting (at approx. tens of Gauss). The energy axis is not to scale.

Figure 11.7 (a) Illustration of a combined 140 GHz,

cylindrical ESR cavity and an open helical structure for RF excitation. (Maly et al. 2008 [31]. Reproduced with permission of AIP.) (b) Sketch of a Fabry–Perot resonator facilitating a movable planoconcave mirror made from a gold coated‐fused silica lens. (Morley et al. 2008 [80]. Reproduced with permission of AIP.) (c) Schematic of a Fabry–Perot resonator for liquid‐state DNP at 9.2 T and a stripline resonator for NMR detection.

Figure 11.8 Scanning electron microscope (SEM) image of a micromachined all‐silicon Fabry–Perot resonator based on cylindrical Bragg mirrors. (a) Fabry–Perot architecture showing four silicon layers as well as the alignment trenches for the optical fibers coupling to the cavity. (b) Highlighted details of the cavity elements. A fiber rod lens has been introduced for focusing the light beam in another transverse direction.

Figure 11.9 (a) SEM image of a photonic crystal resonator for operation at 100 GHz with loaded

‐factors of 5000. A high‐resistivity silicon substrate was perforated with holes arranged in triangular lattice by means of deep reactive ion etching. (Otter

et al

. 2014 [94]. Reproduced with permission of Elsevier.) (b) Photonic resonator inside a metallic ridge terahertz waveguide. (top) Schematic profile if the waveguide and design of the corrugation. (bottom) Photograph of the fabricated waveguide including a zoomed‐in detail of the resonant structure.

Figure 11.10 (a) Schematic of a THz resonator based on an inhomogeneous Bragg grating incorporated into the walls of a metallic slit waveguide. In the center of the gap, a thin plate of silicon provides improved lateral field confinement. (Gerhard

et al

. 2010 [102]. Reproduced with permission of OSA Publishing.) (b) Schematic of sub‐millimeter per terahertz‐integrated dielectric waveguide based on high‐resistive silicon. The silicon substrate is structured by DRIE forming a dielectric waveguide suspended by periodic support beams, followed by a thermal bond to a glass substrate.

Figure 11.11 (a) Spin and concentration sensitivity for one‐turn planar microcoils as a function of the coil diameter

. (Boero

et al

. 2003 [107]. Reproduced with permission of AIP.) (b) Sketch of a variety of planar MW resonators. (1) Planar ohm‐shaped resonator. (Narkowicz

et al

. 2005 [109]. Reproduced with permission of Elsevier.) (2) Split‐ring resonator. (Pendry

et al

. 1999 [110]. Reproduced with permission of IEEE.) (3) Half‐wavelength resonator. (Torrezan

et al

. 2009 [111]. Reproduced with permission of AIP.) (4) Surface loop‐gap resonator.

Figure 11.12 (a) A membrane‐supported MW resonator based on microstrip topology. (Left) Schematic cross‐sectional view of the resonator fabricated from a three‐wafer process employing a combination of RIE and anisotropic wet etching. (Right) Photograph of a fabricated resonator that operates at 37 GHz with measured unloaded

‐factors of around 400. (Brown

et al

. 1999 [115]. Reproduced with permission of Wiley.) (b) Illustration of a quasi‐planar MW resonator. The basic design resembles a half‐wavelength air‐filled cavity supporting the

mode at 26 GHz and unloaded

‐factors of 400.

Figure 11.13 (a) Layout of a planar substrate‐integrated cavity resonator for permittivity measurements employing a microstrip feed. The resonator operates at frequencies of around 8 GHz with unloaded

‐factors of around 700. (b) Simulated electric field distribution for the fundamental

resonant mode as well as photographs of fabricated resonators.

Figure 11.14 (a) Assembly flow chart of a microfabricated high‐

evanescence resonator based on a gold‐coated silicon cavity including a capacitive post for resonator loading. (b) Schematic of the tuning mechanism. Frequency tuning is accomplished by adjusting the capacitive load of the cavity by means of an electrostatically actuated diaphragm.

Figure 11.15 (a) Experimental setup. Gaseous mixture containing

p

‐H

2

and propene flows through the catalyst bed packed inside the inlet capillary close to the connection between the inlet and the outlet capillaries. Three types of capillaries of different diameters (ID 800, 405, and 150 mm) were used as the inlet capillary. The diameter of the outlet capillary was constant in all experiments (ID 150 mm). The catalyst bed was placed inside the encoding coil. (b, c)

1

H NMR spectra of the reaction mixture measured by the detection microcoil in the experiments with (b)

p

‐H

2

and (c) normal hydrogen at 22

C. The reaction was carried out in R‐800‐5 reactor. (d, e)

‐Encoded RD TOF images acquired for the R‐800‐5 reactor at

C. Normal H

2

and

p

‐H

2

were used for acquiring the images (d) and (e), respectively. The catalyst bed and the outlet capillary regions are shown with white dashed lines in image (e).

Figure 11.16 (a) Experimental setup from Ref. [138]. The

para

‐hydrogen/propene mixture flows through the hydrogenation catalyst layer packed between plugs of glass wool inside a heated quartz tube. After the hydrogenation reaction, the polarized propane gas flows from the tube into the microfluidic chip inside an NMR magnet. (b) Remote detection MRI experimental setup. (c)

H NMR spectrum of hyperpolarized propane gas measured by the detection coil. (d) Remote detection MRI pulse sequence, in which phase encoding of spatial coordinates is carried out in the

y

and

z

directions.

Figure 11.17 A

129

Xe microfluidic chip polarizer. A

129

Xe + N

2

gas mixture is pumped from a main gas manifold through the inlet, pump, probe, and finally outlet chambers, respectively.

87

Rb metal was preloaded into the pump chamber. The total footprint of the chip is

, with a thickness 1 mm. The pump and probe chambers have dimensions

and

, respectively.

Figure 11.18 A 3D‐printed NMR cell incorporating a Xe

(g)

bubble chamber. From left to right: the principle of the bubble pump, a 3D drawing of the entire cell, and a full cell including a

129

Xe solenoid coil.

Chapter 12

Figure 12.1 Schematic of different modes of small‐volume NMR hyphenation. (a) Hyphenation to chemical microseparations in which the separation column is placed outside the magnet, and the separated compounds flow into the magnet via a small‐diameter transfer capillary: the separation can either by pressure‐driven or electro‐osmotically driven. The RF coil can be perpendicular (solenoid) or parallel (e.g., Helmholz or stripline) to the main magnetic field. Detection can be continuous or stopped‐flow. (b) Hyphenation of NMR detection to segmented flow, in which several samples are spaced by a chemically inert compound and studied in turn. (c) Integration of NMR with an on‐chip separation. The RF coil is usually a planar structure, or two planar structures placed either side of the separation column.

Figure 12.2 Four different types of radio‐frequency coil used in small‐volume hyphenated NMR. A capillary containing the sample is indicated by the dotted rectangle. In each case, the

B

0

field of the magnet is in the vertical direction. (a) A solenoidal coil, (b) a spiral planar RF coil, (c) a stripline resonator, and (d) a microslot resonator.

Figure 12.3 Stopped‐flow

1

H spectrum of the first eluted peak (calyosin‐7‐

O

‐β‐

D

‐glucopyranoside) obtained with CHPLC‐NMR from an extract of

Radix astragali

using a 1.5 µl solenoidal coil. The spectrum was recorded with 30 000 signal averages and referenced to the solvent signal of residual acetonitrile at 1.93 ppm. ()

Figure 12.4 (a) Basic setup for hyphenating capillary electrophoretic separations with microcoil NMR detection. (b) Figure showing the effect of the applied voltage on the spectral linewidth of the water resonance in a buffer solution. (c) Periodic stopped‐flow CE‐NMR spectra of a mixture of arginine and triethylamine (TEA) (both initially 50 mM). A solenoidal microcoil with active volume of 8 nl was used on a 300 MHz NMR spectrometer. A volume of 290 nl is injected gravimetrically (height of 20 cm for 22 s). Arginine resonances appear at 3.1 and 1.5 ppm and TEA resonances at 2.5 and 1.0 ppm. A voltage of 7.0 kV is applied for 15 s between acquisitions. The first 48.5 min of migration time has been omitted for clarity. () (d) (Top): Spectrum of 100 mM sucrose in D

2

O acquired with a current of 30 mA passing through the capillary: the linewidth of the water resonance was 9.3 Hz. (Bottom) Spectrum obtained using wavelet decomposition and reference deconvolution. ()

Figure 12.5 Instrumentation setup for coupling CEC/CZE or CHPLC with microcoil detection. ()

Figure 12.6 (a) Principle of separation or sample concentration using capillary isotachophoresis. (b) Structure of the neurotoxin from

Calliostoma canaliculatum

, the disulfide‐bonded dimer of 6‐bromo‐2‐mercaptotryptamine (BrMT). (c) Stopped‐flow CITP/NMR spectrum of the focused analyte band containing the neurotoxin (the dimer of BrMT, 1) and a minor component. ()

Figure 12.7 Expansion of the CITP microcoil NMR spectra measured for 9 nmol of doxepin with a LE of 160 mM sodium acetate at pD 5.0 and a TE of 160 mM acetic acid‐d6. The LE and TE both contain 4.5 mM β‐cyclodextrin. The resonances of doxepin near the beginning of the analyte band have chemical shifts similar to those of free doxepin with the

Z

‐isomer as the dominant species in this region. The broad resonances in this diffuse region suggest intermediate exchange between the free and bound doxepin. Resonances in the latter portion of the band are primarily due to the β‐CD complexed

E

‐isomer. Note the pronounced increase in the intensity of the β‐CD as it co‐concentrates with doxepin in the latter portion of the band. ()

Figure 12.8 (a) Comparison of different methods of flow‐loading microsamples into the NMR coil, flow injection analysis (FIA), segmental flow analysis (SFA), and zero‐dispersion SFA. () (b) Apparatus for segmented flow analysis NMR. In the center is the sample loop valve of a sample loader. The “deliver” position shown places the sample loop in line between the sample loader pump and the transfer line connected to the NMR probe. In the “fill” position, the sample handler syringe draws sample plugs into the loop via a 200 µm i.d. capillary threaded through the sample handler needle. The sample plugs are formed by alternately drawing samples in DMSO, the immiscible fluorocarbon FC 43, and wash plugs of clean solvent. The transfer line to the probe is 3 m long (43 µl). ()

Figure 12.9

1

H spectrum of 80 ng of a DNA adduct,

N

‐(2′‐deoxyguanosin‐8‐yl)‐2‐acetylaminofluorene 5′‐monophosphate (AAF‐dGMP), structure shown. 100 000 scans were acquired in a solenoidal microcoil with 1 µl observe volume over 40 h. The inset shows the 7.4 ppm peak from the first 4 h (start) and from the last 4 h (end). Signal assignments: 6.3 ppm (CH at ribose C‐1), 4.2 ppm (CHs at ribose C‐3 and C‐4), 3.9 ppm (bridge CH

2

of fluorene and CH

2

at ribose C‐5), and 2.2 ppm (acetyl and CH

2

at ribose C‐2). ()